Airline Microeconomics [1 ed.] 1527584984, 9781527584983

Citation preview

Airline Microeconomics

Airline Microeconomics By

Tony Webber

Airline Microeconomics By Tony Webber This book first published 2022 Cambridge Scholars Publishing Lady Stephenson Library, Newcastle upon Tyne, NE6 2PA, UK British Library Cataloguing in Publication Data A catalogue record for this book is available from the British Library Copyright © 2022 by Tony Webber All rights for this book reserved. No part of this book may be reproduced, stored in a retrieval system, or transmitted, in any form or by any means, electronic, mechanical, photocopying, recording or otherwise, without the prior permission of the copyright owner. ISBN (10): 1-5275-8498-4 ISBN (13): 978-1-5275-8498-3

This book is dedicated to the amazing care that has been provided to me by my wife. She rescued me from a knock-out punch, and I will always be indebted to her.

TABLE OF CONTENTS

LIST OF FIGURES ........................................................................................................................................... xiv LIST OF TABLES ............................................................................................................................................ xvii FOREWORD ................................................................................................................................................... xviii PREFACE .......................................................................................................................................................... xix PART A: INTRODUCTION AND AVIATION LANGUAGE CHAPTER 1 ......................................................................................................................................................... 2 THE DIFFICULTIES OF CONSISTENTLY MAKING MONEY IN AVIATION 1.1 RICHARD BRANSON AND WARREN BUFFETT ................................................................................................ 2 1.2 AIRCRAFT ARE AN EXPENSIVE INVESTMENT ................................................................................................. 3 1.3 UNCONTROLLABLE MACROECONOMIC FORCES AFFECT REVENUE FORECASTS ........................................... 3 1.4 MICROECONOMIC VARIABLES AFFECTING REVENUE FORECASTS ................................................................ 6 1.5 FORECASTING COST FOR THE LIFE OF AIRCRAFT INVESTMENTS ................................................................... 7 1.6 AIMS OF THIS BOOK AND THE REMAINING CHAPTERS .................................................................................. 8 QUIZ 1.1 WHY IT IS DIFFICULT TO CONSISTENTLY MAKE MONEY IN AVIATION? .............................................. 9 CHAPTER 2 ....................................................................................................................................................... 11 THE LANGUAGE OF THE AIRLINE BUSINESS 2.1 KEY AIRLINE OPERATIONAL METRICS AND NOTATION .............................................................................. 11 2.2 LOAD METRICS ........................................................................................................................................... 11 2.2.1 Passengers Carried (PAX) ................................................................................................................ 11 2.2.2 Aircraft Movements (MOV) .............................................................................................................. 14 2.2.3 Revenue Passenger Kilometres (RPK).............................................................................................. 16 2.2.4 Cargo Carried (FRT) ........................................................................................................................ 17 2.2.5 Revenue Freight Tonne Kilometres .................................................................................................. 18 2.2.6 Revenue Passenger Tonne Kilometres (RPTK)................................................................................ 18 2.2.7 Revenue Tonne Kilometres (RTK) .................................................................................................... 18 QUIZ 2-1 LOAD METRICS.................................................................................................................................. 20 2.3 CAPACITY METRICS .................................................................................................................................... 20 2.3.1 Seats Carried (SEATS)...................................................................................................................... 21 2.3.2 Available Seat Kilometres (ASK) ...................................................................................................... 21 2.3.3 Available Freight Tonne Kilometres ................................................................................................ 21 2.3.4 Available Passenger Tonne Kilometres (APTK) .............................................................................. 22 2.3.5 Available Tonne Kilometres (ATK) .................................................................................................. 22 2.3.6 Block Hours (HOURS) ..................................................................................................................... 22 2.3.7 Fleet Units (FLEET) ......................................................................................................................... 23 QUIZ 2-2 CAPACITY METRICS .......................................................................................................................... 23 2.4 CAPACITY UTILISATION METRICS .............................................................................................................. 27 2.4.1 Passenger Seat Factor (PSF) ............................................................................................................ 27 2.4.2 Freight Load Factor .......................................................................................................................... 27 2.4.3 Total Load Factor ............................................................................................................................. 27 QUIZ 2-3 CAPACITY UTILISATION .................................................................................................................... 27 2.5 PRODUCTIVITY METRICS ............................................................................................................................ 28 2.5.1 Labour Productivity .......................................................................................................................... 28 2.5.2 Fuel Productivity ............................................................................................................................... 29 2.5.3 Fleet Productivity .............................................................................................................................. 29 QUIZ 2-4 PRODUCTIVITY METRICS ................................................................................................................... 30 2.6 AIRLINE OPERATIONAL PERFORMANCE METRICS....................................................................................... 31 2.6.1 On-Time Performance ...................................................................................................................... 31 2.6.2 Reliability ........................................................................................................................................... 32 QUIZ 2-5 AIRLINE PERFORMANCE METRICS ..................................................................................................... 32

viii

Table of Contents

2.11. YIELD ...................................................................................................................................................... 32 2.11.1 Passenger Yield ............................................................................................................................... 32 2.11.2 Passenger RASK.............................................................................................................................. 33 2.11.3 Freight Yield .................................................................................................................................... 33 2.11.4 RASK ............................................................................................................................................... 33 2.11.5 Total Yield ....................................................................................................................................... 34 2.11.6 Airline Example .............................................................................................................................. 34 2.12 UNIT COST ................................................................................................................................................ 35 2.12.1 Cost per ASK (CASK)...................................................................................................................... 35 2.12.2 Cost per ASK Excluding Fuel (CASK Ex-Fuel) ............................................................................ 35 2.12.3 Numerical Example ........................................................................................................................ 36 QUIZ 2-7 YIELD AND UNIT COST METRICS ....................................................................................................... 36 PART B: AVIATION DEMAND AND REVENUE CHAPTER 3 ....................................................................................................................................................... 40 PASSENGER DEMAND 3.1 DEMAND BY PURPOSE ................................................................................................................................ 40 3.2 MODE OF TRANSPORT SUBSTITUTES IN SHORT DISTANCE DOMESTIC MARKETS ....................................... 41 3.3 COMPLEMENTS TO AIR TRAVEL AND THE RELATIVE IMPORTANCE OF AIRFARES ....................................... 42 3.3.1 Accommodation ................................................................................................................................. 42 3.3.2 Land Transport.................................................................................................................................. 43 3.3.3 Food and Beverages .......................................................................................................................... 43 3.3.4 Entertainment.................................................................................................................................... 43 QUIZ 3-1 PURPOSE OF TRAVEL, SUBSTITUTES AND COMPLEMENTS ................................................................. 44 3.4 ORGANIC AIR TRAVEL DEMAND ................................................................................................................ 44 3.5 AIRLINE SPECIFIC DEMAND ........................................................................................................................ 45 3.5.1 An Introduction to Airline Demand ................................................................................................. 45 3.5.2 Airline Demand Functions ............................................................................................................... 45 3.5.3 Airline Demand Curves..................................................................................................................... 46 3.6 MARKET DEMAND ...................................................................................................................................... 49 3.6.1 Market Demand Function ................................................................................................................ 49 3.6.2 Market Inverse Demand.................................................................................................................... 50 QUIZ 3-2 ORGANIC, MARKET AND AIRLINE LEVEL DEMAND ........................................................................... 51 3.7 THE AIRFARE ELASTICITY OF AIR TRAVEL DEMAND ................................................................................. 52 3.7.1 Demand Elasticity Versus Demand Sensitivity ................................................................................ 52 3.7.2 Market Fare Elasticity of Air Travel Demand ................................................................................. 52 3.7.3 Drivers of the Market Fare Elasticity of Air Travel Demand .......................................................... 55 3.7.4 Cross-Mode Market Elasticity of Air Travel Demand ..................................................................... 56 3.7.5 Cross-Destination Market Elasticities of Air Travel Demand ......................................................... 56 3.7.6 Complement Market Elasticity of Air Travel Demand..................................................................... 58 3.7.7 Airline-Specific Fare Elasticities of Demand ................................................................................... 59 QUIZ 3-3 AIRFARE ELASTICITY OF AIR TRAVEL DEMAND................................................................................ 60 3.8 DEMAND OVER TIME .................................................................................................................................. 62 3.8.1 Drivers of the Trend .......................................................................................................................... 62 3.8.2 Cycle .................................................................................................................................................. 62 3.8.3 Seasonality ......................................................................................................................................... 63 3.8.4 Structural Events............................................................................................................................... 64 3.9 DEMAND ACROSS CITY PAIRS AND THE GRAVITY MODEL ......................................................................... 65 3.10 DEMAND AND NETWORK EFFECTS ........................................................................................................... 68 3.10.1 What is a Network Effect? .............................................................................................................. 68 3.10.2 Networks, Indirect Services and Hub Points .................................................................................. 71 3.10.3 Before, Beyond and Trunk Routes ................................................................................................. 72 QUIZ 3-4 CHANGES IN DEMAND OVER TIME AND NETWORK EFFECTS ............................................................. 74

Airline Microeconomics

PART C: AVIATION REVENUE CHAPTER 4 ....................................................................................................................................................... 76 SHORT RUN AIRLINE REVENUE 4.1 AIRLINE TIME HORIZONS – SHORT, MEDIUM AND LONG RUNS .................................................................. 76 4.1.1 Short Run Decision Variables .......................................................................................................... 76 4.1.2 Medium Run Decision Variables ...................................................................................................... 77 4.1.3 Long Run Decision Variables ........................................................................................................... 77 4.2 OBSERVATIONS ABOUT THE PASSENGER SEAT FACTOR ............................................................................. 78 4.3 SHORT RUN REVENUE DEFINITIONAL RELATIONSHIPS ............................................................................... 78 4.4 RELATIONSHIP BETWEEN AVERAGE AIRFARES AND THE PASSENGER SEAT FACTOR .................................. 80 4.4.1 The Logic ........................................................................................................................................... 80 4.4.2 Linear Relationship between the Average Airfare and the Seat Factor .......................................... 81 4.4.3 Numerical Example .......................................................................................................................... 82 4.5 ESTIMATING THE PARAMETERS OF THE LINEAR FUNCTION ........................................................................ 83 QUIZ 4-1. SHORT RUN PASSENGER REVENUE AND THE PSF ............................................................................. 86 4.6 SHORT RUN REVENUE WITH LINEAR AVERAGE AIRFARE FUNCTION.......................................................... 87 4.6.1 Numerical Example .......................................................................................................................... 87 4.6.2 A More General Illustration ............................................................................................................. 89 4.7 ELASTICITY OF THE AVERAGE AIRFARE TO THE SEAT FACTOR – A LITTLE MORE ADVANCED ................... 90 4.7.1 Change in Revenue and the Airfare Elasticity ................................................................................. 90 4.7.2 Calculating the Airfare Elasticity ..................................................................................................... 90 4.7.3 A Case Study Using Rex Express Airlines Data............................................................................... 92 4.8 CONSTANT-ELASTICITY AVERAGE AIRFARE FUNCTION ............................................................................. 95 QUIZ 6-2. SHORT RUN PASSENGER REVENUE AND THE PSF ............................................................................. 96 CHAPTER 5 ....................................................................................................................................................... 99 MEDIUM RUN AIRLINE REVENUE 5.1 MEDIUM RUN CAPACITY ............................................................................................................................ 99 5.2 MEDIUM RUN REVENUE ANALYTICS ........................................................................................................ 100 5.3 MODELLING PRASK AND MEDIUM RUN REVENUE .................................................................................. 101 5.3.1 Linear PRASK ................................................................................................................................. 101 5.3.2 Cobb-Douglas PRASK .................................................................................................................... 102 QUIZ 5-1. LINEAR AND COBB-DOUGLAS PRASK ........................................................................................... 104 5.4 DEEPER MEDIUM RUN REVENUE ANALYTICS........................................................................................... 106 5.4.1 General Analytics ............................................................................................................................ 106 5.4.2 An Airline Illustration – Southwest Airlines .................................................................................. 107 5.4.3 Drivers of the Own Yield Elasticity ................................................................................................. 109 QUIZ 5-2. PRASK AND OWN ASKS ............................................................................................................... 109 5.5 IMPACT OF ASKS ON MEDIUM RUN REVENUE IN THE PRESENCE OF COMPETITORS ................................. 110 5.5.1 Passenger Revenue Effects - Logic................................................................................................. 110 5.5.2 Passenger Revenue Effects – Analytics .......................................................................................... 111 5.5.3 Airline Illustration – Frontier, Southwest, and Spirit.................................................................... 112 5.5.4 Passenger Revenue Effects – Analytics Percentage Change in Revenue ..................................... 112 5.5.5 Airline Illustration – Southwest Airlines and Spirit....................................................................... 113 QUIZ 5-3. PRASK, OWN ASKS AND COMPETITOR REACTION ....................................................................... 115 5.6 AIRLINE GROUP REVENUE AND CANNIBALISATION EFFECTS ................................................................... 115 5.6.1 Examples of Airline Groups ........................................................................................................... 115 5.6.2 Cannibalisation Effect Flows ......................................................................................................... 116 5.6.3 Cannibalisation Analytics ............................................................................................................... 118 5.6.4 A Numerical Example – Qantas and Jetstar .................................................................................. 119 5.6.5 A Detailed Illustration – Singapore Airlines Group ...................................................................... 120 QUIZ 5-4. CANNIBALISATION EFFECTS ........................................................................................................... 123 5.7 IS IT REALISTIC TO BELIEVE THAT AIRLINE REVENUE CAN FALL, WHEN ASKS INCREASE?..................... 124

ix

x

Table of Contents

PART D: AVIATION COST AND PROFIT CHAPTER 6 ..................................................................................................................................................... 130 AIRLINE COST 6.1 WHAT ARE THE DIFFERENT COSTS THAT AIRLINES FACE? ....................................................................... 130 6.2 CHANGING IMPORTANCE OF FUEL AND NON-FUEL COSTS ....................................................................... 134 QUIZ 6-1: COST CATEGORIES ......................................................................................................................... 137 6.3 THE LINK BETWEEN PRODUCTIVITY, RESOURCE PRICES AND AIRLINE UNIT COST .................................. 137 6.3.1 Schematic Representation of the Linkage ...................................................................................... 137 6.3.2 Airline Production and Productivity ............................................................................................... 139 QUIZ 6-2: AIRLINE PRODUCTIVITY, RESOURCE PRICES AND UNIT COST ........................................................ 141 6.4 AIRLINE FUEL COSTS ................................................................................................................................ 142 6.4.1 Basic Analytics ................................................................................................................................ 142 6.4.2 Airline Illustrations of Time Series Movements in FCASK........................................................... 143 QUIZ 6-3: FUEL COSTS ................................................................................................................................... 146 6.5 AIRLINE LABOUR COSTS........................................................................................................................... 147 6.5.1 Labour Resources at Airlines ......................................................................................................... 147 6.5.2 Airline Wage Rates .......................................................................................................................... 149 6.5.3 Comparing Manpower Costs Across Airlines and Redundancies ................................................. 149 6.5.4 A Simple Model of Airline Labour Costs ....................................................................................... 150 QUIZ 6-4: MANPOWER COSTS......................................................................................................................... 151 6.6 AIRCRAFT CAPITAL COSTS ....................................................................................................................... 152 6.6.1 Owned Aircraft ................................................................................................................................ 152 6.6.2 Leased Aircraft ................................................................................................................................ 154 QUIZ 6-5: AIRCRAFT CAPITAL COSTS ............................................................................................................. 156 6.7 AIRPORT CHARGES ................................................................................................................................... 157 6.7.1 Airport Charges and the Building Block Approach ....................................................................... 157 6.7.2 Single-Till Versus Dual-Till Approaches to Airport Charges........................................................ 158 QUIZ 6-6: AIRPORT AND NAVIGATION CHARGES............................................................................................ 159 6.8 MAINTENANCE COSTS .............................................................................................................................. 160 QUIZ 6-7: MAINTENANCE COSTS .................................................................................................................... 161 6.9 GROUND HANDLING ................................................................................................................................. 162 QUIZ 6-8: GROUND HANDLING COSTS ........................................................................................................... 164 6.10 AIRLINE COST FUNCTIONS AND RUNS .................................................................................................... 164 6.10.1 Total Long Run Cost Function ..................................................................................................... 164 6.10.2 Medium Run Cost Function ......................................................................................................... 164 6.10.3 Short Run Function ...................................................................................................................... 165 QUIZ 6-9: COST FUNCTIONS AND RUNS .......................................................................................................... 167 6.11 NON-FUEL COST PER ASK ..................................................................................................................... 168 6.11.1 What is Non-Fuel Cost per ASK? ................................................................................................. 168 6.11.2 Analytical Representation of NFCASK ........................................................................................ 170 6.11.3 What Drives Non-Fuel Cost Per ASK over Time? ....................................................................... 171 6.12 ADJUSTING NON-FUEL COSTS PER AVAILABLE SEAT KILOMETRE FOR SECTOR LENGTH ....................... 172 QUIZ 6-10: NON-FUEL COST PER ASK ........................................................................................................... 173 6.13 AIRLINE ECONOMIES OF SCALE AND SCOPE ........................................................................................... 175 6.13.1 Airline Economies of Scale ........................................................................................................... 175 6.13.2 Airline Economies of Scope .......................................................................................................... 176 QUIZ 6-12: BELLY FREIGHT COSTS AND ECONOMIES OF SCALE AND SCOPE .................................................. 176 CHAPTER 7 ..................................................................................................................................................... 177 MAXIMISING SHORT RUN AIRLINE PROFIT 7.1 LINEAR AVERAGE AIRFARE FUNCTION .................................................................................................... 177 7.1.1 Theory .............................................................................................................................................. 177 7.1.2 Airline Illustration – Aegean Airlines ............................................................................................ 179 QUIZ 7-1. SHORT RUN PROFIT ........................................................................................................................ 182 7.2 COBB DOUGLAS AVERAGE AIRFARE FUNCTION ....................................................................................... 183 7.2.1 Theory .............................................................................................................................................. 183 7.2.2 Aeroflot Illustration ........................................................................................................................ 185 QUIZ 7-2 SHORT RUN PROFIT FUNCTION WITH COBB DOUGLAS AVERAGE AIRFARE FUNCTION ................... 187 7.3 BREAK-EVEN PASSENGER SEAT FACTOR .................................................................................................. 187 7.3.1 Traditional Calculation and its Problems ...................................................................................... 187 7.3.2 Singapore Airlines ........................................................................................................................... 187 7.3.3 A Superior Breakeven Formula ..................................................................................................... 188

Airline Microeconomics

7.3.4 Multiple Breakeven Seat Factor ..................................................................................................... 189 7.3.5 Breakeven Seat Factor when the Average Airfare is a Cobb-Douglas Function ......................... 190 7.3.6 Illustration – All Nippon Airways ................................................................................................... 190 QUIZ 7-3. BREAKEVEN SEAT FACTORS........................................................................................................... 190 CHAPTER 8 ..................................................................................................................................................... 193 MAXIMISING MEDIUM RUN AIRLINE PROFIT 8.1 MEDIUM RUN AIRLINE PROFIT MAXIMISATION – LINEAR PRASK FUNCTION ......................................... 193 8.1.1 Theory and Analytics ...................................................................................................................... 193 8.1.2 Numerical Illustration – Qantas Domestic..................................................................................... 196 QUIZ 8-1. MEDIUM RUN AIRLINE PROFIT MAXIMISATION WITH LINEAR PRASK .......................................... 199 8.2 COBB-DOUGLAS PRASK FUNCTION ........................................................................................................ 200 8.2.1 Theory and Analytics ...................................................................................................................... 200 8.2.2 Numerical Example – Allegiant Air ............................................................................................... 201 QUIZ 8-2. MEDIUM RUN AIRLINE PROFIT MAXIMISATION – COBB-DOUGLAS PRASK .................................. 203 8.3 MEDIUM RUN AIRLINE PROFIT MAXIMISATION WITH EXOGENOUS COMPETITOR CAPACITY AND LINEAR PRASK ...................................................................................................................................... 203 8.3.1 Theory and Analytics ...................................................................................................................... 203 8.3.2 Numerical Illustration – Copenhagen to London Heathrow......................................................... 205 QUIZ 8-3. MEDIUM RUN AIRLINE PROFIT MAXIMISATION WITH EXOGENOUS COMPETITOR CAPACITY AND LINEAR PRASK ...................................................................................................................................... 207 8.4 MEDIUM RUN AIRLINE PROFIT MAXIMISATION WITH EXOGENOUS COMPETITOR CAPACITY AND COBB-DOUGLAS PRASK ........................................................................................................................ 208 8.4.1 Theory and Analytics ...................................................................................................................... 208 8.4.2 Numerical Example – Emirates and Lufthansa............................................................................. 209 QUIZ 8-4. MEDIUM RUN AIRLINE PROFIT MAXIMISATION WITH EXOGENOUS COMPETITOR CAPACITY AND COBB-DOUGLAS PRASK ........................................................................................................................ 211 8.5 MEDIUM RUN AIRLINE PROFIT MAXIMISATION WITH COMPETITION AND CONJECTURAL VARIATION ..... 211 8.5.1 Ex-post Conjectural Variation ........................................................................................................ 211 8.5.2 Ex-ante Conjectural Variation ....................................................................................................... 213 8.5.3 Illustrative Example Conjectural Variation – JetBlue Airways .................................................... 214 QUIZ 8-5. MEDIUM RUN AIRLINE PROFIT MAXIMISATION WITH CONJECTURAL VARIATION ......................... 216 PART E: TOPICS IN AVIATION ECONOMICS CHAPTER 9 ..................................................................................................................................................... 220 MONOPOLY AND OLIGOPOLY AIRLINE COMPETITION 9.1. COMPETITION AND MARKET DEFINITION ................................................................................................ 220 9.1.1 Competition Basics .......................................................................................................................... 220 9.1.2 Market Definition ............................................................................................................................ 221 QUIZ 9-1 COMPETITION AND THE MARKET..................................................................................................... 223 9.2 MODELS OF COMPETITION ........................................................................................................................ 223 9.3 MONOPOLY AIRLINE MODEL OF ROUTES ................................................................................................. 225 9.3.1 A Brief Introduction to the Theory ................................................................................................. 225 9.3.2 Numerical Example ........................................................................................................................ 227 QUIZ 9-2 MODELS OF COMPETITION AND MONOPOLY ................................................................................... 228 9.4 HOMOGENEOUS DUOPOLY ........................................................................................................................ 230 9.4.1 Theory and Analytics ...................................................................................................................... 230 9.4.2 Numerical Example ........................................................................................................................ 232 9.5 DIFFERENTIATED DUOPOLY...................................................................................................................... 233 9.5.1 Theory .............................................................................................................................................. 233 9.5.2 NUMERICAL EXAMPLE ........................................................................................................................... 236 9.6 N-AIRLINE MODELS .................................................................................................................................. 237 9.6.1 Theory and Analytics ...................................................................................................................... 237 9.6.2 Numerical Example ........................................................................................................................ 238 QUIZ 9-3 OLIGOPOLISTIC AIRLINE COMPETITION ........................................................................................... 239 9.7 MEASURING COMPETITION ....................................................................................................................... 242 9.8 UNDERSTANDING WHY FARES DIFFER ACROSS ROUTES AND CABINS ...................................................... 244 9.8.1 Why do Fares Differ Across Routes? ............................................................................................. 244 9.8.2 Understanding why Fares Differ Across Cabins ........................................................................... 245 QUIZ 9-4 MEASURING COMPETITION, WHY FARES DIFFER ACROSS ROUTES AND CLASSES .......................... 247

xi

xii

Table of Contents

CHAPTER 10.................................................................................................................................................... 249 AIRLINE RELATIONSHIPS AND BUSINESS MODELS 10.1 SPECTRUM OF AIRLINE RELATIONSHIPS ................................................................................................. 249 10.2 INTERLINING ........................................................................................................................................... 251 10.2.1 What is an Interline Agreement?.................................................................................................. 251 10.2.2 Interline Prorating ........................................................................................................................ 252 10.2.3 Advantages and Disadvantages of Interline Agreements............................................................. 255 10.3 CODESHARE ............................................................................................................................................ 256 10.3.1 What is a Codeshare Agreement? ................................................................................................. 256 10.3.2 Types of Codeshare Agreements ................................................................................................... 257 10.3.3 Motivations for Codeshare Agreements ....................................................................................... 257 10.3.4 Codeshare Agreement Route Types .............................................................................................. 259 QUIZ 10-1 INTERLINE AND CODESHARE RELATIONSHIPS ............................................................................... 259 10.4 ALLIANCES ............................................................................................................................................. 261 10.5 REVENUE SHARING RELATIONSHIPS ....................................................................................................... 263 10.6 JOINT VENTURES AND PARTNERSHIPS .................................................................................................... 264 10.6.1 An Introduction to Airline Joint Ventures ................................................................................... 264 10.6.2 Initial Trans-Atlantic Joint Ventures ........................................................................................... 265 10.6.3 Qantas and Emirates Joint Ventures ............................................................................................ 266 10.6.4 SkyTeam and Star Alliance Trans-Atlantic Joint Ventures ........................................................ 267 10.6.5 Asian Joint Ventures ..................................................................................................................... 268 10.6.6 Latin America Joint Venture ........................................................................................................ 268 QUIZ 10-2 ALLIANCES, REVENUE SHARING AND JOINT VENTURES ................................................................ 268 10.7 AIRLINE EQUITY INVESTMENTS .............................................................................................................. 269 10.8 MERGERS AND ACQUISITIONS ................................................................................................................ 270 10.8.1 Mergers .......................................................................................................................................... 270 10.8.2 Acquisitions ................................................................................................................................... 273 QUIZ 10-3 AIRLINE EQUITY INVESTMENTS, MERGERS AND ACQUISITIONS .................................................... 273 10.9 LOW-COST VERSUS FULL-SERVICE CARRIER BUSINESS MODELS .......................................................... 273 10.9.1 List of Low-Cost Carriers by Region ............................................................................................ 273 10.9.2 Low-Cost Airline Spectrum and Key Characteristics ................................................................... 274 10.9.3 LCC Versus FSA Cost Categories ................................................................................................ 278 10.9.4 Full-Service Versus Low-Cost Unit Cost Comparisons ............................................................... 278 QUIZ 10-4 FSA VERSUS LCC AIRLINES ......................................................................................................... 281 CHAPTER 11.................................................................................................................................................... 282 AVIATION CHARGES, TAXES AND A PRICE ON CARBON 11.3 FUEL SURCHARGES ................................................................................................................................. 282 11.3.1 What is a Fuel Surcharge? ........................................................................................................... 282 11.3.2 Revenue Recovery with Varying Loads ........................................................................................ 285 11.3.3 Problems with Fuel Surcharges.................................................................................................... 290 QUIZ 11-3 FUEL SURCHARGES ....................................................................................................................... 291 11.4 DEPARTURE TAXES................................................................................................................................. 293 11.4.1 What is a Departure Tax? ............................................................................................................. 293 11.4.2 A Model of the Impact of a Departure Tax .................................................................................. 294 QUIZ 11-4 DEPARTURE TAX ........................................................................................................................... 300 11.5 CARBON TAXES AND EMISSIONS TRADING SCHEMES ............................................................................. 301 11.5.1 Carbon Taxes Imposed on Aviation Markets ............................................................................... 301 11.5.2 A Model of Carbon Taxes on Aviation Markets........................................................................... 303 11.5.3 A Model of Emissions Trading Schemes in Airline Markets....................................................... 305 QUIZ 11-5 CARBON TAX AND EMISSIONS TRADING SCHEME ......................................................................... 307 CHAPTER 12.................................................................................................................................................... 308 THE ECONOMICS OF OIL AND JET FUEL MARKETS 12.1 THE DEMAND FOR OIL ............................................................................................................................ 308 QUIZ 12-1 OIL DEMAND ................................................................................................................................. 310 12.2 OIL SUPPLY ............................................................................................................................................ 312 12.3 OIL PRICING BENCHMARKS .................................................................................................................... 315 QUIZ 12-2 CRUDE OIL PRICE BENCHMARKS................................................................................................... 317 12.4 AN ECONOMIC MODEL OF PRICE OUTCOMES IN THE OIL MARKET ......................................................... 318 12.4.1 Theory and Analytics .................................................................................................................... 318 12.4.2 Numerical Illustration................................................................................................................... 320 QUIZ 12-3 THE OIL MARKET .......................................................................................................................... 322

Airline Microeconomics

12.5 THE JET FUEL MARKET .......................................................................................................................... 324 12.5.1 Refining Process............................................................................................................................ 324 12.5.2 Jet Fuel Benchmark Prices........................................................................................................... 325 12.5.3 Historical Movements in the Jet Fuel Price ................................................................................. 326 12.5.4 The Demand for Jet Fuel .............................................................................................................. 326 12.6 A MODEL OF THE JET FUEL MARKET...................................................................................................... 328 12.6.1 Theory and Analytics .................................................................................................................... 328 12.6.2 Numerical Example ...................................................................................................................... 330 12.7 JET REFINING MARGINS .......................................................................................................................... 332 12.8 AN ECONOMIC MODEL OF THE JET FUEL CRACK MARGIN ..................................................................... 333 12.8.1 Theory and Analytics .................................................................................................................... 333 12.8.2 Numerical Example ...................................................................................................................... 336 QUIZ 12-4 THE JET FUEL MARKET ................................................................................................................. 337 REFERENCES ................................................................................................................................................. 339 GLOSSARY ...................................................................................................................................................... 349 INDEX ............................................................................................................................................................... 354 AIRLINE INDEX ............................................................................................................................................. 362

xiii

LIST OF FIGURES

FIG. 1-1: VIRGIN ATLANTIC PROFIT BEFORE TAX 2005 TO 2018 ...................................................................................... 2 FIG. 1-2: ECONOMIC GROWTH IN THE USA DURING THE GLOBAL FINANCIAL CRISIS ...................................................... 4 FIG. 1-3: TOTAL REVENUE OF THE TOP 8 AIRLINES IN THE USA ....................................................................................... 5 FIG. 1-4: QANTAS PASSENGER REVENUE PER ASK VERSUS THE ASX/S&P 200 STOCK MARKET INDEX ......................... 5 FIG. 1.5: GULF COAST JET FUEL PRICE 1990 TO 2019 ....................................................................................................... 7 FIG. 1-6: US TRADE WEIGHTED INDEX (MAJOR CURRENCIES) 1990 TO 2019 ................................................................... 8 FIG. 2-1: PASSENGERS CARRIED AND LEGS ..................................................................................................................... 12 FIG. 2-2: PASSENGERS CARRIED AND MULTIPLE LEGS ................................................................................................... 13 FIG. 2-3: PASSENGERS CARRIE ACROSS MULTIPLE LEGS – SYDNEY, LOS ANGELES, AND NEW YORK ............................ 14 FIG. 2-4: REVENUE PASSENGER KILOMETRES ................................................................................................................. 15 FIG. 2-5: SINGAPORE AIR CARGO, FREIGHT CARRIED ..................................................................................................... 16 FIG. 2-6: REVENUE FREIGHT TONNE KILOMETRES CALCULATION .................................................................................. 17 FIG. 3-1: AIRLINE DEMAND CURVES FOR AIRLINES A AND B ......................................................................................... 46 FIG. 3-2: CEBU DEMAND CURVE ON MANILA-JAKARTA ................................................................................................. 48 FIG. 3-3: PHILIPPINE AIRLINES DEMAND CURVE ON MANILA-JAKARTA ......................................................................... 48 FIG. 3-4: AIRLINE DEMAND CURVES WITH AN IMPROVEMENT IN THE PRODUCT OF AIRLINE A ...................................... 49 FIG. 3-5: CEBU DEMAND CURVE ON MANILA-JAKARTA AFTER PRODUCT IMPROVEMENT .............................................. 49 FIG. 3-6: MARKET DEMAND CURVE FOR AIR TRAVEL .................................................................................................... 50 FIG. 3-7: MARKET DEMAND CURVE FOR AIR TRAVEL – MANILA TO JAKARTA ............................................................... 50 FIG. 3-8: MARKET INVERSE DEMAND CURVE FOR AIR TRAVEL ...................................................................................... 51 FIG. 3-9: AIRFARE ELASTICITY OF AIR TRAVEL DEMAND ON THE MARKET DEMAND CURVE ........................................ 53 FIG. 3-10: AIRFARE ELASTICITY OF AIR TRAVEL DEMAND VARIES ALONG THE MARKET DEMAND CURVE ................... 53 FIG. 3-11: AIRFARE ELASTICITY OF AIR TRAVEL DEMAND AND SHIFTS IN THE DEMAND CURVE ................................... 55 FIG. 3-12: DEMAND FUNCTION FOR TRAVEL FROM ORIGIN A TO DESTINATION B .......................................................... 57 FIG. 3-13: EMIRATES AIRLINES PASSENGERS CARRIED ................................................................................................... 62 FIG. 3-14: CYCLE IN REX EXPRESS AIRLINES RPKS........................................................................................................ 63 FIG. 3-15: VUELING AIRLINES TOTAL SYSTEM SEAT FACTOR ........................................................................................ 63 FIG. 3-16: A TYPICAL TIMELINE OF ADVERSE SHOCKS ................................................................................................... 64 FIG. 3-17: IMPACT OF THE SARS VIRUS FOR AUSTRALIAN INBOUND TOURISM FROM ASIA ........................................... 65 FIG. 3-18: FLIGHT FROM MILWAUKEE TO LOS ANGELES VIA DENVER ........................................................................... 72 FIG. 3-19: FLIGHT FROM ADELAIDE TO COFFS-HARBOUR VIA SYDNEY .......................................................................... 73 FIG. 3-20: FLIGHT FROM EDINBURGH TO NAGASAKI VIA FRANKFURT AND TOKYO ........................................................ 73 FIG. 4-1: GLOBAL PASSENGER SEAT FACTORS ................................................................................................................ 79 FIG. 4-2: STRAIGHT LINE OR LINEAR AIRFARE FUNCTION – AIR FRANCE EXAMPLE ....................................................... 81 FIG. 4-3: STRAIGHT LINE OR LINEAR AIRFARE FOR THE AEROMÉXICO EXAMPLE ........................................................... 83 FIG. 4-4: AVERAGE AIRFARE AND PASSENGER SEAT FACTOR - STRAIGHT LINE SHIFTING OUT OVER TIME .................. 84 FIG. 4-5: REX EXPRESS AIRLINES AVERAGE AIRFARE VERSUS THE PASSENGER SEAT FACTOR FY06 TO FY18 ............. 84 FIG. 4-6: REX EXPRESS AIRLINES AVERAGE AIRFARE FUNCTION IN FY18 ..................................................................... 85 FIG. 4-7: SHORT RUN AVERAGE AIRFARE AND REVENUE FUNCTIONS ............................................................................ 88 FIG. 4-8: AIRFARE ELASTICITY OF DEMAND AND THE REVENUE FUNCTIONS.................................................................. 91 FIG. 4-9: AIRFARE ELASTICITY OF DEMAND AND THE REVENUE FUNCTIONS FOR REX EXPRESS IN FY18 ...................... 93 FIG. 4-10: SHIFT IN THE REX EXPRESS AVERAGE AIRFARE FUNCTION TO THE RIGHT ..................................................... 93 FIG. 4-11: SHIFT IN THE AVERAGE AIRFARE ELASTICITY AND SHIFT TO THE RIGHT IN THE REVENUE CURVE ................ 94 FIG. 4-12: AIRFARE AND REVENUE FUNCTIONS COBB DOUGLAS AIRFARE FUNCTION.................................................... 96 FIG. 5-1: IMPACT ON MEDIUM RUN REVENUE OF AN INCREASE IN CAPACITY ............................................................... 100 FIG. 5-2: IMPACT ON MEDIUM RUN REVENUE OF AN INCREASE IN CAPACITY ............................................................... 101 FIG. 5-3: QANTAS DOMESTIC MEDIUM RUN PRASK AND REVENUE FUNCTIONS ......................................................... 102 FIG. 5-4: COBB-DOUGLAS PRASK FUNCTION .............................................................................................................. 103 FIG. 5-5: MEDIUM RUN FUNCTION FOR COBB-DOUGLAS PRASK FUNCTION ............................................................... 104 FIG. 5-6: IMPACT OF OWN ASKS ON REVENUE WITH COMPETITOR REACTION ............................................................. 110 FIG. 5-7: IMPACT OF OWN ASKS ON REVENUE WITH CANNIBALISATION AND COMPETITOR REACTION ....................... 117 FIG. 5-8: AIR NEW ZEALAND REVENUE AND CAPACITY ............................................................................................... 125 FIG. 5-9: CHINA EASTERN REVENUE AND CAPACITY .................................................................................................... 125 FIG. 5-10: CHINA SOUTHERN REVENUE AND CAPACITY ................................................................................................ 126 FIG. 5-11: ALL NIPPON AIRWAYS REVENUE AND CAPACITY......................................................................................... 126 FIG. 5-12: UNITED AIRLINES REVENUE AND CAPACITY ................................................................................................ 126

Airline Microeconomics

xv

FIG. 5-13: INTERNATIONAL AIRLINES GROUP REVENUE AND CAPACITY ...................................................................... 127 FIG. 6-1: MAJOR COST BREAKDOWN – QANTAS GROUP FY20...................................................................................... 132 FIG. 6-2: MAJOR COST BREAKDOWN SINGAPORE AIRLINES 12 MONTHS TO MARCH 31, 2020 ..................................... 133 FIG. 6-3: MAJOR COST BREAKDOWN – TURKISH AIRLINES 12 MONTHS TO DECEMBER 31, 2019 ................................. 135 FIG. 6-4: NON-FUEL COSTS AS A PERCENTAGE OF QANTAS GROUP TOTAL COST ......................................................... 135 FIG. 6-5: NON-FUEL COSTS AS A PERCENTAGE OF SOUTHWEST AIRLINES TOTAL COST ............................................... 136 FIG. 6-6: NON-FUEL COSTS AS A PERCENTAGE OF TOTAL COST LUFTHANSA GROUP ................................................... 136 FIG. 6-7: THE LINKAGE BETWEEN AIRLINE COST, PRODUCTION AND UNIT COST ......................................................... 138 FIG. 6-8: JETBLUE PILOT AND CABIN CREW BLOCK HOURS AND FUEL CONSUMPTION – 2001 TO 2020 ....................... 140 FIG. 6-9: JETBLUE FLIGHT DEPARTURES AND FLEET UNITS – 2001 TO 2019 ................................................................ 140 FIG. 6-10: QANTAS GROUP FUEL COST PER ASK VERSUS THE SPOT PRICE OF JET FUEL ............................................. 144 FIG. 6-11: SINGAPORE AIRLINES FUEL COST PER ASK VERSUS THE SPOT PRICE OF JET FUEL ..................................... 145 FIG. 6-12: TURKISH AIRLINES FUEL COST PER ASK VERSUS THE SPOT PRICE OF JET FUEL ......................................... 145 FIG. 6-13: DELTA AIR LINES FUEL COST PER ASM VERSUS THE WTI CRUDE PRICE.................................................... 146 FIG. 6-14: QANTAS GROUP FULL TIME EQUIVALENT STAFF ......................................................................................... 148 FIG. 6-15: QANTAS GROUP AVAILABLE SEAT KILOMETRES .......................................................................................... 148 FIG. 6-16: INTERNATIONAL AIRLINES GROUP STAFF COST PER AVAILABLE SEAT KILOMETRE .................................... 151 FIG. 6-17: INTERNATIONAL AIRLINES GROUP FTE PER AVAILABLE SEAT KILOMETRE................................................. 151 FIG. 6-18: ANN NIPPON AIRWAYS BOEING 787-800 PAYMENT SCHEDULE AND EXCHANGE RATE MOVEMENTS.......... 153 FIG. 6-19: AIR FRANCE/KLM LANDING FEES AND EN-ROUTE CHARGES PER PASSENGER ............................................ 159 FIG. 6-20: SINGAPORE AIRLINES HANDLING CHARGES PER PASSENGER ....................................................................... 163 FIG. 6-21: ILLUSTRATION OF THE MEDIUM RUN COST FUNCTION ................................................................................. 165 FIG. 6-22: AN ILLUSTRATION OF THE SHORT RUN COST FUNCTION .............................................................................. 166 FIG. 6-23: QANTAS GROUP NON-FUEL COST PER ASK TIME SERIES ............................................................................ 169 FIG. 6-24: SINGAPORE AIRLINES NON-FUEL COST PER ASK TIME SERIES ................................................................... 169 FIG. 6-25: SOUTHWEST AIRLINES NON-FUEL COST PER ASK TIME SERIES .................................................................. 170 FIG. 6-26: AIR FRANCE/KLM NON-FUEL COST PER ASK TIME SERIES........................................................................ 170 FIG. 6-27: NON-FUEL COST PER ASK IN CY14 – FULL-SERVICE AIRLINES IN ASIA ..................................................... 173 FIG. 6-28: NON-FUEL COST PER ASK IN CY14 – LOW-COST AIRLINES IN ASIA........................................................... 173 FIG. 6-29: NON-FUEL COST PER ASK VERSUS AVERAGE SECTOR LENGTH IN CY14 – FULL-SERVICE AIRLINES IN ASIA ................................................................................................................................................................ 174 FIG. 6-30: NON-FUEL COST PER ASK VERSUS AVERAGE SECTOR LENGTH IN CY14 – LOW-COST AIRLINES IN ASIA . 174 FIG. 7-1: AN ILLUSTRATION OF THE SHORT RUN PROFIT FUNCTION ............................................................................. 178 FIG. 7-2: PROFIT MAXIMISING PASSENGER SEAT FACTOR ............................................................................................ 179 FIG. 7-3: AVERAGE AIRFARE AND SHORT RUN REVENUE FUNCTIONS AEGEAN AIRLINES ............................................ 181 FIG. 7-4: SHORT RUN COST AND REVENUE FUNCTIONS AEGEAN AIRLINES .................................................................. 182 FIG. 7-5: SHORT RUN PROFIT FUNCTION AEGEAN AIRLINES ......................................................................................... 183 FIG. 7-6: THE SHORT RUN PROFIT FUNCTION WITH COBB-DOUGLAS AVERAGE AIRFARE ............................................ 184 FIG. 7-7: OPTIMAL SHORT RUN PROFIT WITH EXPONENTIAL AVERAGE AIRFARE ......................................................... 185 FIG. 7-8: SHORT RUN REVENUE, COST AND PROFIT FUNCTIONS FOR AEROFLOT .......................................................... 186 FIG. 7-9: INTERPRETATION OF THE TRADITIONAL BREAKEVEN SEAT FACTOR CALCULATION ...................................... 188 FIG. 7-10: BREAKEVEN SEAT FACTOR REPORTED BY SINGAPORE AIRLINES ................................................................. 188 FIG. 7-11: SHORT RUN BREAKEVEN SEAT FACTOR ....................................................................................................... 189 FIG. 7-12: SHORT RUN BREAKEVEN SEAT FACTOR WITH BEND IN THE REVENUE CURVE ............................................. 189 FIG. 7-13: SHORT RUN BREAKEVEN SEAT FACTOR WITH COBB-DOUGLAS AVERAGE AIRFARE FUNCTION .................. 190 FIG. 7-14: SHORT RUN PROFIT ANA DOMESTIC FLIGHT WITH 180 SEATS .................................................................... 191 FIG. 8-1: MEDIUM RUN PROFIT FUNCTION .................................................................................................................... 194 FIG. 8-2: MEDIUM RUN MARGINAL REVENUE EQUALS MARGINAL COST CONDITION .................................................. 195 FIG. 8-3: QANTAS PRASK, MR, REVENUE AND OWN YIELD ELASTICITY .................................................................... 197 FIG. 8-4: NUMERICAL ILLUSTRATION OF THE QANTAS DOMESTIC MEDIUM RUN PROFIT FUNCTION ............................ 198 FIG. 8-5: NUMERICAL ILLUSTRATION OF QANTAS DOMESTIC REVENUE AND COST FUNCTIONS ................................... 198 FIG. 8-6: ALLEGIANT AIR ESTIMATED MEDIUM RUN PROFIT FUNCTION ...................................................................... 202 FIG. 8-7: ALLEGIANT AIR MARGINAL REVENUE AND MARGINAL COST ....................................................................... 202 FIG. 8-8: PROFIT MAXIMISING CONDITION WITH COMPETITOR CAPACITY .................................................................... 204 FIG. 8-9: CAPACITY REACTION FUNCTION OF THE OWN AIRLINE ................................................................................. 205 FIG. 8-10: BRITISH AIRWAYS PROFIT ON LHR-CPH ILLUSTRATION ............................................................................. 206 FIG. 8-11: BRITISH AIRWAYS MARGINAL REVENUE VS MARGINAL COST CONDITION ON LHR-CPH ILLUSTRATION .. 206 FIG. 8-12: BRITISH AIRWAYS REACTION FUNCTION ...................................................................................................... 207 FIG. 8-13: OWN AIRLINE REACTION FUNCTION WITH COBB-DOUGLAS PRASK ........................................................... 209 FIG. 8-14: EMIRATES REACTION FUNCTION ON FRA-DXB ILLUSTRATION ................................................................... 210 FIG. 8-15: EMIRATES PROFIT FUNCTION ON FRA-DXB ILLUSTRATION ........................................................................ 211 FIG. 8-16: JETBLUE PROFIT FUNCTION ON EWR-MCO ILLUSTRATION ........................................................................ 214 FIG. 8-17: JETBLUE PROFIT FUNCTION WITH EX-ANTE CONJECTURAL VARIATION ...................................................... 216

xvi

List of Figures

FIG. 9-1: HOW A MONOPOLIST AIRLINE SETS PRICE AND OUTPUT ................................................................................ 227 FIG. 9-2: PRICING IN THE AIRLINE DUOPOLY ................................................................................................................ 231 FIG. 10-1: THE SPECTRUM OF AIRLINE COOPERATION .................................................................................................. 250 FIG. 10-2: MAP OF A TRIP FROM BOGOTA TO PARIS VIA BARCELONA .......................................................................... 251 FIG. 10-3: MAP OF A TRIP FROM PORT MACQUARIE TO LAS VEGAS VIA LOS ANGELES ............................................... 252 FIG. 10-4: MAP OF A TRIP FROM AUSTRALIA’S EAST COST TO DESTINATIONS BEYOND LAX IN THE U.S.A. ............... 258 FIG. 10-5: ILLUSTRATIVE SPECTRUM OF MAIN LOW-COST CARRIERS .......................................................................... 276 FIG. 10-6: QANTAS AND JETSTAR EBIT LEVEL COST PER ASK .................................................................................... 279 FIG. 10-7: SINGAPORE AND SCOOT AIRLINES COST PER ASK ....................................................................................... 280 FIG. 10-8: AIR FRANCE/KLM AND TRANSAVIA AIRLINES COST PER ASK .................................................................... 280 FIG. 10-9: IBERIA AND VUELING AIRLINES COST PER ASK ........................................................................................... 281 FIG. 11-1: CALENDAR ANNUAL SPOT PRICE OF JET FUEL – 1990 TO 2019 .................................................................... 284 FIG. 11-2: CONCAVITY OF PASSENGER REVENUE AS A FUNCTION OF THE FUEL SURCHARGE ....................................... 287 FIG. 11-3: CONCAVITY OF THE AEROMÉXICO PASSENGER REVENUE AS A FUNCTION OF THE FUEL SURCHARGE.......... 289 FIG. 11-4: NET TOURISM AS A FUNCTION OF THE PRICE OF TRAVEL – OUTBOUND SITS HIGHER THAN INBOUND ......... 296 FIG. 11-5: DEPARTURE TAX REVENUE FUNCTION ......................................................................................................... 297 FIG. 11-6: U.K. INBOUND AND OUTBOUND TOURISM SPENDING FUNCTIONS 2019 ....................................................... 298 FIG. 11-7: U.K. INBOUND AND OUTBOUND TOURISM SPENDING FUNCTIONS 2019 WITH A DEPARTURE TAX INCREASE............................................................................................................................................................. 299 FIG. 11-8: U.K. DEPARTURE TAX REVENUE FUNCTION ................................................................................................ 299 FIG. 11-9: N-PLAYER COURNOT MODEL WITH A CARBON TAX ..................................................................................... 304 FIG. 11-10: N-PLAYER COURNOT MODEL WITH A NON-BINDING EMISSIONS TRADING SCHEME .................................. 306 FIG. 11-11: N-PLAYER COURNOT MODEL WITH A BINDING EMISSIONS TRADING SCHEME ........................................... 307 FIG. 12-1: GLOBAL OIL DEMAND GROWTH BETWEEN 1966 AND 2019.......................................................................... 309 FIG. 12-2: GLOBAL OIL DEMAND AND WORLD GDP GROWTH ..................................................................................... 309 FIG. 12-3: OIL CONSUMPTION BY COUNTRY IN 2019 .................................................................................................... 311 FIG. 12-4: OIL CONSUMPTION BY REGION IN 1965 ........................................................................................................ 312 FIG. 12-5: OIL CONSUMPTION BY REGION IN 2020 ........................................................................................................ 312 FIG. 12-6: WORLD OIL PRODUCTION GROWTH 1966 TO 2019 ....................................................................................... 312 FIG. 12-7: OIL PRODUCTION BY COUNTRY .................................................................................................................... 313 FIG. 12-8: OIL PRODUCTION BY REGION IN 1965 ........................................................................................................... 313 FIG. 12-9: OIL PRODUCTION BY REGION IN 2020 ........................................................................................................... 314 FIG. 12-10: OPEC OIL PRODUCTION – 1965 TO 2019 .................................................................................................... 314 FIG. 12-11: OPEC OIL PRODUCTION AS A PERCENTAGE OF TOTAL WORLD PRODUCTION ............................................ 315 FIG. 12-12: CRUDE OIL GLOBAL BENCHMARK PRICES.................................................................................................. 316 FIG. 12-13: EQUILIBRIUM PRICE IN THE DOMINANT-FIRM/FRINGE FIRMS OIL MODEL ................................................. 320 FIG. 12-14: IMPACT ON EQUILIBRIUM OIL PRICES OF AN INCREASE IN GDP ................................................................. 320 FIG. 12-15: IMPACT ON EQUILIBRIUM OIL PRICES OF MORE SENSITIVE FRINGE FIRM SUPPLY ..................................... 321 FIG. 12-16: IMPACT ON EQUILIBRIUM OIL PRICES OF MORE SENSITIVE FRINGE FIRM SUPPLY ..................................... 322 FIG. 12-17: STYLISED VERSION OF THE DISTILLATION PROCESS ................................................................................... 325 FIG. 12-18: U.S. REFINED PRODUCT PRODUCTION IN 2020 ........................................................................................... 325 FIG. 12-19: U.S. GULF COAST JET FUEL PRICES (US$ PER BARREL ............................................................................. 326 FIG. 12-20: GULF COAST JET FUEL AND WEST TEXAS INTERMEDIATE CRUDE OIL PRICES........................................... 327 FIG. 12-21: WORLD JET FUEL CONSUMPTION ............................................................................................................... 327 FIG. 12-22: GROWTH IN WORLD JET FUEL CONSUMPTION ............................................................................................ 327 FIG. 12-23: GROWTH IN SOUTHWEST JET FUEL CONSUMPTION VERSUS CAPACITY GROWTH....................................... 328 FIG. 12-24: U.S. DOMESTIC MARKET JET FUEL DEMAND CURVE ................................................................................. 331 FIG. 12-25: U.S. DOMESTIC MARKET JET FUEL ELASTICITY OF DEMAND ..................................................................... 332 FIG. 12-26: WEST TEXAS INTERMEDIATE CRUDE OIL PRICES AND THE JET CRACK MARGIN........................................ 333 FIG. 12-27: GULF COAST JET FUEL PRICE TO WEST TEXAS INTERMEDIATE CRUDE OIL PRICE CRACK RATIO ............. 333 FIG. 12-28: EQUILIBRIUM CONDITION FOR CRUDE OIL DEMAND, JET FUEL PRODUCTION AND PRICES ........................ 335 FIG. 12-29: OIL REFINERY PROFIT FUNCTION ............................................................................................................... 337

LIST OF TABLES

TABLE 1-1: LIST PRICES FOR BOEING AND AIRBUS AIRCRAFT IN 2018 ............................................................................. 3 TABLE 2-1: ABBREVIATIONS USED FOR AIRLINE OPERATIONAL METRICS ...................................................................... 12 TABLE 2-2: ALL NIPPON AIRWAYS AIRCRAFT IN SERVICE MARCH 2016 ........................................................................ 23 TABLE 2-3: HAWAIIAN AIRLINES REVENUE AND OPERATING DATA, DECEMBER QUARTER 2016 .................................. 34 TABLE 2-4: HAWAIIAN AIRLINES COST DATA, DECEMBER QUARTER 2016 .................................................................... 36 TABLE 3-1: DEMAND SCHEDULE FOR CEBU PACIFIC ON MANILA-JAKARTA ................................................................... 47 TABLE 3-2: TOP 100 CITY PAIRS BY PASSENGERS CARRIED ........................................................................................... 66 TABLE 3-3: SIZE OF AIRLINE NETWORKS ........................................................................................................................ 69 TABLE 4.1: GLOBAL PASSENGER SEAT FACTORS IN 2017 BY AIRLINE TYPE................................................................... 78 TABLE 5.1: GLOBAL AIRLINE GROUPS ......................................................................................................................... 116 TABLE 5.2: SINGAPORE TO DENPASAR FLIGHTS FOR SINGAPORE AIRLINES AND SCOOT............................................... 120 TABLE 5.3: SINGAPORE TO BANGKOK FLIGHTS FOR SINGAPORE AIRLINES AND SCOOT................................................ 121 TABLE 5.4: SINGAPORE TO TAIPEI FLIGHTS FOR SINGAPORE AIRLINES AND SCOOT ..................................................... 121 TABLE 6-1: LIST OF MAJOR AND MINOR AIRLINE COST CATEGORIES........................................................................... 131 TABLE 6-2: QANTAS GROUP OPERATING COST BREAKDOWN 12 MONTHS TO 30 JUNE 30, 2020 .................................. 132 TABLE 6-3: SINGAPORE AIRLINES OPERATING COST BREAKDOWN FULL YEAR ENDED MARCH 31, 2020 ................... 133 TABLE 6-4: TURKISH AIRLINES OPERATING COST BREAKDOWN 12 MONTHS TO DECEMBER 31, 2019......................... 134 TABLE 6-5: LIST OF NON-OPERATIONAL AIRLINE DEPARTMENTS ................................................................................ 148 TABLE 6-6: DIMINISHING VALUE METHOD DEPRECIATION SCHEDULE FOR A BOEING B787-800 ................................. 153 TABLE 6-7: SINGAPORE AIRLINES BALANCE SHEET SEPTEMBER 2018, 2019, AND 2020 .............................................. 155 TABLE 6-8: DNATA REVENUE AND COSTS .................................................................................................................... 163 TABLE 6-9: ESTIMATES OF PASSENGER VARYING COST AS % OF TOTAL COST ............................................................ 167 TABLE 9-1: SEAT DIMENSIONS ON THE SINGAPORE AIRLINES AIRBUS A380 AIRCRAFT ............................................... 246 TABLE 9-2: ALLOCATED COST PER SEAT BY CABIN SINGAPORE AIRLINES A380 FLIGHT SINGAPORE TO LOS ANGELES ILLUSTRATION ..................................................................................................................................................... 246 TABLE 10-1: INTERLINE PARTNERS OF AVIANCA AIRLINES .......................................................................................... 251 TABLE 10-2: CODESHARE PARTNERS PRODUCT ENHANCEMENT ................................................................................... 258 TABLE 10-3: STAR ALLIANCE MEMBERS 2020 .............................................................................................................. 261 TABLE 10-4: SKYTEAM MEMBERS 2020 ....................................................................................................................... 262 TABLE 10-5: ONEWORLD MEMBERS 2020 ..................................................................................................................... 263 TABLE 10-6: AIRLINES NOT PART OF AN ALLIANCE ..................................................................................................... 263 TABLE 10-7: EXAMPLES OF AIRLINE JOINT VENTURE ................................................................................................... 265 TABLE 10-8: QANTAS MARKETED FLIGHTS FROM SYDNEY TO LONDON VIA DUBAI – ECONOMY CABIN...................... 266 TABLE 10-9: EMIRATES MARKETED FLIGHTS FROM SYDNEY TO LONDON VIA DUBAI – ECONOMY CABIN ................... 267 TABLE 10-10: BIGGEST AIRLINE MERGERS IN RECENT YEARS ..................................................................................... 271 TABLE 10-11: LOW-COST CARRIERS IN ASIA AND OCEANIA ........................................................................................ 274 TABLE 10-12: LOW-COST CARRIERS IN EUROPE, THE MIDDLE EAST AND AFRICA ....................................................... 275 TABLE 10-13: LOW-COST CARRIERS IN THE AMERICAS ................................................................................................ 275 TABLE 10-14: SECONDARY AIRPORTS USED BY LCCS.................................................................................................. 277 TABLE 11-1: FUEL SURCHARGE STRUCTURE OF THE QANTAS GROUP AS AT APRIL 2012 ............................................. 282 TABLE 11-2: FUEL SURCHARGE STRUCTURE OF SINGAPORE AIRLINES AS AT MARCH 2012 ......................................... 283 TABLE 11-3: FUEL SURCHARGE MODEL ZONES AND PASSENGERS ............................................................................... 283 TABLE 11-4: THE UK AIR PASSENGER DUTY RATES .................................................................................................... 293 TABLE 11-5: COUNTRIES THAT IMPOSE A CARBON TAX, YEAR IMPOSED AND 2019 VALUE ......................................... 302 TABLE 11-6: COUNTRIES THAT IMPOSE AN EMISSION TRADING SCHEME AND YEAR IMPOSED ..................................... 305 TABLE 12-1: PRODUCTS THAT HAVE OIL AS A BASIS .................................................................................................... 308 TABLE 12-2: CHARACTERISTICS OF DIFFERENT CRUDE OIL BENCHMARKS .................................................................. 315 TABLE 12-3: JET FUEL CONSUMPTION, CAPACITY AND JET FUEL PRODUCTIVITY IN THE US DOMESTIC MARKET – 12 MONTHS TO SEPTEMBER 2020 ............................................................................................................................. 330

FOREWORD

I was intrigued when Tony told me of his plans to write this book on air transport economics, as I’ve long admired his analytical approach to the industry’s key economic issues. During my 17 years as Chief Economist at the International Air Transport Association I have come across few industry insiders that have applied to air transport the analytical tools of economics with such enthusiasm, vigour, and insight. I first met Tony when he was heading up the economics team at Qantas. When he returned to academia and consulting, we were quick to use his modelling talents to tackle some of the issues we were grappling with at the time. This is a key moment for air transport, both for its users and providers. The COVID pandemic virtually stopped the air transport of passengers (but not cargo) and, at the time of writing, shows little sign of allowing a return to the previrus normal. Preferences for air travel, particularly for business, may change following the forced use of video conferencing as a substitute. A second challenge to air transport’s future is climate change, or rather the response to it. The industry is a small cause of rising temperatures today, but business-as-usual expansion over coming decades would use up much of the atmospheric carbon budget that climate models estimate is left, if we are to keep global temperature rise to 1.5 degrees. Air transport may get more expensive as a result – though to fully understand what may happen you need the tools contained in this book. This book will show you, in depth and with many useful examples drawn from Tony’s experience, how the analytical tools of economics can be applied to the myriad of issues around airline revenue, cost, competition and profitability, as well as topics like the carbon taxes or cap-and-trade schemes facing the industry because of climate change. The first part is structured around the airline profit and loss account, with a deep dive into the detail and modelling of unit revenues, costs, and key concepts such as breakeven load factors. For any analyst wanting to better understand or predict the profitability of firms in the airline sector this is required reading. There is a comprehensive assessment of demand, the markets of various kinds for air transport service, services to transport goods as well as people. The section on elasticities – how travellers and shippers respond to changes in prices and incomes at various market levels is particularly useful. Of course, what matters to firms is revenue not just the number of customers, and there is a comprehensive tour through the intricacies of modelling these to help decision-making over different time horizons. How this translates to the financial success or otherwise of the air transport firm depends critically on the economics of the production process, how costs are affected by economies of scale and scope, about how this interacts with competition to drive route profitability. Much detail is provided here about how to apply analytical tools to model these key determinants of profitability. The book continues using the same analytical approach to tackle issues like competition, fuel surcharges and oil markets. This book contains a first-class toolkit for anyone seeking to apply the analytical tools of economics to model many of the key issues in air transport, written with an insiders’ experience and replete with useful illustrations. Brian Pearce 12 December 2021

PREFACE

When I first started at Qantas, I felt like I didn’t fit in. I was the new kid on the block with a mountain of microeconomic and statistical modelling expertise under my belt but very little aviation expertise. The Qantas people who had worked at Qantas for ten, twenty, thirty and in some cases forty years, who had been completely immersed in the complex world of aviation since they were just out of high school, didn’t understand my language and I didn’t understand theirs; at least not at the start. I realised quickly that I was never going to hold my own against these experienced aviation warriors on practical aviation knowledge, so I needed to win them over in another way. I needed to win them over based on my strengths not their strengths. This certainly wasn’t easy because my strengths can often confound people, largely because my strengths are boundup in the analytics of aviation economics and finance. Most people, including the good people working at Qantas, are very binary – they either enjoy an analytical approach to solving complex problems, or they see this type of approach and dissolve like an aspirin in water. The majority of the very experienced aviation people at Qantas were the latter, and so when they saw my analytical approach to tackling complex problems, rather than try and understand it they often would completely write it off. “It doesn’t work”; “This is silly”; “The world doesn’t work like that”; “You can’t model this complexity using analytical approaches”; “The elasticity doesn’t exist”, were just some of the responses I would get. To try and break down these walls and views I needed to get some early wins with my modelling, and that I did. Probably the first win was in relation to something that is called fuel hedge effectiveness testing. This is a test that airline accountants must conduct to determine whether the airline’s fuel hedging has been good or bad. When I first started at Qantas, probably within a week or two, I was thrown straight into this complex problem of coming up with a new test because the test that Qantas was using at the time simply wasn’t working. As it turns out, this was probably the best thing that could have happened to me. It meant that, not only was I forced into learning probably the most important and complicated financial aspects of an airline’s business, that being fuel hedging, but I also had a complex mathematical problem to solve which is my strength. I ended up building a test that worked called the WHET, or the Webber Hedge Effectiveness Test, which Qantas continued to use even after I left the organisation. From that point on I gained some respect from the Treasury Risk Management team at Qantas and that respect seemed to spread across several departments, albeit after a period of time. Another big win came when I correctly forecast that Qantas’ revenue would continue to fall for another year halfway through the impact of the Global Financial Crisis. This was met by enormous scepticism by some very senior people at Qantas, as my views were presented to both the Qantas Board and the senior executive committee of the airline. But it turned out to be a reasonably accurate position. This position came after I spotted a strong correlation between Qantas’ revenue and the Australian stock market. Using the strength of this correlation and knowing that the Australian equities market was continuing to head downward, I knew that Qantas was likely to be headed for a second year of revenue weakness and this turned out to be accurate. I named the two years of revenue weakness as wealth and income effects, with the wealth effect in the first year and the income effect in the second year. It was a reference that the current CEO of the Qantas Group, Allan Joyce, started using in presentations and speeches and so my analysis had rubbed-off on the right people. Both the fuel hedge effectiveness testing and revenue forecasting wins set off a chain rection of wins which allowed me to garner much more respect within the airline than when I started, even amongst quite senior aviation people who had been a part of the Qantas furniture for decades. These wins occurred across a range of departments, including Treasury, Finance, Strategy, Yield Management, Government Relations, Marketing, Communications and Revenue Forecasting to name but a few. I mention this in this preface because the analysis and the modelling that was an integral part of the worked performed for these departments are a key part of the contents of this book. All the major wins that I had at Qantas use techniques that are documented in this book. And I want to share this with you the reader.

PART A: INTRODUCTION AND AVIATION LANGUAGE

CHAPTER 1 THE DIFFICULTIES OF CONSISTENTLY MAKING MONEY IN AVIATION

1.1 Richard Branson and Warren Buffett Richard Branson, founder of airline businesses such as Virgin Atlantic, Virgin Express (now Brussels Airlines), Virgin Australia, Virgin Nigeria (now Air Nigeria) and Virgin America (now part of Alaska Airlines), once famously said about investing in airlines “If you want to be a Millionaire, start with a billion dollars and launch a new airline” (Metcalf, Pendleton, and Mak 2020)

Meaning that if you invested a billion dollars in an airline then you will eventually wind up with something materially less than 1 billion dollars because of the losses the airline is expected to make. One of the world’s most successful share market investors in history, Warren Buffet, was also just as scathing about airline investments “Investors would have saved millions of dollars if someone had shot down the Wright Brother’s plane” (Levine-Weinberg 2021)

Why would two of the most astute and successful investors and entrepreneurs in modern history recommend, in the strongest possible way, to invest in anything else but airlines? In the case of Buffett, it is quite understandable. He lost a considerable amount of money back in the late 1980s investing in an airline called U.S. Air, which was a major airline operating in the U.S. international and domestic markets until it merged with American Airlines in 2013. According to an article published in the New York Times on October 3, 1995, Mr Buffett’s company, Berkshire Hathaway, bought preferred shares in the airline for US$358m in 1989. By March 1995, however, the value of those shares fell to just US$89.5m, resulting in Mr Buffet bearing a considerable capital loss on his U.S. Air investment (Feder 1995). This was to be Mr Buffet’s last foray into airline investments for decades. Virgin Atlantic Profit Before Tax (GBP) m £150 £100 £50 £0 -£50

(1)

(2)

(5)

-£100

(8)

(9)

(4)

-£150 -£200

(7)

(3) (6) 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018

Source: Airline Intelligence and Research Database 2021

Fig. 1-1: Virgin Atlantic Profit Before Tax 2005 to 2018 Mr Branson’s view, on the other hand, was shaped by his role as a founder and owner of airlines, the biggest of which, and his first airline, being Virgin Atlantic. As can be seen in Figure 1-1 above, the earnings of Virgin Atlantic between 2005 and 2018 were miserable. Figure 1-1 shows that the airline lost money in 9 years out of 14, with the numbers in parentheses in the figure indicating those annual periods in which the airline made a loss. On aggregate between 2005 and 2019, the airline had lost £271m, which means that the airline’s good years have not been enough to compensate for the bad years. When looking at these numbers, it is not at all surprising that Mr Branson has arrived at his dismal view of airline investments. Despite the underwhelming performance of his airline ventures, Mr Branson has maintained much of his investment and interest in airlines, in addition to flying into space, because of his clear passion for the sector.

The Difficulties of Consistently Making Money in Aviation

3

1.2 Aircraft are an Expensive Investment The easiest way to understand why Warren Buffett was wrong with his airline investments, and Richard Branson was unable to consistently make any money from airline operations, is to understand the wide variety of factors that determine the return on an investment in aircraft by airlines. The topic of this book, airline microeconomics, helps us to identify these factors, and understand how unpredictable they can be, how they impact airline earnings, and how airlines should respond to them in a way that maximises profit. The first and most important factor that determines an airline’s return on investment in an aircraft is the cost of that investment as represented by the price of aircraft. As Buffett has pointed out, one of the reasons why it is difficult to make money in airlines is that it is capital intensive, and the capital that is bought by airlines is expensive (Zhang 2017a). To buy an aircraft the cost to airlines is in the hundreds of millions of dollars. Table 1-1 below provides the 2018 list price of a range of popular Boeing and Airbus aircraft. The prices of these popular aircraft range from just over US$100m for the Boeing 737-800 to almost US$450m for the Airbus A380. Airbus 2018 List Price (US$m) A320neo US$110.6 A321neo US$129.5 A330-200 US$238.5 A350-900 US$317.4 A380 US$445.6 Aircraft

Boeing 2018 List Price (US$m) B737-800 US$102.2 B737 Max 9 US$124.1 B787-800 US$239m B787-900 US$281.6 B747-800 US$402.9 Aircraft

Source: McNutt 2018

Table 1-1: List Prices for Boeing and Airbus Aircraft in 2018 When airlines decide to invest in a new aircraft, it is a small fortune that they must pay. In addition to the cost of the aircraft they must also pay for the funds that must be raised to pay for the aircraft. Airlines will raise funds to make this purchase by borrowing money from financial institutions, or sometimes a group of institutions, ask airline owners for more funding, or use funds that the airline has accumulated from its operations. These sources of capital come at a cost, such as an interest rate that is paid to a financial institution or a dividend paid to owners of share capital. The cost of capital that airlines pay is relatively high because it depends on the riskiness of aircraft investments, which is considered high.

1.3 Uncontrollable Macroeconomic Forces Affect Revenue Forecasts Aircraft have an average life of around 20 years. This means that when an airline buys a plane it is a long-term investment. The business case that the airline must build to justify making such a long-term investment involves determining the revenue and costs from operating the aircraft over a 20-year, forward looking horizon. Forecasting the revenue and the costs associated with operating an aircraft is difficult one day, one week and one month into the future, let alone 20 years into the future. In fact, it would be fair to say that airlines almost always get these forecasts wrong not because they are terrible forecasters but because it is simply too difficult to forecast revenue and costs 20 years into the future. The reason it is so difficult to forecast the revenue that is generated by selling seats to passengers and space to freight, is that it is affected by so many different variables. This includes macroeconomic and microeconomic variables that the airline cannot control, and microeconomic variables that the airline can control but whose effect on revenue is difficult to quantify. Let us spend a few moments discussing some of these variables. Probably the most influential variable that uncontrollably affects the revenue generated from airline investments is economic growth. Economic growth is a measure of the extent to which the income and the production of an economy grows over time. Economic growth can be measured for a country, a state within a country or a major city within a country. It can also be measured for a group of countries that share a similar geographical location, such as the countries of Asia and Europe. Economic growth affects an airline’s passenger and freight revenue because it influences the ability of a passenger to pay for a seat on a plane, and it influences the amount of goods that are produced in the economy, which in turn affects the demand for space on aircraft to carry freight. Airlines cannot control economic growth and economic growth is unpredictable. We typically expect at least one and sometimes two severe economic downturns every ten years. Each time there is a severe economic downturn this results in depressed airline yields, weak passenger and freight loads, and airline revenue that is in the doldrums. This means that over the life of an aircraft we would expect up to four years of weak economic growth, which feeds into subdued airline revenue. Knowing this, how does the airline factor this into the revenue it expects to earn from operating an aircraft? How many periods of economic weakness does it assume? When are these periods of weakness assumed to occur? How long after the economic weakness is airline revenue impacted? How deep will be the impact of economic

Chapter 1

4

weakness on the aircraft’s revenue? These are extremely difficult questions to answer, but they are questions to which the airline must provide answers if it is to quantify the revenue it is expected to obtain from an aircraft investment. Unfortunately, these questions are so difficult to answer that airlines rarely get them right, and understandably so. Weak economic growth can have a devastating impact on airline revenue. To see this, let us examine the impact that the Global Financial Crisis had on airlines in the U.S. in 2009. Figure 1-2 presents the movement over time in the economic growth of the U.S. economy. US Real GDP Growth % 5 4 3 2 1 0 -1 -2 -2.5

-3 2004

2005

2006

2007

2008

2009

2010

2011

2012

2013

Source: Federal Reserve Bank of St Louis 2019

Fig. 1-2: Economic Growth in the USA During the Global Financial Crisis We can see that during the 2009 Global Financial Crisis the U.S. economy went backwards by 2.5%, when it usually grows by between 2% and 3% per annum. This effectively means that the income for the U.S. economy fell in 2009. When an economy’s income falls, the households and businesses in that economy spend less on goods and services, and they especially spend less on discretionary items such as air travel and a holiday. Less spending on air travel is precisely what happened during the Global Financial Crisis. As can be seen in Figure 1-3, the money that passengers and freight distributors spent on air travel with the top 8 airlines in the U.S. fell by 16.1% in 2009, which represents around US$21b. To put this drop in revenue into perspective, a decline in revenue of 1% to 2% is usually considered significant and unusual for an airline. This was, however, one of the most significant economic downturns since the great depression in both the U.S. and for the global economy, and unusually weak revenue was to be expected. It presents a clear example of the extent to which an unpredictable, adverse economic event can have an impact on the fortunes of the airline business, and deeply affect the business case for investing in aircraft. For airlines that invested heavily in aircraft assets, and received delivery of aircraft assets, in and around the time of the Global Financial Crisis, this would not have augured well for the aircraft meeting the desired or budgeted return on the airline’s aircraft investment. Gross Domestic Product does not capture the entire impact of economic activity on the decisions made by passengers to travel or not, and the revenue streams of airlines. This is because it does not capture the capacity and willingness to pay for air travel of a large segment of passengers that are likely to pay for their travel from their savings, and the returns to those savings. These retired passengers will pay for their grey nomad journeys through the cash they have in the bank, the investments they have made in property and the share market, and the money they have tucked away in superannuation and pension funds. The variable that often captures these types of effects, which I call wealth effects, is the movement in the stock market. As the movement in the stock market captures the wealth and confidence of around one third of the population that is retired, particularly in developed countries, we find that it is an important driver of the underlying demand for air travel and therefore the revenue that airlines earn from aircraft investments. This is reflected in the observation that airline yields, which we will talk in more detail about in chapter 2, are often highly correlated with movements in the stock market, or as investors like to call them, equity indices. For example, in Figure 1-4 below, we present the movement through time in Qantas Group passenger revenue per ASK (PRASK) versus the aggregate equity index for the Australian economy, which is the ASX/S&P 200.1 1 As we will see in chapter 2, PRASK is a particular measure of airline yield that is determined by dividing passenger revenue by a measure of airline seat capacity called available seat kilometres. In jurisdictions that use miles it is referred to as PRASM, or passenger revenue per available seat mile.

The Difficulties of Consistently Making Money in Aviation

5

Top 8 US Airline Revenue (US$ m) 160,000 150,000 140,000 $128,865

130,000 120,000

16.1% Decline

110,000

$108,183

100,000 90,000 80,000 2004

2005

2006

2007

2008

2009

2010

2011

2012

2013

Source: Airline Intelligence and Research Database 2021

Fig. 1-3: Total Revenue of the Top 8 Airlines in the USA Qantas Group PRASK (AU$)

PRASK

ASX/S&P 200

ASX/S&P 200

6700

0.110

6200

0.105

5700

0.100

5200 0.095

4700

0.090

4200 3700

0.085

3200

0.080

2700 0.075

2200

0.070

1700 2018

2017

2016

2015

2014

2013

2012

2011

2010

2009

2008

2007

2006

2005

2004

2003

2002

2001

2000

1999

1998

1997

1996

1995

1994

1993

Source: Airline Intelligence and Research Database 2021, Yahoo Finance ASX/S&P 200 2019

Fig. 1-4: Qantas Passenger Revenue per ASK Versus the ASX/S&P 200 Stock Market Index The Qantas Group is the biggest airline Group in Oceania and indeed one of the biggest airline Groups in the world. It has its main headquarters in Mascot, Sydney as well as a smaller head office in Melbourne, Australia. The airline Group operates two flying brands – Qantas Mainline, which is a full-service airline, and the Jetstar Group, which is a low-cost carrier, both operating extensively in domestic Australian and international aviation markets. Between 1993 and 2018 there was a +88% correlation between Qantas Group passenger revenue per available seat kilometre and the ASX/S&P 200. This means that a stronger stock market in Australia, which led to greater consumer and business wealth and confidence, coincided with higher Qantas Group yields and revenue.

6

Chapter 1

The fact that there is a strong positive correlation between stock market indexes and airline yields, and stock market indices are highly volatile, highlights how difficult it is to attempt to forecast the 20-year forward revenue stream of an aircraft investment. To do so with any accuracy would require the airline to guess what is likely to happen to the share market for each year, 20 years into the future. If the airline was able to do this with any accuracy, then it would give up the airline business and open-up a remarkably successful stock broking firm or investment bank. The exchange rate is another, important variable that an airline must forecast to predict the revenue that an aircraft is likely to generate over its life, particularly if that aircraft is used for international services carrying passengers from different currency zones. Aircraft used on international routes are likely to generate revenue for the airline in a variety of currencies, such as the US dollar, the Euro, the Japanese Yen, the Chinese Yuan, the Great British Pound, the Canadian dollar, the Swiss Franc, and the Australian dollar. For example, in the case of the British low-cost carrier easyJet, which flies services throughout the United Kingdom and Europe, the airline generated 42% of its revenue in Pound Sterling, 47% in Euro, 1% in US dollars, and 10% in mostly Swiss Franc over the 12 months to 30 September 2020 (easyJet Annual Report 2020, 58). When these currencies change against the local currency of the airline (the Great British Pound in the case of easyJet), so does the revenue that is earned by the aircraft investment. For the airline to accurately predict the 20-year revenue stream of its aircraft investment, it will need to accurately predict the movement in a range of currencies. Most very experienced exchange rate analysts and economists find it exceptionally difficult to predict what is likely to happen to currencies a month or a quarter ahead let alone each year for 20 years ahead. In fact, the exchange rate is usually so difficult to predict that one of the best models of the exchange rate says that the forecast of the exchange rate tomorrow is the exchange rate today (Meese and Rogoff 1983, 3). This is a model of the exchange rate called the random walk (Hamilton, 1994, 436). The revenue generated by aircraft is also affected by completely random, usually adverse events. These events include the outbreak of deadly viruses such as the Coronavirus and SARS. It also includes the disruption to flying due to ash that is present in the sky at flying altitude that is the result of volcanic eruptions, such as the Eyjafallajökul eruption in April 2010 which heavily impacted air travel in Europe for several days. Adverse shocks to air travel also come about because of the impact of earthquakes in tourism areas, and the subsequent flooding generated by tsunamis such as the Indian Ocean earthquake and tsunami in 2004. They also include events of terror, such as the events of 9-11 in September 2001 and the Bali bombings in 2002 and 2005. Events such as these most certainly cannot be predicted by airlines in determining the revenue that aircraft investments are likely to generate over their lifetime. This is despite history telling us that over the life of an aircraft asset it may be affected by such events two to three times. The macroeconomic variables discussed above which affect aircraft revenue streams, are beyond the control of the airline. The airline has no control over the extent to which economies grow, the movement in the stock market, the value of different currencies, and extreme events of nature. These forces are not only uncontrollable, but they are also highly volatile or unpredictable, some more than others. Economic growth, share markets and exchange rates are exceptionally difficult to predict, especially over a 20-year horizon.

1.4 Microeconomic Variables Affecting Revenue Forecasts The lifetime revenue stream of an aircraft investment will also depend on variables that are within the control of the airline. The airline will determine the number of seats on the aircraft, and the mix of first, business, premium economy, and economy class seats. It will control the prices that it offers passengers and freight forwarders and the number of seats that it makes available for sale at different prices. It determines the number of seats on the plane that are occupied by passengers, also called the passenger seat factor. It will determine the frequency of services of the aircraft, the timing of services and where the aircraft will be flown. The airline will also determine how much it will spend on marketing and the channels it uses for marketing, such as TV, radio, social media, magazines, billboards, newspapers, and emails. It determines the extent to which it trains cabin crew, which in turn affects the customer experience. It determines the training that flight crew and engineers receive, which in turn influences the ability of the aircraft to arrive safely. The airline also determines the points that frequent flyer members need to book a seat on a plane, or upgrade to a better seat, and it determines the points that frequent flyer members earn when they fly with the airline or spend money on a credit card. While the airline has control over these variables, and they affect the revenue stream of an aircraft over its lifetime of operation, this does not mean that the airline knows exactly how these variables will affect said revenue. This too adds to the difficulty in predicting a new aircraft’s revenue stream. For example, if an airline decides to sell more seats at cheap fares, it does not know exactly the extent to which this will encourage passengers to fly on the airline, which will have a significant impact on the revenue that the airline earns. When an airline decides to advertise its lower fares on Facebook, it does not know how this will influence the demand for seats and thus revenue. When an airline decides to form a relationship with another airline on a route, such as a codeshare or interline arrangement which we will discuss in more detail in chapter 10, it does not know how this will impact its revenue and earnings. There are also variables that are industry in nature, or microeconomic, which affect an airline’s revenue that are outside of the control of the airline. The first and probably most significant is competition from, and decisions made by, other airlines. Airlines may battle against aggressive competitors with materially lower costs and better products, in many of its markets or routes. It can come up against carriers that are owned by Governments, with aggressive growth underwritten by their owners. Or it can come up against new types of carriers, such as low-cost carriers and ultra-low-

The Difficulties of Consistently Making Money in Aviation

7

cost carriers. Over a 20-year period, the number of competitors, the types of competitors and the quality of competitors can change markedly. All are difficult to predict and have a deep impact on the yields, demand, and revenue of an aircraft investment. Other microeconomic variables that can have a significant impact on the revenue stream of aircraft assets are taxes, many of which we will discuss in chapter 11. Income, departure, carbon, payroll and goods and service taxes can all affect the potential revenue stream generated by aircraft assets. It is often the case that governments make changes to taxes that are both unpredictable and have a material impact on the revenue that can be earned by an aircraft investment.

1.5 Forecasting Cost for the Life of Aircraft Investments Not only is the revenue stream of a new aircraft asset difficult to predict over the life of the asset, but so too is the cost stream. The most difficult component of an airline’s cost to predict is its fuel cost, and the most unpredictable component of an airline’s fuel cost is the fuel price. The fuel price that is relevant to aircraft costs is the jet fuel price. As indicated in Figure 1-5 below, since the early 2000 period the jet fuel price has trended rapidly upward and moved in a volatile manner around that upward trend. Gulf Coast Jet Fuel Price (US$/bbl) $145

$123

$125 $105 $85 $65 $45 $29

$25 $5

2019 2018 2017 2016 2015 2014 2013 2012 2011 2010 2009 2008 2007 2006 2005 2004 2003 2002 2001 2000 1999 1998 1997 1996 1995 1994 1993 1992 1991 1990 Source: Energy Information Administration Spot Prices 2021

Fig. 1.5: Gulf Coast Jet Fuel Price 1990 to 2019 When airlines purchased aircraft for hundreds of millions of dollars back in 2003 their expectation was that the jet fuel price would cycle up and down around US$29 per barrel for the next 20 years, as it had done prior to 2003. Instead of this occurring the jet fuel price decided to surge upward to around the US$100 per barrel mark. It would be hard to imagine that any of the business cases built in 2003 for aircraft purchases would have conceivably anticipated the jet fuel price increasing to over $100. To examine the impact that a higher jet fuel price would have had back in 2003, consider an Emirates Boeing 777 aircraft that consumes around 300,000 barrels of jet fuel per year. At a jet fuel price of US$30 per barrel, consuming 300,000 barrels of jet fuel per year amounts to an annual fuel cost of US$9m. At an average jet fuel price of US$85 per barrel, which is the average between 2003 and 2019, the fuel cost for the aircraft for a year of jet fuel consumption becomes US$25.5m – more than double the average fuel cost prior to 2003. The exchange rate not only makes it difficult to predict the revenue stream of an aircraft investment but also the cost stream. In fact, it usually has a deeper impact on cost than revenue because of the concentration of costs denominated in US dollars, most notably fuel and aircraft capital costs (depreciation and operating lease costs). As indicated in Figure 1-6 below, the US dollar fell by 33% between 2003 and 2011 but then appreciated by 18% thereafter. Such exchange rate volatility would have resulted in materially lower fuel and aircraft capital costs between 2003 and 2011 but higher costs thereafter. This type of unpredictability and volatility makes it extremely difficult to accurately put together business cases for aircraft investments.

Chapter 1

8

US Trade Weighted Index (Major Currencies) 110

107.85

105 100 95

91.79

90 85 80 75 70.91

70

2019 2018 2017 2016 2015 2014 2013 2012 2011 2010 2009 2008 2007 2006 2005 2004 2003 2002 2001 2000 1999 1998 1997 1996 1995 1994 1993 1992 1991 1990 Source: Federal Reserve Bank of St Louis 2019

Fig. 1-6: US Trade Weighted Index (Major Currencies) 1990 to 2019

1.6 Aims of this Book and the Remaining Chapters The airline microeconomics topics that we present in this textbook aim to draw attention to the abovementioned factors and variables that make it difficult to consistently make money in aviation, and difficult to predict the return that an airline is expected to make on an investment in an expensive aircraft. It will present topics that help us understand how these variables and factors influence demand, revenue, costs and therefore airline profitability. It will also analyse how airlines should adjust the variables that they can control, such as capacity and prices, in response to changes in these variables and factors to maximise the profit that they earn. The first step in meeting these aims is to understand the acronyms and language that aviation people use – this is the focal point of chapter 2. Without this understanding graduates starting out in aviation will be left behind in conversations, meetings, discussions, presentations, media releases, external media pieces on aviation, and internal communications. Communication is an important part of aviation strategy, which is why airlines and airports often have many people working within the Corporation Communications Department and the Head of Corporate Communications is often a member of the executive committee of the airline. Without understanding the language and acronyms of the aviation industry this will delay the development process of young people who enter the industry. The passenger demand side of the airline business is then examined in chapter 3. To understand the complexity of the airline business it is necessary to understand the forces that drive the demand for passenger travel by air. This will include market drivers such as macroeconomic variables as well as airline specific drivers such as the quality of the airline product and the fares that airlines charge. Probably the most important parameter in aviation is the way that the average airfare impacts airline demand, which is often referred to by economists as the airfare elasticity of air travel demand. This parameter will form a key part of the analysis of chapter 3, as well as other chapters in this book. The passenger demand for air travel combined with the impact of airfare movements on demand enable us to model the revenue earned by airlines, which is the topic of chapters 4 and 5. The way that revenue is modelled is different in the short and medium runs for airlines. The short run focuses on the relationship between the seat factor, the average airfare and revenue for a fixed level of airline capacity. The medium run analyses the relationship between airline capacity, yields and revenue for a given number of fleet units. In both cases there are ‘bends’, or non-linearity to be technical, in the relationship between revenue and these variables that is due to increases in airline supply causing a decline in price. The cost side of the airline business is then examined in chapter 6. The dominant costs include fuel, manpower., aircraft capital costs and airport charges. What this tells us is that airlines face significant fixed costs. One of the challenges on the cost side of the airline business is constructing a useful measure of unit cost. While most airlines quite rightly use cost per available seat kilometre to measure unit cost, this is a troublesome measure for airlines that have a significant freight side to their business. There are ways to overcome this problem, but you will have to read chapter 6 for more insights. To understand how airlines can make better decisions, we make use of some useful tools from microeconomics in chapters 7 and 8. In the short run, airlines wish to choose the seat factor and the average airfare to maximise profit. By understanding how the seat factor influences the average airfare, revenue, and cost, we can determine the optimal level of the seat factor and the average airfare. We will show in this book how this is dependent on the proportion of costs

The Difficulties of Consistently Making Money in Aviation

9

that vary with the load on the aircraft, the sensitivity of average airfares to the load, and the willingness of consumers to pay for a seat on the plane. By understanding how airline capacity influences revenue, cost and thus profitability, we can calculate the level of available seat kilometres or capacity that maximises profit. This depends on the proportion of costs that vary with capacity, the elasticity of yield to capacity, and underlying demand. By identifying profit maximising capacity, the airline maximises medium run profit, which places the airline in the best position to overcome the random shocks that affect the business. This is covered in detail in chapter 8. Chapter 9 investigates how we model competition between airlines. It focusses on two standard models of competition as set out in the economics literature of industrial organisation, specifically monopoly and oligopolistic competition. These two models of competition are the focal points because it is these two models, particularly the latter, that is most likely to accurately explain the reality of airline competition on most routes. The models can then be used to understand why airfares differ across routes and classes of travel. The key findings of the models are that airfares differ across routes and cabins because of unit costs, the number and quality of competitors, and the sensitivity of demand to a change in the average airfare. One of the ways that airlines can improve their performance is by identifying airlines with which they can build relationships to improve yields, demand, and revenue and to reduce costs. This is the topic of investigation of chapter 10. Airline relationships may take the form of interline and codeshare relationships, joint ventures, revenue sharing, equity investments, mergers, and acquisitions. Airlines will investigate the possibility of making these relationships come to fruition subject to approval from competition authorities. Governments tax individuals and companies to finance spending on public infrastructure and welfare. Like most other companies, airlines pay their fair share of taxes, including company tax, goods and services tax, departure tax, carbon tax amongst many other taxes. Changes in these taxes can affect the incentives that airlines face to invest in more capacity. This is the topic of discussion of chapter 11. In chapter 12, we present to you the economics of the oil market. The oil market is pivotal to understanding the financial performance of airlines, given that it is such an important driver of airline costs. In chapter 12 we take you through the demand side of the oil market, including the main consumers of oil globally and the role that world Gross Domestic Product plays in driving changes in the global demand for oil over time. The demand for, and the cost of oil, are important inputs into a model of the global oil market called the dominant-firm/fringe-firm model, which attempts to capture the impact of OPEC (Organisation of Petroleum Exporting Countries) on the oil market. This model will be analysed in detail in chapter 12 and will be used to understand the changes in the price of key oil benchmarks, such as Brent and West Texas Intermediate crude, over time. Crude oil is refined into jet kerosene, which is the product that is consumed by commercial aircraft. Chapter 12 analyses the process of converting crude oil into refined products such as jet kerosene and the forces that determine the margin between the price of oil and the price of jet kerosene. I hope you enjoy reading this book as much as I have enjoyed writing it. Most importantly I hope you learn something, and that you can use it in your first or next job in aviation.

Quiz 1.1 Why it is Difficult to Consistently Make Money in Aviation? 1. (a) (b) (c) (d)

In which of the following airlines did Richard Branson NOT have an equity interest? Virgin Atlantic. Virgin Australia. Virgin Singapore. Virgin America.

2. (a) (b) (c) (d)

Which of the following reasons is behind equity investor guru Warren Buffet’s view that investors should steer away from investing in airlines? He was a CEO of an airline and therefore understands airline industry history of poor returns. He was a pilot and therefore understands how volatile the passenger and freight loads are likely to be on aircraft. He invested in the airline U.S. Air and lost money. Airline share prices are negatively correlated with global economic growth.

3. (a) (b) (c) (d)

What has Warren Buffet indicated as one of the potential reasons why it is difficult to make money in airlines? It is energy intensive, and subject to fluctuations in the price of energy. Aircraft must be imported and therefore subject to exchange rate fluctuations. It is labour intensive and expensive to buy labour. It is capital intensive, and expensive to buy aircraft.

4. (a) (b) (c) (d)

Which of the following is NOT a source of capital for funding aircraft investments? Raising funds from airline owners. Borrowing money from financial institutions. Using the proceeds of tax revenue raised by Government. Using cash reserves that have been built-up because of the airline’s operations.

Chapter 1

10

5. (a) (b) (c) (d)

What is the approximate average life of an aircraft? 10 years. 15 years. 20 years. 25 years.

6. (a) (b) (c) (d)

What must an airline forecast into the future to construct a business case for an aircraft investment? Revenue and cost streams generated by the aircraft’s operations. The share price of the airline. The weather patterns 20 years ahead. Whether there will be an event such as a Coronavirus over the life of the aircraft.

7.

Which of the following is NOT an uncontrollable macroeconomic variable that influences the revenue stream of an aircraft investment? Economic Growth. Share market movements. Exchange rate movements. Jet fuel price movements.

(a) (b) (c) (d)

CHAPTER 2 THE LANGUAGE OF THE AIRLINE BUSINESS

You are reading this textbook because you are hoping to one day work in aviation – for an airline, an airport, a safety regulator, an air services operator, or a government agency whose focal point is air transport or tourism. When you start your new job in aviation, one of the very first things that you will notice is the number of acronyms that are used. In fact, the first few days, weeks and months can be incredibly daunting because you are listening to conversations that are full of these acronyms, many of which will be completely foreign to you – almost like a different language. ASKs, RPKs, ATKS, RTKs, RFTKs, AFTKs, IATA, ICAO, Yield, EBIT, PAT …… the list is endless. One of the first things that I did when I started at Qantas back in 2004 as the General Manager of Microeconomics was to sit for a three-day course in the language of aviation and aviation data. It was incredibly useful because after the course, I began to better understand conversations and discussions in meetings, and eventually I could become more deeply involved in those conversations and discussions. Learning the acronyms and language short cuts used in aviation will not happen overnight, however, because it is a slow–absorber exercise. I estimate that to be incredibly competent in the language of aviation it will take about 6 months. I start-off this chapter by taking you through the operational language and metrics of aviation. This includes available seat kilometres (ASKs) and revenue passenger kilometres (RPKs), two of the most frequently used operational metrics in aviation. I will also introduce you to some operational metrics that are not very frequently used but are still useful, such as Revenue Passenger Tonne Kilometres (RPTK). Where these metrics need to be calculated, I will also show you how this is done, and will demonstrate these calculations using real-life examples based on data that is published by airlines. I will then take you through the most important financial language and metrics used in aviation. This includes the language that is used to describe the prices that passengers pay, such as RASK and PRASK. It will also involve examining the different measures that are used to understand an airline’s unit cost, such as CASK, CASM and CATK. I will then examine the different measures that airlines and airports use when they are assessing the earnings performance of the airline, such as EBIT, EBITDAR and PBT. Enjoy chapter 2. Make sure after you have read it that you strike-up an aviation conversation with someone who works for an airline, otherwise use it to impress people at aviation society events!

2.1 Key Airline Operational Metrics and Notation Table 2-1 below presents a summary of the main operational metrics used as language by airline management and analysts, including the abbreviations that are used in written and oral communications. The content of this table will be discussed at length in the subsections below. Airline management and analysts use these operational metrics so that they can be precise about what they are talking about, and so they can have efficient conversations that cover as much detail as possible in the least time possible. It is imperative that airline workers starting their aviation career learning terms like this. If you do, you will hit the ground running on day 1.

2.2 Load Metrics 2.2.1 Passengers Carried (PAX) Passengers carried or PAX measures the number of passengers carried over a single leg. A leg in air travel involves an aircraft flying from one airport to another, for example travel from Sydney (SYD) to Melbourne (MEL), or London Heathrow (LHR) to Dublin (DUB), or Singapore (SIN) to Beijing (PEK) represents one leg.2 An airline that flies from LHR to DUB and return, which we can abbreviate by LHR-DUB-LHR, involves two flight legs – the first leg being LHRDUB and the second leg being DUB-LHR. An airline that flies from Sydney to Los Angeles (LAX) and then from LAX to John F. Kennedy New York Airport (JFK) and then flies on a return service from JFK to LAX and then LAX to SYD flies 4 legs. Leg 1 is SYD-LAX, leg 2 is LAX-JFK, leg 3 is JFK-LAX and leg 4 is LAX-SYD.

2

In this illustration you will notice that I have used abbreviations for the airports. For example, I have used the abbreviation SYD for Sydney Airport. This abbreviation represents the three-letter code for airports developed by the International Air Transport Association (IATA), which is a trade association for the world’s airlines. It is important that you learn as many of these airport codes as possible, which is why I have introduced you to these airport abbreviations in this chapter and will continue to use these abbreviations throughout this textbook.

Chapter 2

12

Operation Metric

Abbreviation

Load Metrics Passengers Carried

PAX

Revenue Passenger Kilometres/Miles

RPK/RPM

Revenue Passenger Tonne Kilometres/Miles

RPTK/RPTM

Revenue Freight Tonne Kilometres/Miles

RFTK/RFTM

Freight Carried

FRT

Revenue Tonne Kilometres/Miles

RTK/RTM

Aircraft Movements

MOV

Capacity Metrics Seats Carried

SEATS

Available Seat Kilometres/Miles

ASK/ASM

Available Passenger Tonne Kilometres/Miles

APTK/APTM

Available Freight Tonne Kilometres/Miles

AFTK/AFTM

Available Tonne Kilometres/Miles

ATK/ATM

Block Hours

HOURS

Fleet Units

FLEET

Utilisation Passenger Seat Factor

PSF

Freight Load Factor

FLF

Total Load Factor

TLF

Labour Metrics Block Hours

HOURS

Full-Time Equivalent Staff

FTE

Productivity Metrics Block Hours per Fleet Unit per Day Available Seat Kilometres/Miles per Barrel of Fuel Consumed Available Seat Kilometres/Miles per Full Time Equivalent Staff

HOURS FLEET ൈ 365 ASK QFUEL ASK FTE

Table 2-1: Abbreviations used for Airline Operational Metrics As an example of how we calculate passengers carried over different legs, consider the simple case of a single passenger who flies from SYD to MEL and then the return journey MEL to SYD, as represented in Figure 2-1 below. One-Way trip = 1 passenger carried

SYD

MEL Return trip = 1 passenger carried

Fig. 2-1: Passengers Carried and Legs

The Language of the Airline Business

13

This passenger flies on two legs on her journey. The first leg is SYD-MEL, and the second leg is MEL-SYD. This passenger’s journey is counted as two passengers carried because the same passenger flies on two legs. There will be passengers travelling by air between SYD and MEL who only travel in one direction. This means there will be passengers travelling from SYD to MEL only, and passengers travelling only from MEL to SYD. These one-way passengers represent one passenger carried because they fly on just one leg. A passenger’s journey can include multiple legs flying between more than two airports. For example, a passenger who flies from Singapore (SIN) to Kuala Lumpur (KUL) to Jakarta (CGK) and then back to SIN flies on a journey from SIN to CGK and then back to SIN that involves three legs. SIN-KUL is the first leg; KUL-CGK is the second leg; and CGK-SIN is the last leg. For a passenger that travels on each of these legs, this is recorded as 3 passengers carried or 3 PAX. Consider another example. A passenger travels from Hobart (HBA) to Perth (PER) via MEL and returns with a direct flight to HBA from PER – refer to Figure 2-2 below which illustrates this passenger’s journey.3 1 PAX

MEL

HBA

2 PAX

PER 3 PAX

Fig. 2-2: Passengers Carried and Multiple Legs The journey presented in Figure 2-2 represents 3 passengers carried or 3 PAX – the first leg travelled by the passenger between HBA and MEL is one passenger carried; the journey from MEL to PER is two passengers carried; and the return leg from PER to HBA represents 3 passengers carried. The key complication that we experience when calculating passengers carried occurs because each flight, or each leg that is flown by an airline, carries passengers that are at different stages of their journey. To see what I mean by this statement, consider the following more complicated example. Qantas Mainline flies the QF1 service from Sydney (SYD) to London Heathrow (LHR) via Singapore (SIN).4 The airline carries passengers who board the flight in SYD and fly all the way to LHR. It carries passengers who board the flight in SYD and disembark in SIN, and it flies passengers who board in SIN and disembark in LHR. To determine the PAX for the flight, QF1, the airline must split the flight into legs, specifically SYD to SIN and SIN to LHR, and aggregate over the passengers that fly on each leg. For example, let us assume for this flight that there are 200 passengers who board the aircraft in Sydney and fly all the way to London Heathrow, there are 150 passengers who board the aircraft in Sydney and disembark in Singapore, and there are 75 passengers who board the aircraft in Singapore and disembark in London Heathrow. This means that the PAX carried for the QF1 flight is: Passengers CarriedQF1 = 200u2 + 150 + 75 = 625 We can see in this case that the QF1 service carries 200 passengers over two legs, SYD-SIN and SIN-LHR, which is why we multiply these 200 passengers by two. The remaining 225 passengers are carried over just one leg. The following two numerical examples will further describe some of the complexities associated with calculating passengers carried. Numerical Example 1 Air Asia X flight D7 206 departs Kuala Lumpur (KUL) at 23.55 and arrives on the Gold Coast (OOL) at 09:55+1 with a total flying time of 8 hours.5 Air Asia X is a long-haul low-cost carrier that is owned by the airline Group Air Asia and is based in Malaysia with affiliations in Thailand and Indonesia. The airline operates an A330-300 aircraft on KUL-OOL with 377 seats. The same aircraft type is used for the return flight D7 207 departing OOL at 22.20 and arriving in KUL at 04:55+1. Both flights are operated daily. On average the airline fills 78% of its seats with passengers over the space of a year on both flights combined. What is the total number of passengers carried by Air Asia flights D7 206 and D7 207 over the space of a year? The answer to this question is: 3 Hobart is the capital city of the state of Tasmania in Australia, Melbourne is the capital city of the State of Victoria in Australia, and Perth is the capital city of the State of Western Australia in Australia. 4 QF is the two-letter designator, or airline reservation code, that has been developed by IATA to describe flights by Qantas Mainline. Throughout this textbook I will introduce you to the IATA airline codes for a wide variety of airlines. It is important that you learn as many of these codes as possible. There is nothing more impressive in my view than a person beginning a job with an airline who has a strong knowledge of these two-letter airline flight codes and three-letter airport codes. 5 The arrival time notation 09:55+1 indicates that the flight has arrived at 09:55 on the day after the day of departure. The arrival time notation 09:55+2 indicates that the flight arrives at 09:55 two days after the day of departure.

Chapter 2

14

Passengers CarriedD7 206+D7 207 = 365 u 2 u 377 u 0.78 = 214,664 You will note in this answer that we use the term 365 u 2 because the flight involves a daily service and there are 365 days in one year, and the daily service involves a return flight, which is why we multiply by 2. Numerical Example 2 Qantas flight QF11 leaves Sydney at 09:50 and arrives in New York (JFK) at 16:40 while the QF12 flight leaves New York at 18:10 and arrives in Sydney at 06:20 + 2. Both flights use the Airbus A380-800 aircraft. The Qantas A380-800 has 484 available seats on board. The service is a daily service. An average QF11 flight picks up 400 passengers in Sydney, drops-off 200 passengers in Los Angeles (LAX), and picks up 100 passengers in LAX destined for JFK. An average QF12 flight picks up 300 passengers in New York, drops off 100 passengers in Los Angeles and picks up 200 passengers in Los Angeles who have Sydney as their final destination. How many passenger movements in total are there throughout the year for the QF11 and QF12 services? Figure 2-3 presented below may help you to answer this question as it describes the flow of passengers on the various legs between Sydney, Los Angeles, and New York. 400 PAX

LAX

300 PAX

SYD

JFK 300 PAX

400 PAX LAX

Fig. 2-3: Passengers Carrie across Multiple Legs – Sydney, Los Angeles, and New York The answer to the question is: Total Passengers CarriedQF11+QF12 = 365 u (400 + 300 + 300 + 400) = 511,000 You will notice in this illustration that it is not necessary to multiply by 2 because we have split the passenger numbers up into one-way and return passengers.

2.2.2 Aircraft Movements (MOV) An aircraft movement is an aircraft landing or take-off. When an aircraft lands this is referred to as one aircraft movement and when an aircraft takes-off this is also referred to as one aircraft movement. When an aircraft takes-off from São Paulo Guarulhos Airport (GRU) and lands at Rio de Janeiro Galeão Airport (GIG) this represents two aircraft movements. Movements can be scheduled or unscheduled. A scheduled movement is one in which seats are offered for sale to the public at regular time intervals. These are often referred to as RPT (regular passenger transport) services. For example, Qantas flight QF409 from Sydney to Melbourne leaves at 09:30 on every day of the week at the time of writing – this represents a scheduled service or an RPT service. Just because a flight is scheduled, however, does not mean that it is flown. Sometimes an airline will cancel a scheduled RPT service, in which case the number of services that are scheduled will exceed the number of services that are flown. An unscheduled service is also referred to as a charter service, where an individual, a group of individuals or a company approach an airline to fly them by air from port A to port B at an agreed time and date that is not part of a regular service. For example, the Australian airline, Alliance Airlines (IATA code QQ), transports employees who work in the Goonyella Riverside Mine in the town of Moranbah, which is in the state of Queensland Australia, between Moranbah and the capital of Queensland, Brisbane, where many of the mine employees live. One of the flights on this route is the QQ 4351 service, which is chartered by the mining company BHP (Goonyella Riverside Mine 2002). When an airline on an RPT service stops at a midpoint this provides added complexity to the determination of aircraft movements. In this case the airline has 4 movements per return service. For example, an Emirates (EK) flight from LHR to MEL via Dubai (DBX) represents four aircraft movements as there is a take-off and landing for the LHR-DXB leg and a take-off and landing for the DXB-MEL leg. Emirates Airline is one of the biggest full-service carriers in aviation, it is owned by the Government of the United Arab Emirates and operates out of headquarters in Dubai. Aviation analysts also use alternative language to describe aircraft take-off and landing. When an airline uses the word flights or departures or sectors flown these are one half of an aircraft’s movements because flights, departures and sectors flown only reference the departure of a flight and not the landing. For example, on the Emirates flight LHRDXB-MEL the numbers of flights, departures and sectors flown is 2, while there are 4 aircraft movements. On a flight

The Language of the Airline Business

15

from Houston (IAH) to Orlando (MCO) and back to IAH there are four aircraft movements (two take-offs and two landings) but just 2 sectors flown, departures and flights. A sector flown, or simply a sector, is a flight that departs from a particular port and lands at a particular port. It is the same as a leg. For example, when an airline flies from LHR to DXB and then from DXB to Auckland (AKL), the LHR to DXB flight is a sector with a certain sector length, while the flight DXB to AKL is a sector with a certain sector length. If we add together the LHR-DXB sector length and the DXB-AKL sector length and divided by 2 then we obtain the average sector length for the journey LHR-DXB-AKL.

2.2.3 Revenue Passenger Kilometres (RPK) RPKs is a measure of the distance that revenue passengers are carried. It is the most popular measure of passenger load that is used by airlines. RPKs are defined according to the following formula: RPK = PAX u PASL

(2.1)

where PASL = average distance flown by the passenger aircraft, or it is often referred to as the passenger average sector length (PASL). RPKs will be higher when an airline either carries more passengers over the same distance, or it carries the same passengers over a longer distance. To illustrate how RPKs are calculated using (2.1), suppose that Jetstar (JQ) carries 100 passengers between SYD and MEL. Jetstar is a low-cost carrier that is a part of the Qantas Group, flying services domestically in Australia, New Zealand and Japan, and international services out of bases in Australia and Singapore. The JQ RPKs for the SYD-MEL flight is 100 multiplied by the flying distance between Sydney and Melbourne, which is 706km – see Figure 2-4 below. JQ RPK = 100 u 706 km = 70,600

SYD

MEL

Fig. 2-4: Revenue Passenger Kilometres If 9m return passengers are carried by all airlines per year on SYD-MEL, which is approximately the number of passengers carried on the city pair over the space of a year in 2019, then the RPKs for the SYD-MEL route over a year using (2.1) will be: RPKSYD-MEL = 9,000,000 × 706 = 6.354b RPKs is a superior passenger load statistic then passengers carried because the unit of output sold by the airline is the number of kilometres travelled by a passenger. It is also the case that the airline’s costs depend heavily on the distance that it carries seats and passengers. Having a unit of passenger load that varies with distance therefore also aligns with the dominant driver of costs in the airline business. We often wish to find the average distance that an airline flies on its passenger network. This is also referred to as the PASL. The average length of a passenger sector for an airline’s network can be determined if we know the RPKs and PAX for that network. Specifically, if we divide RPKs by PAX we obtain the PASL, or the average distance that passengers are flown: Passenger Average Sector Length =

ୖ୔୏ ୔୅ଡ଼

ൌ

୔୅ଡ଼ൈ୔୅ୗ୐ ୔୅ଡ଼

ൌ PASL

(2.2)

One of the best websites for determining the flying distance between two city pairs is the Great Circle Mapper website, which can be found at the address http://www.gcmap.com/. All you need to do to find the flying distance between two cities using this website is to click on the web address above, insert the airport codes for the city pair in question with a hyphen between the airport codes (e.g. LHR-DXB) in the white box that appears at the top left of the website, press on the button “Distance” which is located towards the top left of the page, and then press on the “Calculate” button towards the bottom left of the webpage. After you click on the calculate button the website will provide the great circle distance for the city pair. There is also a Units dropdown box in Great Circle Mapper to the right of the calculate button that can convert the distance in nautical miles to a distance that is in kilometres and vice-versa. If you would like to see a map of the city pairs then press on the button called “Map” rather than button “Distance” at the top left-hand corner of the website. If you are unsure about the airport code, (e.g., LHR for London Heathrow), the Great Circle Mapper can provide this information for you. Simply type in the name of the airport in the white box that appears at the top left of the website, and Great Circle Mapper will provide a variety of airport options that include this name. Included in this information will be the three letter IATA airport code. If you click over the top of this airport code Great Circle Mapper will provide a map of the location of the airport, which you can use to determine if the airport that you have selected is correct.

Chapter 2

16

The Great Circle Mapper website generates the circular or arced distance between two cities. It does not generate the straight-line distance between two cities. The circular or arced distance is relevant to flying rather than the straightline distance because an arc, or part of the outside of a circle, is a better approximation to the flight path that an airline takes when it flies between airports. This is the case for two reasons. The first is that the three key stages of flying, namely take-off, cruise and then descent to landing is better approximated by an arc than a straight line. The second reason, which is more relevant to longer distances, is that the earth is round and therefore the flight path between two cities a long distance apart is an arc rather than a straight line.

2.2.4 Cargo Carried (FRT) Cargo carried is the number of tonnes of freight carried over a single leg and is described by the shorthand notation FRT. Cargo can be held in the belly of a passenger aircraft or in the body of dedicated freighter aircraft, or an aircraft that only carries freight. Most freight carried around the world is in the belly of passenger aircraft because of the cost efficiencies that this generates. Cargo carried is a freight load metric that is independent of the distance that the freight is flown and dependent on the weight of the freight in tonnes. Figure 2-5 below presents time series data at an annual frequency between 2003 and 2020 of the cargo carried by one of the biggest commercial airline carriers of air freight in the world, Singapore Air Cargo or SIA Cargo. Singapore Airlines Cargo Carried (m kg) 1350 1250 1150 1050 950 850 750 2020

2019

2018

2017

2016

2015

2014

2013

2012

2011

2010

2009

2008

2007

2006

2005

2004

2003

Source: Airline Intelligence and Research Database 2021 Fig. 2-5: Singapore Air Cargo, Freight Carried We can see in 2020 that SIA Cargo carried just over 780 million kilograms of air freight, which is a significant drop in air cargo from the 1,245 million kilograms carried in 2019. The significant drop in SIA Cargo’s freight carried between 2019 and 2020 is largely because of the impact of the Coronavirus, which affected the number of passenger airline flights and thus the freight that is carried in the belly of passenger aircraft.

2.2.5 Revenue Freight Tonne Kilometres Revenue Freight Tonne Kilometres (RFTK) is the preferred metric that is used to describe the volume of freight that the airline carries. It is the preferred metric because it takes into consideration both the distance and the weight of the freight that is carried (revenue freight tonne kilometres). This is important because the ability of an aircraft to carry freight is affected by two sets of constraints – the physical space on the aircraft that is available to carry freight and the loaded weight that the aircraft can carry given its fuel capacity and engine technology. RFTKs is a metric that therefore takes into consideration both constraints. RFTKs is calculated according to the following formula: RFTK = FRT u FASL

(2.3)

where FASL or freight average sector length is the average distance that freight is carried. Equation (2.3) says that RFTKs is equal to the volume of freight in tonnes multiplied by the average distance that this freight is moved. The variable FASL in this case is defined in a similar way to the passenger average sector length or PASL in (2.2). As an example of calculating RFTKs, if 20 tonnes of freight are carried from SYD to Auckland (AKL), the number of RFTKs per flight using (2.3) is 20 times the distance between the two cities, which is 2,164 km, as shown in Figure 2-6 below.

The Language of the Airline Business

RFTK = 20 u 2,164 km = 43,280

SYD

17

AKL

Fig. 2-6: Revenue Freight Tonne Kilometres Calculation Consider another example of calculating RFTKs. If the airline carries 3 tonnes of freight between SYD and AKL on a single flight, then the RFTKs for that flight using (2.3) will be: RFTKSYD-AKL = 3×2,164 = 6,492 More complicated calculations of RFTKs arise when freight is carried over more than one sector, which often occurs. In this case it is necessary to determine the RFTKs for each sector and then add these RFTKs together. For example, let us suppose that 10 tons of freight is carried from Los Angeles (LAX) to Hong Kong (HKG) and then from HKG to Shanghai Pudong Airport (PVG). Using the Great Circle Mapper, the distance in kilometres between LAX and HKG is 11,684 km and the distance between HKG and PVG is 1,253 km. The RFTKs for the combined two sectors in this case is equal to: RFTKLAX-HKG-PVG = (11,684 + 1,253) u 10 = 129,370 It is also likely that an aircraft flies across multiple sectors, with different tonnes of freight carried over those different sectors because it sets down freight and picks up freight at different points in its journey. For example, let us suppose that an Emirates (EK) flight carries 15 tonnes of freight from Tokyo Narita (NRT) airport to Dubai (DXB) airport, sets down 5 tonnes of that freight in DXB, and then picks up another 8 tonnes of freight in DXB to carry to Addis Ababa Bole International Airport (ADD) in Ethiopia. There is a great circle distance of 7,994 km for the sector NRT-DXB and a great circle distance for the sector DXB-ADD of 2,515 km. In the case of the journey NRT-DXB-ADD the RFTKs for the combined two flights are: RFTKNRT-DXB-ADD = 15 u 7,994 + 18u2,515 = 119,910 + 45,270 = 165,180 Like the passenger average sector length, it is possible to find the freight average sector length for an airline’s freight network if that airline provides information about its freight carried and RFTKs. This is found by dividing RFTKs into the volume of freight carried as follows: Freight Average Sector Length =

ୖ୊୘୏

(2.4)

୊ୖ୘

For example, in 2020, Singapore Air Cargo carried FRT = 780.4 million kilograms of cargo and 4,321.3 million RFTKs. The FASL in the case of Singapore Air Cargo in 2020 can be found by dividing its RFTKs by its FRT. Before we do this however, it is necessary to convert its cargo carried into freight tonnes rather than kilograms because its RFTKs is measured in tonne kilometres not kilogram kilometres. The FRT of the airline in tonnes is found by dividing its freight in kilograms by 1,000 as follows 780.4/1,000 = 0.7804 million tonnes. The FASL using (2.4) therefore becomes: Freight Average Sector Length =

ସ,ଷଶଵ.ଷ ଴.଻଼଴ସ

= 5,537.29 km

The freight average sector length of Singapore Air Cargo operations in calendar 2020 is therefore 5,537 kilometres.

2.2.6 Revenue Passenger Tonne Kilometres (RPTK) This is a passenger load metric that includes not just the distance that passengers are carried but also the total weight that passengers carry onto the aircraft in tonnes. It is a statistic that is not routinely published by airlines, except for carriers from China and India, which routinely publish RPTK statistics. This metric has been designed so that passenger and freight loads can be measured using the same units (tonne kilometres) and therefore added together to generate a total load statistic (RTK or revenue tonne kilometres, which we will discuss in section 2.1.7 below). RPTK is defined as the average passenger weight that a passenger brings onto the aircraft multiplied by the RPKs that the passenger flies and can be described using the following formula: RPTK = RPK u APW

(2.5)

where APW is the average weight brought onto the aircraft by the passenger defined in tonnes, including own body weight, checked-in baggage, and cabin baggage.

Chapter 2

18

Often airlines don’t publish RPTK statistics, but they publish RTK and RFTK statistics from which the RPTK statistic can be derived. As mentioned earlier, RTK refers to revenue tonne kilometres and represents the sum of RPTK and RFTK. The derivation of RPTK if we know RTK and RFTK is simply: RPTK = RTK – RFTK

(2.6)

where RTK is revenue tonne kilometres, which we will define again in section 2.1.7 below. This expression simply says that the tonne kilometres that are attributable to passengers is the total tonne kilometres due to passengers and freight minus the tonne kilometres attributable to freight. If we know the RPTKs and we know the RPKs, then we can derive the average implied passenger weight, APW, using equation (2.5). To derive this weight in kilograms, we use the following formula: APW (kgs) =

ୖ୔୘୏ ୖ୔୏

× 1000

(2.7)

This formula says that if we wish to know the average passenger weight assumption that is used by an airline, and the airline provides data about its RPTKs and RPKs then to find the average passenger weight we divide the RPTKs by RPKs and multiply this by 1,000. When we multiply by 1,000 in equation (2.7) we are converting the result from tonnes into kilograms. The average passenger weight is an important statistic for airlines because it is an input into determining the total payload of the aircraft, which in turn is a necessary input into determining the total fuel that is required by the aircraft for its flight. The total payload of an aircraft is the total weight of the revenue-generating passengers and freight that are being carried by the airline. It is the total load of the aircraft that is paying for the airline’s costs. To determine APW, airlines will normally use an assumed value for the weight of the passenger. They will then add to this a value for the average weight of baggage that is checked-in by passengers plus the average weight of cabin baggage. Given the average weight of passengers has increased over the past two decades by virtue of increasing waistlines, particularly in large western countries, it has been necessary for airlines to vary the assumptions they use for passenger weight (Ironside 2016). For some airlines that carry a limited number of heavy people on their flights on relatively small aircraft, such as airlines that carry passengers between the Pacific Islands including Fiji, Samoa, and Tonga, where the weight of passengers is a high percentage of the total weight of the aircraft, there is often a need to weigh each passenger and charge each passenger according to the weight they bring onto the aircraft. If you wish to read more about this interesting topic, read the very good article by Smith 2016, which describes how Samoa Air (OL) was the first airline in the world to charge fares according to passenger weight.

2.2.7 Revenue Tonne Kilometres (RTK) Revenue Tonne Kilometres (RTK) is the statistic that is used to describe the total freight and passenger load of the airline. It is a distance and weight statistic, like RFTKs and RPTKs. It is defined according to the following identity: RTK = RPTK + RFTK

(2.8)

RTK is the total tonne kilometre load carried by the aircraft, including both the passenger and freight tonne kilometres. It is usually defined in terms of the revenue earning load, excluding both the passengers and freight that are carried ‘free of charge’ by the airline. It is reasonably unusual for an airline to provide this statistic, although the big Chinese and Indian carriers as well as the Aeroflot Group in Russia and Cathay Pacific in Hong Kong publish it routinely. Airlines in the US and Europe almost never publish data on RTKs although they calculate this metric for their own internal analysis.

Quiz 2-1 Load Metrics 1. Air New Zealand (NZ) flies a 168-gauge Airbus A320-200 aircraft with flight number NZ0123 between AKL and MEL. The flight leaves at 6.35pm and arrives at 8.40pm for a flying time of 3 hours and 5 minutes. The flight then leaves MEL the following morning at 8.40am on flight NZ0722 arriving in AKL at 2.15pm with a flying time of 3 hours and 35 minutes. There are no stops. The flight is a daily service. On average, three quarters of the seats are occupied over a year on the combined one-way and return flights. How many passengers are carried on these flights for Air New Zealand over the space of a year? (a) (b) (c) (d)

168 u 0.75 168 u 0.75 u 365 168 u 365 168 u 0.75 u 365 u 2

The Language of the Airline Business

19

2. Continuing with quiz question 1 in this section, use the Great Circle Mapper and the link below to find the air distance between Auckland and Melbourne (http://www.gcmap.com/). What are the revenue passenger kilometres for the year for the Air New Zealand Airbus A320-200 on flights NZ0123 and NZ0722 combined? (a) (b) (c) (d)

168 u 0.75 u 2,644 u 2 x 365 168 u 0.75 u 365 u 2,644 168 u 365 u 2,164 168 u 0.75 u 365 u 2 u 2,164

3. Jetstar flight JQ952 flies from Sydney to Cairns leaving at 6am and arriving at 9.10am on an Airbus A321 with 180 seats. Passengers then transfer from this flight to JQ15 which is destined for Osaka (KIX), departing at 1.00pm and arriving at 7.30pm on a Boeing B787-800 aircraft with 335 seats – refer to the Figure below which describes this journey. 160 PAX

300 PAX

CNS

SYD

KIX

Flight JQ952 picks-up 160 passengers in Sydney and flies them to Cairns. 100 of those passengers transfer to flight JQ15 to Osaka from Cairns. 200 passengers are picked-up in Cairns to fly to Osaka. Determine the number of passengers carried by the aircraft on the Sydney (SYD) to Osaka (KIX) market on flights JQ952 and JQ15. (a) 100 + 200 (b) 160 + 200 (c) (100 + 200) x 2 (d) 100 4. Continue to use the information in question 3 of this section to answer the following question. Use the great circle distance data obtained from http://www.gcmap.com/ to estimate the revenue passenger kilometres on flights JQ952 and JQ15. (a) (b) (c) (d) (e)

160u1966 + 300u5787 (200 + 160) u (1966 + 5787) 100u1966 + 200u5787 60 + 200 200 u 5787

5. China Eastern Airways provides the following actual monthly data on load in its monthly reporting (see the table below). Complete the two right-hand columns of the table.

Feb-16 Mar-16 Apr-16 May-16 Jun-16

RTK m

RFTK m

RPK m

1469.83 1608.49 1596.379 1593.433 1561.366

288.68 418.51 396.0407 400.0262 387.4611

13508.62 13327.87 13463.31 13366.82 13229.43

RPTK m

Average Passenger Weight (kg)

6. Cathay Pacific provides the following actual monthly operating data (you can find this data by Googling Cathay Pacific investor relations and searching for operating statistics). A component of this includes the freight business.

Chapter 2

20

Cathay Pacific Feb 2017 Traffic Figures CATHAY PACIFIC / CATHAY DRAGON COMBINED TRAFFIC

FEB 2017

% Change VS FEB16

Cumulative FEB 2017

% Change YTD

- Mainland China

720,595

-0.4%

1,489,925

2.2%

- North-East Asia

1,255,504

-0.4%

2,647,398

4.5%

- South-East Asia

1,381,220

-5.2%

2,885,652

-1.8%

544,860

-3.3%

1,171,540

-9.6%

- South-West Pacific & South Africa

1,505,617

-2.5%

3,112,385

-3.2%

- North America

2,325,064

-6.3%

5,317,846

-2.9%

- Europe

1,748,381

7.2%

3,875,387

9.1%

RPK Total (000)

9,481,241

-1.9%

20,500,133

0.1%

Passengers carried

2,681,050

-2.4%

5,647,895

0.1%

Cargo and mail revenue tonne km (000)

763,545

12.0%

1,626,498

5.9%

Cargo and mail carried (000kg)

137,674

17.4%

288,806

9.0%

6,113

-3.6%

12,920

-1.1%

RPK (000)

- India, Middle East, Pakistan & Sri Lanka

Number of flights

Using this information, what is the average sector length for the Cathay Pacific freight business in the month February 2017 ଵଷ଻,଺଻ସ×ଵ଴଴଴ (a) (b) (c) (d)

଻଺ଷ,ହସହ ଻଺ଷ,ହସହ×ଵ଴଴଴ ଵଷ଻,଺଻ସ ଻଺ଷ,ହସହ ଵଷ଻,଺଻ସ ଵଷ଻,଺଻ସ

଻଺ଷ,ହସହ×ଵ଴଴଴

7. The average passenger weighs 80kg, carries 5kg onto the plane as cabin baggage and carries 15kg onto the plane as checked-in baggage. Using this and the Cathay Pacific data, what is the RTKs for the airline in February 2017 in millions of tonne kilometres? (a) 763.545 + (b) 763.545 + (c) 763,545 + (d) 763,545 +

ଽ,ସ଼ଵ,ଶସଵ×(଼଴ାହାଵହ) ଵ,଴଴଴ ૢ,૝ૡ૚.૛૝૚×(଼଴ାହାଵହ) ଵ଴଴଴ ଽ,ସ଼ଵ.ଶସଵ (଼଴ାହାଵହ) ଽ,ସ଼ଵ.ଶସଵ (ఴబశఱశభఱ) భబబబ

2.3 Capacity Metrics 2.3.1 Seats Carried (SEATS) Seats Carried represents the total number of seats carried over a single leg. It is the capacity equivalent to passengers carried. For example, if a Boeing 737-800 with 174 seats flies from Brunei’s Bandar Seri Begawan Airport (BWN) to the main airport in East Timor, which is Presidente Nicolau Lobato International Airport (DIL) then the seats carried for that sector flown will be 174. A return journey represents two seats carried. If our flight above flies the return journey from DIL back to BWN then the return journey will include 348 seats carried. In the above example we have computed the seats carried for a single flight. We often wish to determine the seats carried over a particular period, such as a year or quarter. To calculate the seats carried over a particular period, we need information about the number of flights or aircraft movements of the airline over the relevant period. The following formula will allow you to determine the number of seats carried (SEATS) over a particular period given the number of aircraft movements:

The Language of the Airline Business

SEATS = Average Seat Count u MOV y 2

21

(2.9)

where Average Seat Count is the average number of seats on the plane, or it is sometimes referred to as the average aircraft gauge. As the number of aircraft movements is equal to the number of flights multiplied by 2 then we can also write (2.9) as: SEATS = Average Seat Count u Flights

(2.10)

For example, if the average seat count of an aircraft is 180 and the aircraft makes 20 flights or flies 20 sectors, which is equivalent to 40 aircraft movements, then the seats carried is equal to: SEATS = 180 × 20 = 3,600 The seats carried includes all revenue earning seats on the aircraft, excluding cabin crew seats and seats in the cockpit. Numerical Example Let us return to our Air Asia X example from section 2.2, which I will now repeat for your convenience. Air Asia X flight D7 206 departs KUL at 23.55 and arrives at OOL at 09:55+1 with a total flying time of 8 hours. The airline operates an A330-300 aircraft on the route with 377 seats. The same aircraft type is used for the return flight D7 207 departing OOL at 22.20 and arriving in KUL at 04:55+1. Both flights are operated daily. Instead of determining the number of passengers let us now determine the number of seats carried over a year. This is equal to: SEATS = 365 u 377 u 2 = 275,210

2.3.2 Available Seat Kilometres (ASK) This is the most popular metric used to measure an airline’s passenger capacity. It represents the sum over the distance that an aircraft carries each available seat and is computed using the following formula: ASK = SEATS u PASL

(2.11)

If we do not know the seats carried but we do know the aircraft that is used on a flight and the number of seats this aircraft carries, and we know the frequency of the aircraft’s services then we can determine ASKs using a variation of the formula at (2.11). If we substitute the seats carried at (2.10) into (2.11), then we obtain a formula for ASKs as a function of the average seat count of the aircraft and the number of flights as follows: ASK = Average Seat Count × Flights u PASL

(2.12)

Equation (2.12) tells us that ASKs can increase because an aircraft flies a longer distance (PASL is higher), or a bigger aircraft with more seats is used (Average seat count or gauge is higher), or the aircraft has increased the frequency of services (flights is higher). Aviation management and analysts often refer to the use of a bigger aircraft as up-gauging. Numerical Example Returning to our Air Asia X example once again, let us estimate the annual number of ASKs flown by flights D7 206 and D7 207. To do this, we must first determine the great circle distance between OOL and KUL. Using http://www.gcmap.com/ this is found to be 6,506 km. We can now find the yearly ASKs for these Air Asia X services: ASK = 275,210 u 6,506 = 1,791m More examples of finding ASKs will be provided in Quiz Questions 2-2 below.

2.3.3 Available Freight Tonne Kilometres This is the most popular metric for the capacity to carry air freight over a certain distance. Available Freight Tonne Kilometres (AFTKs) represents the sum over the distance that an aircraft carries each available tonne of freight. This statistic is determined according to the maximum amount of freight that is available for the aircraft to carry, multiplied by the average distance that the aircraft carries this available freight. AFTKs is not a regular metric that is published by airlines, but like many of the weight related metrics discussed in this section and section 2.2 the large Chinese and Indian carriers routinely publish this information, as do the airlines

Chapter 2

22

where freight activity is a relatively important part of their business, such as Cathay Pacific, Korean Air and Singapore Airlines.

2.3.4 Available Passenger Tonne Kilometres (APTK) This is a measure of the capacity of the aircraft to carry the weight of the aggregate of passengers over a certain distance. It is a statistic that is not published by many airlines. APTK is determined according to the following formula: APTK = APW u ASK

(2.13)

The statistic at equation (2.13) is simply the seat capacity of the aircraft defined in kilometre terms multiplied by the average weight brought onto the aircraft by the passenger in tonnes, including own body weight, checked-in baggage weight and cabin baggage weight. We can also compute the APTK if we know the RPTK, and the passenger seat factor as follows: APTK =

ୖ୔୘୏ ୔ୟୱୱୣ୬୥ୣ୰ ୗୣୟ୲ ୊ୟୡ୲୭୰

=

୅୔୛×ୖ୔୏ ୔ୟୱୱୣ୬୥ୣ୰ ୗୣୟ୲ ୊ୟୡ୲୭୰

= APW × ASK

(2.14)

Equation (2.14) tells us that we can also find APTKs by dividing RPTKs by the passenger seat factor. We can also determine APTKs if the airline provides information about total available tonne kilometres or ATKs and AFTKs. As will be shown in section 2.3.5 below, ATKs is the sum of APTKs and AFTKs and represents the available capacity of the aircraft to carry both passengers and freight combined. It follows that if an airline provides information about ATKs and AFTKs then APTKs can be determined according to the following formula: APTK = ATK – AFTK

(2.15)

Numerical Example In our Air Asia X example for services between OOL and KUL, we assume that the average passenger weighs 85kg, carries onto the plane 8kg in cabin baggage and 20kg in checked-in baggage. The APTK for the year for flights D7 206 and D7 207 is calculated in the following way (note that we are also using our ASK estimate from the previous numerical example): APTK = 1,791 u (85+8+20)/1000 = 202 m Note that we divide by 1,000 in this example to convert the combined weight of passengers and their luggage into tonnes from kilograms.

2.3.5 Available Tonne Kilometres (ATK) ATKs are the total available capacity of the airline including both passengers and freight. It is a popular measure of capacity for airlines in which both passengers and freight are an important part of the business, which is the case for Singapore Airlines, Cathay Pacific, and Korean Air. ATK is defined according to the following identity: ATK = APTK + AFTK

(2.16)

ATKs is therefore the sum of passenger capacity and freight capacity. It is more likely that an airline will provide information about ATKs and AFTKs and not APTKs. We can also derive ATKs if an airline provides information about AFTKs and ASKs. This can be done if an assumption is made about the average passenger weight or APW, in which case APTKs equal AFTKs plus ASKs multiplied by APW.

2.3.6 Block Hours (HOURS) This is a measure of the total time that it takes for a plane to leave one gate and arrive at another. It is usually described by aviation people as the difference between the time at which the chocks are taken away from the wheels of the plane at the point of origin and the time the chocks are placed around the wheels of the plane at the destination. The total number of block hours for a route can be computed using the following formula: HOURS = MOV u AVR Time y 2

(2.17)

where AVR Time is the average flight time. For example, on a Hainan Airlines (HU) flight from Shanghai (PVG) to Beijing (PEK), leaving at 10:55 and arriving at 13:35 on flight HU 7612, the total flight time is 2 hours and 40 minutes.

The Language of the Airline Business

23

Hainan Airlines is one of the big four full-service carriers that is domiciled in mainland China, with headquarters in Haikou in Hainan province. If Hainan Airlines completes 365 of these flights in one year, how many block hours does Hainan flight HU 7612 fly? The answer using (2.17) is: HOURS = 365 × (2 + 40/60) = 973 hours Note that block hours differ from flying hours. Flying hours represents the time that aircraft are in flight whereas block hours represent the time that the aircraft is on the move, which includes both flying time, and time in the apron area, on the taxiways, queuing for the runway and the runway itself.

2.3.7 Fleet Units (FLEET) The number of fleet units is simply defined as the total number of aircraft that are owned or leased by the airline and operated by the airline. The airline may also own aircraft that it leases to other airlines, or simply not used (taken out of service). These aircraft are in the airline’s fleet but not operated by the airline. The number of fleet units is either defined at the end of a period, or sometimes an airline will provide the period average number of fleet units. Airlines will often report the number of aircraft that are under operating lease, capital lease or owned. They will also often distinguish between passenger aircraft and aircraft that are dedicated freighters. Dedicated freighters will often be represented by the letter F after the description of the aircraft type. For example, Boeing 767-300F is a Boeing 767-300 aircraft type that is a dedicated freighter. As an illustration, All Nippon Airways (ANA) release their fleet information on a quarterly basis in their quarterly financial presentations as supplementary material. Table 2-2 below is an example of how ANA present their fleet information. ANA is a full-service carrier that is a part of the big two airlines domiciled in Japan, with headquarters in Tokyo. Boeing 777-300ER Boeing 777-300 Boeing 777-200ER Boeing 777-200 Boeing 787-9 Boeing 787-8 Boeing 767-300ER Boeing 767-300 Boeing 767-300F Boeing 767-300BCF Airbus A321-200 Airbus A320-200neo Airbus A320-200 Boeing 737-800 Boeing 737-700ER Boeing 737-700 Boeing 757-500 Bombardier DHC-8-400 Total

Mar 31, 2016 22 7 12 16 11 35 25 13 4 8 0 0 18 36 2 7 20 21 257

Dec 31, 2016 22 7 12 14 21 36 25 12 4 8 3 1 21 36 0 7 18 20 268

Owned 16 7 6 12 21 31 13 12 0 8 0 1 10 24 0 7 18 20 206

Leased 6 0 6 2 0 5 12 0 4 0 3 0 11 12 0 0 0 1 62

Source: ANA Financial Results Presentation 2016, Page 28

Table 2-2: All Nippon Airways Aircraft in Service March 2016

Quiz 2-2 Capacity Metrics 1. Norwegian Air Shuttle flies a 189 seat Airbus A320-100 aircraft between London-Gatwick (LGW) and Copenhagen (CPH) on a daily return service. What are the available seat kilometres for this service over a year? (a) 189 x 365 x 2 (b) 189 x 987 x 365 (c) 189 x 987 x 365 x 2 (d) 189 x 0.75 x 987 x 365 Consider the February 2017 load and capacity operating figures of Cathay Pacific Airlines in the table below, which includes load-related operating statistics presented earlier in this chapter. Use this table of information to answer the following 4 questions.

24

Chapter 2

Cathay Pacific Feb 2017 Traffic Figures CATHAY PACIFIC / CATHAY DRAGON COMBINED TRAFFIC

FEB 2017

% Change VS FEB16

Cumulative FEB 2017

% Change YTD

- Mainland China

720,595

-0.4%

1,489,925

2.2%

- North-East Asia

1,255,504

-0.4%

2,647,398

4.5%

- South-East Asia

1,381,220

-5.2%

2,885,652

-1.8%

544,860

-3.3%

1,171,540

-9.6%

- South-West Pacific & South Africa

1,505,617

-2.5%

3,112,385

-3.2%

- North America

2,325,064

-6.3%

5,317,846

-2.9%

- Europe

1,748,381

7.2%

3,875,387

9.1%

RPK Total (000)

9,481,241

-1.9%

20,500,133

0.1%

Passengers carried

2,681,050

-2.4%

5,647,895

0.1%

Cargo and mail revenue tonne km (000)

763,545

12.0%

1,626,498

5.9%

Cargo and mail carried (000kg)

137,674

17.4%

288,806

9.0%

6,113

-3.6%

12,920

-1.1%

FEB 2017

% Change VS FEB 16

Cumulative FEB 2017

% Change YTD

- Mainland China

923,757

1.1%

1,937,420

2.7%

- North-East Asia

1,520,166

-5.6%

3,216,321

-1.3%

- South-East Asia

1,639,353

-5.7%

3,436,117

-2.3%

673,523

-9.9%

1,434,481

-12.5%

- South-West Pacific & South Africa

1,671,253

-3.1%

3,516,104

-1.4%

- North America

2,933,593

-4.3%

6,269,089

-2.6%

- Europe

2,012,497

4.4%

4,305,933

7.6%

ASK Total (000)

11,374,142

-3.0%

24,115,465

-0.8%

83.4%

1.0pt

85.0%

0.8pt

1,189,317

1.2%

2,582,911

0.9%

64.2%

6.2pt

63.0%

3.0pt

2,270,666

-0.8%

4,875,994

0.1%

RPK (000)

- India, Middle East, Pakistan & Sri Lanka

Number of flights CATHAY PACIFIC / CATHAY DRAGON COMBINED CAPACITY ASK (000)

- India, Middle East, Pakistan & Sri Lanka

Passenger load factor Available cargo/mail tonne km (000) Cargo and mail load factor ATK (000)

2. What is the available passenger tonne kilometres flown by Cathay Pacific in February 2017 in thousands of tonne kilometres? (a) 2,270,666 (b) 2,270,666 – 1,189,317 (c) 11,374,142 u 0.1 (d) 2,270,666 – 11,374,142 (e) 11,374,142

The Language of the Airline Business

25

3. Using the load and capacity data, what is the implied average passenger weight in kilograms? (a) 2,270,666 – 1,189,317 (b) (2,270,666 – 1,189,317)u1000/11,374,142 (c) (2,270,666 – 1,189,317)u1000/9,481,241 (d) (2,270,666 – 1,189,317)u1000/11,374.142 4. What are the revenue passenger tonne kilometres in February 2017 in ‘000? (a) (2270,666 – 1,189,317) u 0.834 (b) (2270,666 – 1,189,317) (c) (2,270,666 – 1,189,317)u1000 (d) 9,481,241 u (2,270,666 – 1,189,317)/11,374,142 5. What are the revenue tonne kilometres in February 2017 in ‘000? (a) [9,481,241 u (2,270,666 – 1,189,317)u1000/9,481,241] + 763,545 (b) 9,481,241 u (2,270,666 – 1,189,317)u1000/11,374,142 + 763,545 (c) 763,545 + 0.09u9,481,241 (d) (2270,666 – 1,189,317) u 0.834 + 763,545 6. Read the following Qantas media release posted on December 16, 2016. Use the information in this article to determine the number of ASKs that will be flown per year on the new Melbourne-Tokyo Narita service. There are 297 seats on the Airbus A330-300. Use http://www.gcmap.com/ to determine the sector length for the route. (a) 297 u 8,144 (b) 297 u 8,144 u 2 (c) 297 u 2 u 365 (d) 297 u 8,144 u 2 u 365 QANTAS’ MELBOURNE-TOKYO FLIGHTS TAKE-OFF MELBOURNE | PUBLISHED ON 16TH DECEMBER 2016 AT 9:32 Australia’s booming tourism industry is set for another boost when Qantas’ first Melbourne-Tokyo Narita service takes off today. The new daily service is being launched in response to strong demand, as travel between Japan and Australia grows at around 20 per cent a year, and will be operated by an upgraded, twoclass Airbus A330-300 aircraft fitted with Qantas’ signature Marc Newson interiors. It takes the number of seats the national carrier offers on Japanese routes each week to 6,000, following the launch of new SydneyTokyo Haneda and Brisbane-Tokyo Narita routes in 2015 – with extensive onward connections in both countries. Inbound flights on the new route are timed to link smoothly to the rest of Qantas’ Australian domestic network, including popular destinations like Hobart, Adelaide, and Perth, while outbound services will connect with Jetstar’s extensive domestic network within Japan. Qantas International CEO Gareth Evans said the new route was the latest example of the national carrier growing its international network to meet demand and unlock opportunities for Australia in Asia. “The clear message from our Japanese customers was that they wanted a direct link into Melbourne – so we’re delighted to be opening up that gateway for them today,” Mr Evans said. “We’ve had a great response from Victorian tourism operators and the many companies in Melbourne who do business with Japan. “With a new Free Trade Agreement in place, there’s tremendous potential to build on the fantastic growth we’ve already seen in this market, and we want the Qantas Group to lead the way in making it happen.” Today’s route launch means Tokyo becomes the eighth international destination Qantas serves out of the Victorian capital. Flights to Christchurch began on 4 December, while Qantas confirmed earlier this week that Melbourne-Los Angeles would be the launch route for its game-changing Boeing 787-9 Dreamliner. Features of the Airbus A330 that serve the new route include the latest generation of its award-winning Recaro seat in Economy, a lie-flat Business Suite that can be reclined during take-off and landing, and more than 1,500 inflight entertainment options. QF79 is set to depart Melbourne at 9:15am and QF80 will depart from Narita at 7pm, with launch celebrations in both cities, including a water cannon salute and ribbon cutting ceremony.

26

Chapter 2

7. The table below is the operating data provided on page 34 of the Qantas Group’s half yearly report for the period ending December 31, 2016. Qantas Operational Statistics For the half-year ended 31 December 2016 (unaudited) Traffic and Capacity QANTAS DOMESTIC (INCLUDING QANTASLINK) Passengers Carried Revenue Passenger Kilometres Available Seat Kilometres Revenue Seat Factor

‘000 M M %

11,332 14,119 18,254 77.3%

11,220 14,189 18,536 76.5%

JETSTAR DOMESTIC Passengers Carried Revenue Passenger Kilometres Available Seat Kilometres Revenue Seat Factor

‘000 M M %

6,831 8,080 9,662 83.6%

6,962 8,273 9,750 84.8%

QANTAS INTERNATIONAL Passengers Carried Revenue Passenger Kilometres Available Seat Kilometres Revenue Seat Factor

‘000 M M %

3,317 26,643 32,756 81.3%

3,200 26,230 31,492 83.3%

JETSTAR INTERNATIONAL Passengers Carried Revenue Passenger Kilometres Available Seat Kilometres Revenue Seat Factor

‘000 M M %

3,135 9,188 11,007 83.5%

2,720 8,481 10,535 80.5%

JETSTAR ASIA Passengers Carried Revenue Passenger Kilometres Available Seat Kilometres Revenue Seat Factor

‘000 M M %

2,142 3,319 4,054 81.9%

2,109 3,480 4,337 80.2%

‘000 M M %

26,758 61,348 75,732 81.0% 308

26,211 60,652 74,650 81.2% 300

# ‘000 ‘000

30,179 4,066 5,019

29,353 4,133 5,086

QANTAS GROUP OPERATIONS Passengers Carried Revenue Passenger Kilometres Available Seat Kilometres Revenue Seat Factor Aircraft in Service EMPLOYEES Full-time equivalent employees at end of period (FTE) RPK per FTE (annualised) ASK per FTE (annualised)

December 2016

December 2015

The table has an assortment of operating statistics for the 6-month period. Identify the following statistics from this data. (a) How many passengers did Qantas Mainline carry over the 6 months to December 2016? (b) How many passengers did the Jetstar Group carry over the 6 months to December 2016? (c) Which Jetstar brand carried the most passengers over the 6 months to December 2016? (d) What is the share of Jetstar ASKs in total Qantas Group ASKs? How has this changed between the 6 months to December 2015 and the 6 months to December 2016? (e) Find the passenger average sector length for Qantas International over the 6 months to December 2016. (f) Determine the Qantas Mainline share of ASKs in the Group’s total domestic ASKs. (g) Determine the Jetstar international share of Jetstar Group international ASKs over the 6 months to December 2016 (include Jetstar Asia ASKs in Jetstar Group ASKs). (h) Determine the growth rate in Qantas domestic ASKs and compare this to the growth rate in Qantas domestic RPKs. (i) Find the share of Group domestic RPKs in Group total RPKs. How has this changed over the year?

The Language of the Airline Business

27

2.4 Capacity Utilisation Metrics 2.4.1 Passenger Seat Factor (PSF) The passenger seat factor represents the percentage of seats on the aircraft that are occupied by revenue earning passengers. It is the most popular metric that is used by airlines to describe the extent to which seats on the aircraft are being utilised by passengers. The passenger seat factor is defined in the following way: PSF =

ୖ୔୏ ୅ୗ୏

=

୔୅ଡ଼×୔୅ୗ୐ ୗ୉୅୘ୗ×୔୅ୗ୐

=

୔୅ଡ଼

(2.18)

ୗ୉୅୘ୗ

Equation (2.18) tells us that the passenger seat factor can be computed by either dividing the distance-based metrics RPK and ASK or by dividing the distance-unrelated load and capacity metrics PAX and SEATS. The passenger seat factor increases because capacity increases at a slower rate than passenger load increases. Numerical Example 1. All Nippon Airways flight NH880 between Sydney and Tokyo (Haneda) on a Boeing 787-900 has 246 available seats and 170 revenue passengers on board. What is the passenger seat factor? PSF =

ଵ଻଴ ଶସ଺

= 69.1%

2. Qantas flight QF1 flies from Sydney to London Heathrow, departing at 15.50 and arriving at 06:55+1. The flight length of the trip is 24 hours and 5 minutes with a stopover in Dubai. The aircraft that is used on the QF1 flight is an A380-800 with a seat count of 484. There are 350 passengers on board the flight in Sydney; 300 of these passengers will continue to London while 30 passengers will join the flight from Dubai to London. Determine the seat factor for the flight between Sydney and Dubai and the flight between Dubai and London. What is the seat factor for the entire journey? PSFSYD-DXB =

ଷହ଴ ସ଼ସ

= 72.3%

PSFDXB-LHR =

ଷଷ଴ ସ଼ସ

= 68.2%

PSFSYD-DXB-LHR =

ଷଷ଴ାଷହ଴ ସ଼ସାସ଼ସ

= 70.2%

2.4.2 Freight Load Factor The freight load factor represents the percentage of freight capacity on the aircraft that is occupied by revenue earning freight. It is a measure of the extent to which the capacity to carry freight on the aircraft is being utilised. It is calculated by dividing the freight load metric RFTK by the freight capacity metric AFTK as follows: Freight Load Factor =

ୖ୊୘୏

(2.19)

୅୊୘୏

The freight load factor increases because freight capacity or AFTKs increase at a slower rate than the freight load, RFTKs, increase.

2.4.3 Total Load Factor The total load factor represents the percentage of passenger and freight capacity of an aircraft that is occupied by revenue earning passengers and freight. It is computed by dividing revenue tonne kilometres by available tonne kilometres as follows: TLF =

ୖ୘୏

(2.20)

୅୘୏

The total load factor increases when RTKs increase at a faster rate than ATKs.

Quiz 2-3 Capacity Utilisation 1. Consider the Qantas operating statistics over the 6 months to December 2016 presented in Quiz 2-2, question 7 once again. Use this information to answer the following questions. (a) What is the passenger seat factor over the 6 months to December 2016 for Qantas Mainline international services? (b) What is the passenger seat factor over the 6 months to December 2016 for Jetstar domestic services? (c) Compute the passenger seat factor for the Jetstar Group over the 6 months to December 2016. Compare this to the seat factor estimate for the 6 months to December 2015.

Chapter 2

28

(d) Compute the passenger seat factor for Qantas Mainline over the 6 months to December 2016. Compare this to the seat factor estimate for the 6 months to December 2015. (e) Find the seats carried for Qantas domestic over the 6 months to December 2016. Compare this to the seats carried for the 6 months to December 2015. (f) Find the seats carried for Jetstar Asia over the 6 months to December 2016. Compare this to the seats carried for the 6 months to December 2015. 2. Consider once again the Cathay Pacific operating statistics for February 2017 presented in Quiz 2-2, questions 2 to 5. Estimate the following using this data. (a) The seat factor for China Mainland services in February 2017. (b) The seat factor for South-East Asian services in February 2017. (c) The revenue tonne kilometres in February 2017 (note that you will need to perform a few steps to arrive at the answer to this question). (d) The seats carried at the total network level in February 2017. (e) The passenger seat factor for the Asia region in February 2017. (f) The percentage of Cathay Pacific ASKs flown in the Asia Region in February 2017.

2.5 Productivity Metrics 2.5.1 Labour Productivity The number of labour units employed by the airline is usually described by a metric called the number of full-time equivalent (FTE) staff. The number of FTEs is equal to the number of full-time employees plus the number of casual employees multiplied by the time that casual employees work relative to full-time staff. FTE is calculated using the following formula: FTE = Full-Time Employees + Casual Employees× ቀ

୅୴ୣ୰ୟ୥ୣ ୛ୣୣ୩୪୷ ୌ୭୳୰ୱ ୠ୷ େୟୱ୳ୟ୪ ୉୫୮୪୭୷ୣୣୱ

ቁ

୅୴ୣ୰ୟ୥ୣ ୛ୣୣ୩୪୷ ୌ୭୳୰ୱ ୠ୷ ୊୳୪୪ି୘୧୫ୣ ୉୫୮୪୭୷ୣୣୱ

(2.21)

Typically, the denominator of the right-hand side of (2.21) is 40 hours, but this may differ across airlines. If the numerator of the right-hand side of (2.21) were, for example, 30 this means that casual employees work around 75% of the time that full-time employees work. This means that casual employees are converted into full-time equivalent employees by multiplying the number of casual employees by 75%. FTE is used as an input into determining the productivity of labour. Labour productivity is usually proxied by the average product of labour. This is found by dividing ASKs by FTEs: Average Product of Labour =

୅ୗ୏

(2.22)

୊୘୉

The average product of labour is a measure of output divided by a measure of labour input. It measures the average productivity of all labour units. In this case the measure of output is ASKs. If an airline also produces a significant amount of freight capacity, and has a number of labour units employed in freight functions, then it can compute labour productivity using ATKs in the numerator as follows: Average Product of Labour =

୅୘୏

(2.23)

୊୘୉

Airlines such as Cathay Pacific, Korean Air, Singapore Airlines, Emirates and Lufthansa should use (2.23) to determine company-wide labour productivity because these airlines have a significant air freight business. We have also used FTEs as the metric that represents the number of labour units in (2.22) and (2.23). It is also feasible to use the number of block hours as the volume of labour. Labour productivity when block hours is used to represent the volume of labour is computed using the following formula: Average Product of Labour =

୅ୗ୏ ୌ୓୙ୖୗ

or

୅୘୏ ୌ୓୙ୖୗ

(2.24)

The measures of labour productivity at (2.22) to (2.24) are referred to as the average product of labour found by dividing the volume of output by the volume of labour. Another way to compute labour productivity involves calculating the marginal product of labour. The marginal product of labour is determined by examining the change in the volume of output divided by the change in the volume of labour. It is calculated in the following way: Marginal Product of Labour =

'୅ୗ୏ '୊୘୉

(2.25)

The Language of the Airline Business

29

The marginal product of labour is a more difficult concept of productivity to implement because to implement it properly it is necessary to estimate a functional relationship between output and employment, which is difficult for most airlines to do. Labour productivity can also be measured for different types of labour within the airline, such as pilots, cabin crew and engineers. To estimate the productivity of pilots and cabin crew the output driver is ASKs or ATKs, and the labour input is measured as the hours that cabin and pilot crew work. To estimate the labour productivity of engineers, output is measured as the completion of a certain quantity of maintenance work, and the input is the number of engineering hours required to complete the certain quantity of maintenance work.

2.5.2 Fuel Productivity Fuel productivity is measure by the average product of fuel. It is found by dividing ASKs by the quantity of fuel that is consumed by the airline as follows: Average Product of Fuel =

୅ୗ୏ ୊୳ୣ୪ େ୭୬ୱ୳୫୮୲୧୭୬

(2.26)

In (2.26) we use the output metric ASKs but airlines can also use ATKs in the numerator of (2.26) if freight is an important part of their business. When ATKs are used in the numerator the average product of fuel becomes: Average Product of Fuel =

୅୘୏ ୊୳ୣ୪ େ୭୬ୱ୳୫୮୲୧୭୬

(2.27)

The volume of fuel used in the denominator of our fuel productivity expressions (2.26) and (2.27) is usually measured in gallons, barrels, or tonnes. Gallons are often used by U.S. airlines, barrels by Asian airlines and tonnes by European airlines. There are 42 gallons of fuel in a barrel, and around 7.93 barrels of jet fuel in a metric tonne so that we can write the various jet fuel volume conversions as: Fuel Consumption in Gallons = 42 × Fuel Consumption in Barrels Fuel Consumption in Barrels = Fuel Consumption in Metric Tonnes u 7.93 All these measurements of fuel consumption are interchangeable if the above conversions are used. We can also determine fuel productivity by aircraft type if fuel and ASKs are split by aircraft type. This is usually not externally reported by airlines, but airline internal data systems will routinely collect this data.

2.5.3 Fleet Productivity Fleet productivity is a measure of the extent to which fleet units are used by the airline. It is calculated using the following formula: Fleet Productivity =

୆୪୭ୡ୩ ୌ୭୳୰ୱ ୊୪ୣୣ୲ ୙୬୧୲ୱ×୒୳୫ୠୣ୰ ୭୤ ୈୟ୷ୱ

(2.28)

Equation (2.28) determines the average number of hours that aircraft are flown per day. The higher is this number the more that aircraft are flown by the airline. Higher fleet productivity allows the airline to reduce unit costs, as fixed aircraft capital costs are spread across more ASKs and block hours. Utilisation is heavily affected by the network that the airline flies, especially if the network involves flights across time zones and flights that involve airports which are affected by night-time flying curfews. We can also determine fleet productivity by using ASKs or ATKs in the numerator of (2.28) as follows: Fleet Productivity = Fleet Productivity =

୅ୗ୏ ୊୪ୣୣ୲ ୙୬୧୲ୱ×୒୳୫ୠୣ୰ ୭୤ ୈୟ୷ୱ ୅୘୏ ୊୪ୣୣ୲ ୙୬୧୲ୱ×୒୳୫ୠୣ୰ ୭୤ ୈୟ୷ୱ

(2.29) (2.30)

The formulae (2.29) and (2.30) use ASKs or ATKs as the representations for output, or the numerator of the productivity formula, as opposed to block hours. Airlines will tend to report the hours measure of fleet productivity because it is easier for aviation analysts to interpret this metric. As there is a strong correlation between block hours, ASKs and ATKs all three measures of fleet productivity are reasonably interchangeable. These fleet productivity metrics can also be determined by aircraft type. Airlines will usually capture the block hours by aircraft type which enables the airline to determine the level of fleet productivity for certain aircraft types and by

30

Chapter 2

aircraft registration number. Airlines do not usually split block hours by aircraft type in their external reporting, with a notable exception Alaska Airlines which routinely provides this information.

Quiz 2-4 Productivity Metrics The following table represents the operational output in the September quarter 2016 of two low-cost airlines operating in the U.S.A., Allegiant Air and JetBlue. Allegiant Air Comparative Consolidated Operating Statistics The following tables present operating statistics for the periods indicated: Three Months Ended September 30 2016 2015 Operating statistics (unaudited) Total system statistics: Passengers Revenue passenger miles (RPMs) (thousands) Available seat miles (ASM) (thousand) Load factor Operating expense per ASM (cents) Fuel expense per ASM (cents) Operating CASM, excluding fuel (cents) ASMs per gallon of fuel Departures Block hours Average stage length (miles) Average number of operating aircraft during the period Average block hours per aircraft per day Full-time equivalent employees at end of period Fuel gallons consumed (thousand) Average fuel cost per gallon

2,939,055 2,645,533 3,121,762 84.7% 8.22 2.22 6.00 70.6 21,384 47,739 864 84.0 6.2 3,287 44,187 $1.57

2,420,819 2,235,683 2,597,658 86.1% 8.58 2.63 5.95 69.2 17,330 39,347 878 74.7 5.7 2,654 37,518 $1.82

Scheduled service statistics: Passengers Revenue passenger miles (RPMs) (thousands) Available seat miles (ASM) (thousand) Load factor Departures Block hours Total scheduled service revenue per ASM (TRASM) (cents) Average fare - scheduled service Average fare – ancillary air-related charges Average fare - total Average stage length (miles) Fuel gallons consumed (thousand) Average fuel cost per gallon Percent of sales through website during period

2,904,295 2,603,849 2,997,529 86.9% 20,398 45,740 10.54 61.07 43.83 108.78 869 42,439 1.59 94.6%

2,383,556 2,204,760 2,526,292 87.3% 16,563 38,094 11.38 71.32 45.12 120.59 894 36,458 1.83 95.2%

JETBLUE AIRWAYS CORPORATION COMPARATIVE OPERATING STATISTICS (unaudited)

Revenue passengers (thousands) Revenue passenger miles (millions) Available seat miles (millions) Load factor Aircraft utilisation (hours per day) Average fare Yield per passenger mile (cents)

Three Months Ended September 30, 2016 9,953 11,905 13,796 86.3% 12.2 $157.87 13.20

Three Months Ended September 30, 2015 9,237 11,063 12,976 85.3% 12.2 $167.96 14.02

The Language of the Airline Business

Passenger revenue per ASM (cents) Revenue per ASM (cents) Operating expense per ASM (cents) Operating expense per ASM excluding fuel and related ( ) Operating expense per ASM excluding fuel (cents) Departures Average stage length (miles) Average number of operating aircraft during period Average fuel cost per gallon, including fuel taxes Fuel gallons consumed (millions) Average number of full-time equivalent crew members

11.39 12.55 9.99 7.86 7.53 86,801 1,091 219.6 $1.48 198 15,521

31

11.96 13.01 10.3 7.67 7.31 82,989 1,094 209.0 $1.85 185 14,418

Use this information to answer the following questions. (a) How many gallons of fuel did Allegiant and JetBlue consume in the September 2016 quarter? Convert the fuel consumption levels into barrels of jet fuel. (b) Estimate ASMs per barrel of jet fuel consumed for the two airlines in the September 2016 quarter. What might be driving the difference between them? (c) Identify the average number of operating aircraft during the period for both Allegiant Air and JetBlue. (d) Identify the block hours flown by Allegiant Air in the September quarter 2016. (e) Estimate the block hours per fleet unit per day for Allegiant Air and identify the same for JetBlue. (f) Estimate the block hours for JetBlue for the September quarter 2016. (g) Estimate the ASM per fleet unit per day for the September quarter 2016 for both airlines. Compare the results. What may be driving the difference? (Hint: estimate the average passenger sector lengths for the two airlines). (h) Identify the full-time equivalent employees for Allegiant Air and the average number of full-time crew members for JetBlue. (i) Estimate the ASM per FTE for Allegiant Air in the September quarter 2016, and the ASM per average crew member in the case of JetBlue in the September quarter 2016. Compare and explain your results.

2.6 Airline Operational Performance Metrics 2.6.1 On-Time Performance On-time performance is a measure of the extent to which an airline arrives at its destination and/or departs from its origin on-time. It is computed using the following formula: On െ Time Performance =

OnିTime DeparturesାOnିTime Arrivals Movements

(2.31)

Departures and arrivals on-time, which is the numerator of (2.31), typically mean departures and arrivals that are within 15 minutes of the scheduled departure and arrival times. The denominator of (2.31) represents the number of arrivals and departures, which is also equal to the number of aircraft movements. For a single flight, there is a departure and an arrival. This means for a single flight the denominator of (2.31) is equal to 2. If on this flight the departure is on-time and the arrival is on-time, then the numerator of (2.31) is 2 and the denominator is 2, resulting in an on-time performance that is equal to 2/2 or 100%. If neither the departure nor the arrival is on-time, then the numerator of (2.31) is 0 and the on-time performance is 0%. If the departure is on-time but the arrival is late, or vice-versa, then the numerator of (2.31) is 1 and the on-time performance is 50%. Equation (2.31) is the on-time performance statistic for a combination of arrivals and departures. It is also possible to compute on-time arrivals and on-time departures separately. On-time arrivals is equal to: On-Time Arrivals =

୓୬ି୘୧୫ୣ ୅୰୰୧୴ୟ୪ୱ ୒୳୫ୠୣ୰ ୭୤ ୅୰୰୧୴ୟ୪ୱ

The on-time departures are determined according to the following calculation: On-Time Departures =

୓୬ି୘୧୫ୣ ୈୣ୮ୟ୰୲୳୰ୣୱ ୒୳୫ୠୣ୰ ୭୤ ୈୣ୮ୟ୰୲୳୰ୣୱ

Given that aircraft which arrive late are also more likely to depart late, then on-time arrivals and departures are likely to be highly correlated. It is also the case that airlines will often only publish either on-time arrivals or on-time departures but not both or they combine them using equation (2.31) because they convey almost the same on-time information. The on-time performance statistic is incredibly important for airlines with a high business-purpose, high net-worth passenger demographic because these passengers value their time more than other passenger types, such as low-income

Chapter 2

32

leisure passengers. Business passengers also travel to attend meetings and conferences that start and finish at particular times of the day, and so there are strong motivations for them to arrive on time.

2.6.2 Reliability Reliability is a measure of the extent to which an airline sticks to its scheduled services. It is computed using the following formula: Reliability % =

Sectors Flown Sectors Scheduled

=

Sectors Flown

(2.32)

Sectors Flown + Sectors Cancelled

The reliability percentage is therefore equal to the percentage of scheduled flights that are actually flown. The denominator of (2.32) is equal to the number of flights that is scheduled. This is greater than or equal to the number of flights actually flown, as the scheduled flights can either be flown or they may be cancelled. If an airline cancels services that are scheduled to be flown, then the reliability percentage will be low. Conversely, if an airline flies all services that it has scheduled, or has planned to fly, then its reliability rate will be 100%. The reliability rate for airlines globally is typically somewhere between 95% and 100%. Airlines will often cancel services because of poor weather conditions (strong crosswinds, tropical cyclones, hurricanes, snowstorms, lightning, and fog), mechanical faults with aircraft, pilots or cabin crew who call in sick leaving the airline undermanned, or air traffic control which is undermanned for similar reasons.

Quiz 2-5 Airline Performance Metrics The following information is taken from the Australian Government Bureau of Infrastructure, Transport and Regional Economics (BITRE) at the website www.bitre.gov.au. Click on “statistics”, “aviation” and then “domestic on-time performance”, “latest monthly report”, “complete monthly report” to get access to the data. Sydney-Melbourne Market On-Time Performance Data by Quarter – Calendar 2016

Mar-16 Jun-16 Sep-16 Dec-16

Sectors Scheduled (One Way) 13684 13698 14642 14118

Sectors Flown 13326 13214 13894 13444

Departures on Time 11070 11350 11412 10368

Arrivals on Time 10962 10770 10602 9282

Use this information to calculate the following. (a) The reliability percentage for each quarter in 2016 on the Sydney-Melbourne route. (b) The on-time departures percentage for each quarter in 2016 on the Sydney-Melbourne route. (c) The on-time arrivals percentage for each quarter in 2016 on the Sydney-Melbourne route. (d) The on-time performance percentage for each quarter in 2016 on the Sydney-Melbourne route.

2.11. Yield 2.11.1 Passenger Yield Passenger yield is a measure of the price that passengers pay for air travel. Airlines prefer to measure the price that passengers pay in per kilometre terms rather than per passenger terms because the per passenger measure will differ depending on the average sector length of the airline. When the average sector length is higher this naturally causes higher revenue per passenger. Passenger yield is computed using the following formula: Passenger Yield =

Passenger Revenue Revenue Passenger Kilometres

(2.33)

It is usually defined in terms of cents in the case of airlines that have currency denominations in cents (such as the U.S., Australia, New Zealand, Euro Zone, Singapore and Hong Kong) otherwise it is defined in terms of the main currency unit (for example, Chinese Yuan for Chinese carriers, New Taiwan Dollars in the case of the Taiwanese carriers, Thai Bhat in the case of Thai carriers, Japanese Yen in the case of Japanese carriers and United Arab Emirates Dirhams in the case of UAE carriers). Passenger yield increases when passenger revenue increases at a faster pace than revenue passenger kilometres. In jurisdictions in which miles are used as the distance metric rather than kilometres (such as the U.S.A. and Canada), passenger yield is computed using revenue passenger miles in the denominator rather than revenue passenger kilometres as follows:

The Language of the Airline Business

Passenger Yield =

Passenger Revenue

33

(2.34)

Revenue Passenger Miles

A revenue passenger kilometre is around 1.852 times a revenue passenger mile.

2.11.2 Passenger RASK Passenger RASK or PRASK for short is another measure of the price that passengers pay for air travel. It incorporates information about both the passenger seat factor and the passenger yield. It is computed by dividing passenger revenue by available seat kilometres as follows: PRASK =

Passenger Revenue

(2.35)

Available Seat Kilometres

We know that available seat kilometres are revenue passenger kilometres divided by the passenger seat factor, which means we can write PRASK at (2.35) as follows: PRASK =

Passenger Revenue Revenue Passenger Kilometres Passenger Seat Factor

=

Passenger Revenue×Passenger Seat Factor Revenue Passenger Kilometres

(2.36)

Or alternatively, simplifying (2.36) even further: PRASK = Passenger Yield u Passenger Seat Factor

(2.37)

Like passenger yield, PRASK is usually defined in terms of cents. It is constructed as the product of the seat factor and the passenger yield. PRASK is therefore lower than the passenger yield because the seat factor is a number between 0 and 1. Where available seat miles are used as the capacity metric rather than available seat kilometres then the passenger yield metric that we calculate is called PRASM and it is computed using a denominator that is available seat miles as follows: PRASM =

Passenger Revenue

(2.38)

Available Seat Miles

2.11.3 Freight Yield The freight yield of the airline is freight revenue divided by the number of revenue freight tonne kilometres: Freight Yield =

Freight Revenue

(2.39)

Revenue Freight Tonne Kilometres

This is a measure of the price that passenger airlines or logistics companies receive, on a per tonne kilometre basis, for the transport of air freight. The list price of air freight is usually defined on a per tonne or per cubic meter basis but in the context of understanding the performance of an airline’s unit freight revenue it is defined in terms of tonne kilometres. Like passenger yield, freight yield is often defined in terms of cents where it is possible to do so.

2.11.4 RASK RASK is short for revenue per available seat kilometre. Or in countries like the U.S. which use miles rather than kilometres, it is called RASM. RASK is defined as total operating or traffic revenue divided by the number of available seat kilometres as follows: RASK =

Revenue Available Seat Kilometres

(2.40)

In the case of RASM it is defined as: RASM =

Revenue Available Seat Miles

(2.41)

The numerator of these expressions, revenue, is usually operating revenue. This excludes revenue generated from interest income, asset sales or hedging gains and losses, which are non-operational items. RASK and RASM are used as a unit revenue metric by airlines that do not have a significant freight business. If an airline’s freight business is significant, it is better to use a denominator that is available tonne kilometres or revenue tonne kilometres as explained in section 2.11.5 below.

Chapter 2

34

2.11.5 Total Yield Total yield is a measure of the effective price of the combined freight and passenger load on the aircraft. It is computed by dividing operating revenue into revenue tonne kilometres as follows: Total Yield =

Total Revenue

(2.42)

Revenue Tonne Kilometres

In the case where the airline uses revenue tonne miles rather than revenue tonne kilometres the total yield expression is: Total Yield =

Total Revenue

(2.43)

Revenue Tonne Miles

The total yield metric is used for airline businesses in which the freight component of the business is significant. The rule of thumb that I use is that when freight contributes to at least 20% of total revenue then the best yield measure to use for the business is described by (2.42) and (2.43).

2.11.6 Airline Example Table 2-3 below presents information taken from the Quarter 4, December 2016 financial report of Hawaiian Airlines. Hawaiian Holdings, Inc Consolidated Statement of Operations (in thousands, except for per share data) (unaudited) Three Months Ended December 31 2016 2015 Operating Revenue Passenger Other Total

$553,647 $79,326 $632,973

$500,149 $74,005 $574,154

Selected Statistical Data (unaudited) Total Operations (a): (in thousands, except as otherwise indicated) Revenue Passengers Flown RPM ASM Passenger revenue per RPM (cents) Passenger Load Factor (RPM/ASM, %) Passenger Revenue per ASM (PRASM, cents) Operating Revenue per ASM (RASM, cents) Operating cost per ASM (CASM, cents) CASM excluding aircraft fuel and special items (cents) Aircraft Fuel expense per ASM (cents) Revenue Block Hours Operated Gallons of jet fuel consumed Average cost per gallon of jet fuel (actual $) Economic fuel cost per gallon ($)

Three Months Ended December 31 2016 2015 2,730 2,658 3,932,713 3,639,219 4,570,679 4,391,792 14.09 13.76 86.0 82.9 12.13 11.41 13.85 13.07 13.30 10.59 9.12 8.58 2.10 44,627 61,647 1.55 1.51

2.01 42,488 58,008 1.52 1.80

Source: Hawaiian Airlines Quarterly Financial Report 2016

Table 2-3: Hawaiian Airlines Revenue and Operating Data, December Quarter 2016 We can see in the Hawaiian Airlines output in the second set of tables in Table 2-3 that Hawaiian provides information about passenger yield and RASM directly. Let us now show how Hawaiian Airlines arrives at these numbers. Passenger revenue per RPM is indicated by Hawaiian Airlines to be 14.09 over the 3 months to December 2016. This is computed by dividing passenger revenue, which can be found in the upper part of Table 2-3 of the Hawaiian results, by revenue passenger miles, which can be found in the bottom half of Table 2-3 as follows: Passenger Yield =

ହହଷ,଺ସ଻×ଵ଴଴ ଷ,ଽଷଶ,଻ଵଷ

ൎ 14.08

The Language of the Airline Business

35

Notice in this calculation that we multiply the numerator by 100. This is to convert the passenger yield calculation from dollars into cents. PRASM is found by multiplying passenger yield by the passenger seat factor, or by dividing passenger revenue by available seat miles as follows: PRASM =

ହହଷ,଺ସ଻×ଵ଴଴ ସ,ହ଻଴,଺଻ଽ

ൎ 12.11

Once again, we use 100 in the numerator of this expression to convert the PRASM expression into cents from dollars. RASM is found by dividing total revenue by ASM as follows: RASM =

଺ଷଶ,ଽ଻ଷ×ଵ଴଴ ସ,ହ଻଴,଺଻ଽ

ൎ 13.85

2.12 Unit Cost 2.12.1 Cost per ASK (CASK) The most common measure of unit cost in the airline business is CASK or operating cost per ASK in the case of kilometrebased passenger capacity and CASM in the case of miles-based passenger capacity. It is computed by dividing ASKs into total operating cost as follows: CASK =

Total Operating Cost

(2.44)

ASK

Or in the case of using available seat miles (ASM): CASM =

Total Operating Cost

(2.45)

ASM

For many airlines that have a significant freight business, it is better to use ATKs in the denominator of this expression since the activity driver of cost is more closely aligned with ATKs than ASKs. In this case our unit cost expression is CATK rather than CASK: CATK =

Total Operating Cost

(2.46)

ATK

In the case of miles, the measure of unit cost is: CATM =

Total Operating Cost

(2.47)

ATM

2.12.2 Cost per ASK Excluding Fuel (CASK Ex-Fuel) Often, we wish to exclude fuel costs from our CASK calculation because it can be significantly affected by movements in fuel prices. In this case the numerator of the CASK expression is total operating costs less fuel costs as follows: CASK ex Fuel =

Total Operating Cost Excluding Fuel ASK

(2.48)

In the case of using miles this becomes: CASM ex Fuel =

Total Operating Cost Excluding Fuel ASM

(2.49)

Once again, for businesses that have a significant freight component of their business the superior measure of unit cost divides operating costs excluding fuel by ATKs as follows: CATK ex Fuel =

Total Operating Cost Excluding Fuel ATK

(2.50)

It is also the case that some airlines exclude other one-off, volatile items from operating cost to estimate CASK. For example, airlines often exclude asset write-offs to calculate CASK in which case the measure of unit cost we compute is CASK ex Fuel and volatile Items. This is a particularly popular measure of unit cost that is calculated by airlines in the U.S.

Chapter 2

36

2.12.3 Numerical Example If we return to our Hawaiian Airlines data of Table 2-3, we will now supply data related to the cost line of the profit and loss statement (see Table 2-4 below). Hawaiian Holdings, Inc Consolidated Statement of Operations (in thousands, except for per share data) (unaudited) Three Months Ended December 31 2016 2015 Operating Expenses: Aircraft Fuel, including taxes and delivery Wages and benefits Aircraft rent Maintenance materials and repairs Aircraft and passenger servicing Commissions and other selling Depreciation and amortisation Other rentals and landing fees Purchased services Other Total

$95,806 $144,598 $32,330 $62,069 $33,631 $31,795 $26,499 $29,749 $23,385 $33,210 $608,104

$88,399 $129,631 $28,921 $56,136 $29,501 $28,529 $26,804 $24,248 $21,294 $31,845 $465,308

Source: Hawaiian Airlines Quarterly Financial Report 2016

Table 2-4: Hawaiian Airlines Cost Data, December Quarter 2016 We can use this information to derive estimates of CASM as follows: CASM =

଺଴଼,ଵ଴ସ×ଵ଴଴ ସ,ହ଻଴,଺଻ଽ

= 13.30 U.S. cents per ASM

Notice that the numerator of this expression is multiplied by 100 to convert the calculation into cents from dollars. If we exclude fuel costs from the calculation of CASM then we obtain the following: CASM ex Fuel =

(଺଴଼,ଵ଴ସିଽହ,଼଴଺)×ଵ଴଴ ସ,ହ଻଴,଺଻ଽ

= 11.21 U.S. cents per ASM

You will notice that Hawaiian also calculates CASM after excluding fuel and “special items”, which is found to be 9.12.

Quiz 2-6 Yield and Unit Cost Metrics The following is a cost breakdown of Norwegian Air Shuttle over the December quarter 2016, which is presented in its interim report for that quarter. OPERATING COST BREAKDOWN Unaudited Q4 (Amounts in NOK million) Personnel expenses Sales/distribution expenses Aviation fuel Airport and ATC charges Handling charges Technical maintenance expenses Other flight operation expenses Other losses/(gains) – net

2016 1,091.4 171.8 1,437.1 776.9 802.8 563.1 342.7 410.0

2015 862.7 151.1 1,199.9 711.4 613.5 431.1 233.7 304.9

The Language of the Airline Business

Total operating expenses Leasing Total operating expenses incl leases Depreciation and amortisation

37

4,744.4 684.0 5,428.4 337.6

5,023.6 561.7 5,585.3 366.5

OPERATIONAL REVIEW CONSOLIDATED TRAFFIC FIGURES AND RATIOS Q4 Internet bookings ASK (million) RPK (million) Passengers (million) Load Factor Average sector length (km) Fuel Consumption (metric tonnes) CO2 per RPK

2016 77% 15,109 12,959 7.18 85.8% 1,503 308,298 75

2015 76% 11,909 10,107 6.13 84.9% 1,389 247,060 77

The following is a revenue breakdown for Norwegian Air Shuttle: OPERATING REVENUE BREAKDOWN Q4 Unaudited (Amounts in NOK millions) Passenger Revenue Ancillary passenger revenue Other revenue Total

2016 4,795,7 927.1 378.7 6,101.5

2015 4,324.2 774.4 220.4 5,318.9

Use this information to answer the following questions. (a) Find the cost per ASK inclusive of lease and depreciation and amortisation costs for the December quarter 2016 and 2015. What has happened to CASK over the year? (b) Find CASK ex fuel for the December quarter 2016 and 2015. What has happened to CASK ex fuel over the year? Use this estimate to determine fuel cost per ASK in the December quarter 2016 and 2015. (c) Find the passenger yield in the December 2016 quarter and compare this to the passenger yield in the December quarter 2015. (d) Find PRASK in the December 2016 quarter and compare this to the PRASK in the December quarter 2015. (e) Find RASK in the December 2016 quarter and compare this to RASK in the December quarter 2015. (f) What percentage of total revenue is passenger revenue?

PART B: AVIATION DEMAND AND REVENUE

CHAPTER 3 PASSENGER DEMAND

This chapter examines why people travel by air. People travel by air because they need a holiday, they wish to visit friends or relatives, their employer has asked them to hold meetings with clients, they wish to attend a conference, they are starting a new job, or they are studying. Different motivations for air travel are usually driven by different forces. People who need a holiday decide to travel after considering their income, how much it costs to go on a holiday and whether they have enough annual leave to take a holiday. They will also decide between competing destinations. Should I go to Hawaii, Mexico, or the Canary Islands? In determining whether they will visit one location over other potential locations, a key consideration is the cost of those locations. One of the most important determinants of that cost is the exchange rate, because this influences the cost of buying goods and services on arrival at the leisure destination. The US dollar, the Mexican Peso, and the Euro, may all move in quite different directions, influencing the price of a Hawaiian holiday, relative to a holiday to Mexico and the Canary Islands. The leisure traveller is usually highly price sensitive, which means anything that affects the cost of one destination compared to another, including the exchange rate and airfares, are important variables that are considered by the leisure traveller in making destination decisions. People who travel to visit friends and relatives (VFR) also make their decisions by considering the cost of the trip and their income. Since they are likely to be travelling to a destination in which their family and friends reside, without consideration of alternative destinations, then there are fewer variables to consider when determining why these passengers travel compared to leisure travellers. There is also a greater chance that the visitor will stay with friends and relatives, which takes out one of the most significant expenses of travel, accommodation. Accommodation is referred to as a complement to air travel because it is consumed with air travel. People who travel for business usually have most of their expenses paid by their employer, who enjoys a tax deduction for these expenses. The primary driver of this demand for travel is the need to conduct business, which is tied to the income of the business and the likely return that the business will enjoy from its investment in air travel. Businesspurpose travel will drive a high proportion of travel in premium cabins and a high proportion of passengers that pay fares that are flexible in nature, which tend to be high yielding fares. This segment of the market is therefore a much sought-after segment, particularly for full-service airlines. As this chapter will show, there are many different variables that drive the travel behaviour of leisure, business, and VFR travel such as GDP, wealth, employment, exchange rates, airline product quality, on-time performance, and marketing effort to name but a few. The fact that there are a wide variety of variables that drive demand evidences the complexity of trying to predict air travel demand. The aim of this chapter is to reveal, explain and simplify this complexity.

3.1 Demand by Purpose There are many different reasons that explain why people travel by air. The key reasons are (1) leisure or holiday, (2) visiting friends and relatives, (3) business, convention, or conference, (4) education, and (5) employment. By far and away the most important reason for travel is for leisure purposes. Leisure travellers are an important category of air travel because they are the most sensitive to price. It is this segment of the market that airlines are trying to convince to travel by offering more seats at low fares. For low-cost carriers such as Southwest, Ryanair, easyJet, Scoot, Air Asia, Wizz Air, Norwegian Air Shuttle and Jetstar, this is the most important segment of the air travel market. Leisure travellers typically pay for their travel from out of their own household income in the case of passengers who have a job, and from their own household wealth when they are retired. Not only is leisure-purpose travel sensitive to the price of air travel but it is also sensitive to the price of all other goods and services that are purchased while travelling, such as the price of goods and services bought on arrival at the tourism destination. It is for this reason that leisure travel is especially sensitive to changes in the exchange rate since exchange rate changes affect the price of goods and services when the passenger arrives in the foreign country. The leisure traveller will also make decisions about travelling to a particular destination by assessing this destination against other, substitute destinations. For example, Australian and Asian leisure travellers may assess a leisure trip to Bali against a leisure trip to Phuket, Honolulu, Hamilton Island, Guam, or Cairns. American leisure travellers will assess a trip to Maui against a trip to the Caribbean, Mexico, or the beaches of South America. A European leisure traveller will compare a trip to the Greek Islands with travel to the Canary Islands and Ibiza. For income or cash constrained travellers, mode of transport is also an important consideration where other modes of transport are feasible and available (see also section 3.2 below). Air transport is generally more expensive than train, bus and ferry transport and hiring a car, which means that price sensitive and cash and income constrained travellers will often choose the cheaper land and sea travel alternatives to air travel even though the trip will take much longer.

Passenger Demand

41

The business traveller, conversely, typically does not pay for the trip. The employer of the business traveller pays for the trip and receives a tax deduction for the travel expense. For this reason, the business traveller is not as price sensitive as the leisure traveller. The business-purpose passenger travels to attend a meeting, a conference, a convention or to lobby governments. This travel depends more on the state of the economy and the finances of the business paying for the travel rather than the price of air travel. Many business-purpose travellers will travel to their destination and return on the same day if the travel time is sufficiently short (one to two hours). For these travellers, the price of complements such as accommodation, is not a consideration. Of critical importance to the business-purpose traveller is the total travel time. Business-purpose travellers tend to have a relatively high opportunity cost of time. The opportunity cost of time is the benefit that passengers enjoy if, instead of taking the time to travel by air, they used that time for the next best alternative activity, such as working, or sitting on a beach, or spending time with the family (see also section 3.2 below). It is for this reason that airline on-time performance and reliability are key drawcard features for travellers that need to fly by air for business purposes. Business travel usually involves travel from point A to point B without consideration of alternative destinations. As such, an increase in the price of destination B relative to another destination C does not have a strong influence on A-B business travel decisions in the same way that it affects leisure travel decisions, where switching between destinations can be significant. For example, a resident of Los Angeles who must travel to New York for business, doesn’t compare the benefit enjoyed from this trip to the benefit enjoyed from a trip to a substitute destination like Toronto, Canada, because the passenger must travel to meet colleagues located in New York. Passengers that travel to visit friends and relatives have slightly different drivers to leisure travellers because they are more likely to stay with friends and relatives and save on accommodation costs. They are also only interested in one destination – the location in which their friends and relatives reside – meaning there is virtually no chance of choosing an alternate or substitute destination. Visiting friends and relatives travel is strongly driven by migration patterns and changes in the living location of households. For example, the parents of a young couple living in Malmö, Sweden, have migrated to the United States. To visit those parents at Christmas, the young couple must travel to the United States and will not consider another country destination. Travel for education purposes is more complex than other travel purposes because the non-airfare components of the travel bundle, such as accommodation, food and beverages, land transport and entertainment, take on more importance than airfares. This is because the average length of stay of education travellers could be years, which means accommodation and food and beverage costs become far more significant in the total travel costs of the passenger than the cost of an airfare.

3.2 Mode of Transport Substitutes in Short Distance Domestic Markets In the case of international, domestic, and regional travel over relatively short sectors or distances, there is the potential for passengers to use other modes of transport, such as land and sea transport, rather than air travel. These substitute modes of transport for air include own car or car hire (also referred to as self-drive), train, bus, and ferry. Travel by land includes road and rail travel. Road travel may include the use of a passenger vehicle or the use of a bus. Travel by sea includes the use of ferries and ocean liners. When passengers make decisions in relation to the choice of mode of transport, they will consider the net benefit of those decisions. The net benefit is the value they obtain from taking the mode of transport less the cost of the transport mode. The value that a passenger gains from taking a particular mode of transport will depend on the passenger’s personal preferences. This will be affected by, for example, fears of flying and heights, fears of congestion and claustrophobia, the likelihood of being in an accident, the likelihood of sitting next to ill passengers, the thrill of the transport mode (for example the thrill of the take-off when sitting on an aircraft or driving a vehicle fast), amongst many other factors that determine the value a passenger enjoys from consuming a transport type. The cost of different transport modes depends on many factors. They are both explicit and implicit factors. The explicit cost of land travel will include any cost that involves a direct outlay of money, such as train or bus fares, road tolls, petrol or diesel costs, costs of travel to and from bus and train stations, car hire and insurance, parking costs, costs associated with the maintenance of owned vehicles, and the expected costs associated with having an accident, vehicle breakdown or traffic violation. The explicit costs associated with air travel include airfares, costs of travel to and from airports, including parking fees and taxis, and air travel insurance. The implicit cost of travel is the opportunity cost of time associated with the travel type. This is not an out-of-pocket cost, rather it is an opportunity cost. The opportunity cost is the value of the activity in which the traveller would be engaged if the traveller was not travelling by air, land, or sea. This opportunity or implicit cost is usually determined by multiplying the time it takes to complete the air, land, or sea journey by the wage rate of the passenger. The rationale for using the wage rate to determine the opportunity cost is that the time it takes to complete the journey could have been put to more productive use in the form of work, with passengers earning a return on that work which is equal to the wage rate. The implicit cost will depend on the ability of the passenger to use the transport time for productive use. The passenger who travels by air, train and bus, and the non-driver passenger in a motor vehicle, can potentially use a laptop or other device to do work, which reduces the opportunity cost to some extent. In general, the longer is the trip the higher the opportunity cost. Over relatively short sectors, often up to around the 350km mark, the opportunity cost is relatively low even for car, bus, and train travel. For these sectors, land and less often sea transport is a strong substitute for air transport. It is only when the opportunity cost associated with the land

42

Chapter 3

transport travel time becomes too great, which is true for long sectors, that there is a movement from land into air transport. Let us now look at some examples of the existence of mode of transport competition and substitution. Consider first a trip from London to Paris, which is a travel distance of 214 miles or 344 kilometres. The potential modes of transport for this trip are car, bus, rail, and air transport. A car trip will take on average around 4 hours. The car trip will also include a ferry ride if the Dover to Calais ferry is used, or it can include a train ride if the Euro Tunnel is used. The use of the Euro Tunnel in the case of a January 12, 2021, booking date and a March 11, 2021, travel date has a starting price of £81.6 The cost of the Dover to Calais ferry booking through Direct Ferries in the case of a January 12, 2021, booking date and a March 11, 2021, travel date starts at £72 one-way on DFDS Seaways.7 A one-way rail trip on the Eurostar high speed train has a starting price of €79 per passenger in the standard cabin for a January 12, 2021, booking date and February 11, 2021, travel date. Services run around 15-17 times per week and take around 2 hours and 15 minutes with trains travelling at an average speed of 186 miles per hour.8 Bus is by far the least expensive option for travel between London and Paris, with the bus services Eurolines and Ouibus offering transport options9. These services take between 7 and a half to 8 and a half hours with fares beginning at £29.98 oneway. There are many airlines that operate services between London and Paris including British Airways, easyJet and Air France, which operate to a scheduled travel time of between 1 hour and 15 and 1 hour and 20 minutes. This includes travel between London Heathrow, London Luton, and London Gatwick Airports to Charles De Gaulle Airport. Airfares start at around €49 but can escalate quickly if booking occurs only a few days out from departure. For ultra-long-haul sectors such as Perth (PER) to London, and Singapore (SIN) to New York, these routes have no mode of transport competition. The Perth to London Heathrow city pair is serviced by Qantas Airways only and has a great circle distance of 14,500 kilometres. The SIN to Newark (EWR), New York service is operated by Singapore Airlines only and has a great circle distance of 15,344 kilometres. There is no feasible alternative mode of transport that can compete with air travel for these city pairs. This in turn means that there is no mode of transport substitution possible. A mode of transport is considered a substitute for air travel when an increase in the price of that mode of transport results in an increase in air travel. For example, if there is an increase in the price of rail travel between London and Paris on the Eurostar, this will lead to more passengers choosing to travel between London and Paris by air. If there is an increase in the price of the Nozomi very fast train between the Japanese cities of Tokyo and Osaka, which takes 2 hours and 30 minutes, then more passengers will use Peach Airways or Jetstar to fly between the two cities, which has a one-hour shorter journey time.

3.3 Complements to Air Travel and the Relative Importance of Airfares 3.3.1 Accommodation One of the major cost items aside from airfares when a passenger travels by air is the cost of accommodation at the destination. The cost of accommodation has two component parts – the length of stay and the room rate per night. The room rate per night will depend on the type of room that is selected. In the case of many hotels, particularly luxury hotels, there are a wide variety of different room types and room rates. For example, a visitor to London who decides to stay at the Mandarin Oriental Hotel in Hyde Park London between 14 and 21 April 2021 in the case of a booking date of 18 March 2021 has the choice of 13 different room types, with a starting rate of £835 per night and a top price of £5,040 per night.10 In the case of budget hotels, such as the Ibis budget hotel in London Hounslow, there are usually far fewer room types to choose from and fewer room rate options. For the same arrival, departure and booking dates presented above for the Mandarin Oriental Hotel illustration, there are 6 room types and rates per night that are advertised by the Ibis budget hotel London Hounslow starting at £32 per night and rising to £85 per night.11 The purposes of travel that are likely to spend most on accommodation are those in which the average length of stay is longest. This includes education and employment purposes of travel and to a lesser extent leisure travel. Students that travel from overseas to pursue a tertiary education in another country are likely to require accommodation for many years, although they may fly back home several times during their stay abroad. This means that accommodation costs are likely to represent the most significant cost of these air travellers. Passengers that fly to a destination because they 6

Euro Tunnel accessed January 12, 2021, https://www.eurotunnel.com/uk/. Direct Ferries, accessed January 12, 2021, https://www.directferries.co.uk/dover_calais_ferry.htm?&utm_source=google&utm_medium=cpc&gclid=EAIaIQobChMIstW5_SU7gIVwUQrCh0EWQZ6EAAYASAAEgJDgfD_BwE&gclsrc=aw.ds. 8 Eurostar accessed January 12, 2021, https://www.eurostar.com/rw-en. 9 Eurolines, accessed January 12, 2021, https://www.eurolines.de/en/home/, and Ouibus, accessed January 12, 2021. https://www.ouibus.com/. 10 Mandarin Orient Hotel in Hyde Park London, accessed January 12, 2021, https://www.mandarinoriental.com/london/hyde-park/luxury-hotel. 11 Ibis budget hotel London Hounslow, accessed January 12, 2021, https://all.accor.com/hotel/6465/index.en.shtml?utm_campaign=seo+maps&utm_medium=seo+maps&utm_source=google+Maps. 7

Passenger Demand

43

have been offered a permanent job will need to find permanent accommodation at the destination location, which will cost significantly more than the airfares they have paid. In some cases, passengers may fly to a destination for employment purposes only for a limited period such as days, weeks, or months, in which case accommodation costs are significantly less than that paid by employee travellers who stay permanently. The purpose of travel that is likely to have the shortest length of stay is the business purpose, however the price per night that business travellers pay is likely to be higher as they are more likely to demand premium rooms. High accommodation prices may drive business-purpose travellers into shifting from overnight travel to day-trip travel. The business traveller will arrive in the morning, complete meetings, and then leave in the afternoon, thereby avoiding accommodation costs altogether. Visiting friends and relatives (VFR) travellers are in the best position to avoid accommodation costs altogether, as they are more likely to be offered accommodation by friends and relatives.

3.3.2 Land Transport When passengers fly by air, they must catch a taxi, ride-share, train, or bus to the origin airport or ask a friend to take them, and they must use the same land transport means to pick them up from the airport on their return. Passengers also need transportation from the destination airport to their accommodation, or their place of meeting or conference and return. They will also require transport for sightseeing, to visit entertainment venues, to go out for dinner at food and beverage merchants, to buy goods at retail outlets and to travel between business meetings. Land transport can therefore represent a relatively significant cost for the air traveller. Different purposes of travel will use land transport differently. Leisure travellers are more likely to do a lot of walking and use cheaper forms of land transport such as buses, trains, and ferries. Leisure travellers have a lower opportunity cost of time and are therefore happy to use modes of land transport that have a longer travel time. VFR travellers are more likely to borrow family and friends transport, or use the vehicles owned by family members, which reduces land travel costs. Business and convention travellers are likely to use more expensive land transport modes which have the shortest travel time, such as taxies and ride-share operators (for example Uber, Lyft and Gett). They need to be transported from their accommodation location to meetings and conference locations directly, essentially door-to-door, and with the least time possible, which is only feasible if they use a land transport model which delivers this flexibility in pick-up and drop-off options.

3.3.3 Food and Beverages Passengers consume food and beverages while on holidays, and often they do so more when on holidays than not. It follows that spending on food and beverages can represent a relatively significant percentage of the total travel budget. Spending on goods and services not only includes spending on eating out at cafes and restaurants, but also spending at supermarkets. In the case of apartment accommodation, where occupants can make their own meals using the kitchen that is located within their accommodation, spending at the supermarket on groceries can often exceed spending on eating out at cafes and restaurants. Spending on food and beverages is likely to represent a relatively high percentage of total travel spend in the case of leisure-purpose travellers, especially if they are in accommodation where there is no opportunity to cook their own meals. Leisure travellers are more motivated to venture out of their hotel room and try the local cuisine and will often eat-out in high-priced districts that attract tourists. Business travellers, conversely, are more likely to eat breakfast at their hotel, and eat lunch and dinner that is paid for by the employer or business colleagues, or by conference organisers. VFR travellers are more likely to, but not always, eat food and beverages that are supplied by family members and friends.

3.3.4 Entertainment Passengers often travel to destinations to see major sporting events, such as football games, cricket, formula-one racing, tennis, horse racing, summer and winter Olympic games, football and rugby world cups and golf. Travel is more heightened or peaky in the case of semi-final and final matches and will usually take place at certain times of the year. For example, the Wimbledon Tennis Championship takes place in June and July of each year, which is during the European summer. The Singapore Formula-one Grand Prix is held in September and October of each year. The Hong Kong Rugby Sevens tournament takes place in April of each year. Passengers also travel to attend major music events, museums, casinos, boxing matches and zoos. Not only do people travel to attend these events, but the players and the entertainers and their staff also travel by air to these events. For example, sporting teams will travel with coaches, directors, management, physiotherapists, doctors and even chefs. Entertainers and rock stars will travel with their band, security and staff that set up and assemble music equipment. In the case of weekend or overnight trips, the expenses associated with attending these events can often represent a high percentage of total travel costs.

Chapter 3

44

Quiz 3-1 Purpose of Travel, Substitutes and Complements 1.

Define the likely purpose of travel for the following passengers. (a) John decides to fly back home to see his sick father. (b) Tanya wishes to lie on a hammock and sip on a Pina Colada. (c) Johan wishes to ski on the Swiss slopes. (d) George is attending a meeting in Shanghai to sort out problems with the production line. (e) Josephine is heading to the University of Sydney Business School for 3 years from Hong Kong to complete a degree in marketing. (f) Tony is attending an aviation conference in Washington. (g) Harriet is flying to Brisbane to start a new job and a new life. (h) Lukas is flying home for Christmas. (i) Scott is going to see a football match between England and Sweden in Stockholm on the weekend in a world cup qualifier.

2.

Determine whether the following goods and services are substitutes, complements or neither to air travel. (a) Buying a seat that is closer to the front of the plane. (b) Buying extra weight for checked-in baggage. (c) Buying travel insurance. (d) Renting a car to drive to a holiday destination. (e) Renting a car on arrival at the holiday destination. (f) Paying for a taxi to get from the airport to the hotel. (g) Buying a packet of cigarettes. (h) Buying a hotdog at a football game. (i) Buying a snack at the airport. (j) Buying a meal at the hotel at which you are staying. (k) Travelling to Hawaii rather than Bali. (l) Going on a cruise to Vanuatu.

3.4 Organic Air Travel Demand The organic demand for air travel represents the component of air travel demand that is not driven by the average airfare. Organic air travel demand is usually driven by many different forces. Income is the most important force. This includes the income of the passenger or the household, and the income of the employer who is paying employees to travel by air. The income of households and passengers consists of labour and non-labour income. Labour income is the income that passengers gain from being employed. Non-labour income includes interest, dividend, and rental income. This is the income that passengers receive from owning assets that generate a rate of return. When modelling and investigating the drivers of air travel demand in practise, the variables that are often used to approximate income are Gross Domestic Product, household income and labour hours worked. Another key driver of airline organic demand is wealth. The wealth of a passenger or household is the assets less the liabilities of that passenger or household. The main assets include the house owned by the household, the motor vehicles that are driven by the household, and consumer durables used by the household such as TVs and home entertainment products, computer equipment, white goods and mobile telephone products and accessories. The liabilities of the household include mortgage, personal, and credit card debt. There are some passengers that pay for their air travel from their wealth, including passengers who have retired from the workforce. As the wealth of the passenger or household increases this can be used to pay for more air travel. For modelling and analysing air travel demand in practise, the best proxy variable to use for wealth is an index of movements in the stock market. Examples of such indices include the Dow Jones Industrial Average in the U.S., the Shanghai Composite Index in China, the FTSE Index in the U.K., the Nikkei Index in Japan, the Dax Index in Germany, the Strait Times Index in Singapore, and the All-Ordinaries Index in Australia. The size of the population is another important driver of the organic demand for air travel. When there are more people in a population, this increases the pool of people from which air travel demand is drawn. A large population of people, however, will only buy a ticket to travel by air if they can afford that ticket, or if air services are available near to where they live. It is for this reason that it is necessary to combine information about the size of the population and the income per capita of that population to understand how this may influence the organic demand for air travel. The confidence of both individuals and households as well as the business community is an important driver of air travel organic demand. An improvement in confidence, particularly about the security of employment, means that consumers and employers will spend a greater proportion of their extra income or wealth on discretionary spending, which includes air travel. In terms of analysing air travel demand in practise, the data that we use to understand what is happening to confidence will include household and business confidence indicators, as well as indices that capture

Passenger Demand

45

movements in share prices. An example of a consumer confidence indicator is the University of Michigan consumer sentiment index.12 There are many different transport substitutes to air travel as indicated in section 3.2. These include travel by land and travel by sea. A transport mode is a substitute for air travel when an increase in the price of the land or sea transport mode leads to an increase in air travel demand. An increase in the price of substitutes to air transport will increase the organic demand for air travel. Seasonality is another important driver of the organic demand for air travel. Air travel is stronger during the summer months compared to winter months, as leisure and holiday travellers take advantage of school holiday periods and better weather conditions for travelling and holidaying. Passengers who travel to visit friends and relatives do so extensively during festive seasons, such as Christmas, Chinese New Year, Easter, and the Mecca pilgrimage. They also travel on weekends for sporting and other entertainment events, or to take advantage of long weekends. As indicated in section 3.3 above, the organic component of air travel demand is also affected by the price of complements. This includes the price of accommodation, food and beverages and land transport. For example, a resident of Chicago flies from O’Hare International Airport (ORD) to Sangster International Airport Montego Bay (MBJ) on a holiday flying with the large U.S. carrier Delta Air Lines. The passenger pays US$402 each way for the air trip. When the passenger arrives in Montego Bay, she stays at the Hotel Montego paying US$117 per night, staying for 5 nights. If the price of the Hotel Montego were to increase to US$150 per night this will not only reduce the number of nights demanded at the Hotel, but it will also reduce the number of passengers flying with Delta between ORD and MBJ. The exchange rate is another important organic demand driver of air travel demand. This is because changes in the exchange rate effect the prices that travellers pay on arrival at their destination. For example, on arrival in Montego Bay our tourist from Chicago will pay for transport from the airport to Hotel Montego, the tourist will buy food and beverages at restaurants, will be entertained at Montego Bay bars and clubs, and will ride in a taxi to get from the hotel to various tourist attractions. These products and services are paid for in Jamaican dollars. If the US dollar price of 1 Jamaican dollar increases this makes it more expensive for the U.S. tourist to pay for these items while holidaying in Jamaica. This in turn will reduce the number of Chicago and U.S. tourists travelling to Montego Bay which will reduce the demand for air travel between the U.S. and Montego Bay. Conversely, if the US dollar price of 1 Jamaican dollar decreases, then this reduces the cost of buying goods and services in Montego Bay, which promotes more U.S. visitors to travel to Montego Bay and greater air travel demand. There are many households that borrow money to buy their home and/or car and accumulate significant debt on their credit card. When interest rates increase in the economy, including mortgage interest rates, interest rates on personal debt, and interest rates on credit cards, this means the household must pay more to the lender in interest. When more is paid to the lender in interest, for the same household income this means the household has less money to pay for travel by air. This reduces the organic demand for air travel. There are also households, however, that are not in debt. They own their home and car, and they always pay off their credit card on time. These households have a significant store of their assets held as cash in the bank, because they need their assets held in a relatively liquid form to pay for goods and services, including air travel. When interest rates go up, the income of these households go up which in turn increases their spending, including their spending on air travel. Conversely, when interest rates are low, the income these households receive from their stock of assets is relatively low, which reduces their spending on discretionary goods and services, including travel by air.

3.5 Airline Specific Demand 3.5.1 An Introduction to Airline Demand Airline specific demand is the demand for a particular airline on a particular route. For example, on the city pair Manila (MNL) to Jakarta (CGK) there are two airlines operating services – Cebu Pacific (5J) and Philippine Airlines (PR). On this city pair, we can describe the demand for a seat on Cebu Pacific Airlines and the demand for a seat on Philippine Airlines. The demand for Cebu Pacific is an airline-specific demand and the demand for Philippine Airlines on this route is also an airline specific demand. There are five sets of drivers of airline specific demand for a particular route. The first is the airfare charged by the airline in question on the route in question. The second is the airfare charge by other airlines that also operate services on the route. The third is the range of organic demand factors presented in section 3.4 above. The fourth is the marketing and advertising spending of the airline in question. The fifth is the quality of the product of the airline in question compared to the product of the other airlines that operate services on the route.

3.5.2 Airline Demand Functions To describe airline demand we introduce the concept of the airline demand function for a particular route or city pair. The airline demand function is a function that describes the demand for a seat on the airline as a function of the average airfares of the airlines that compete on the route. If there are just two airlines competing on a route, A and B, and the

12

For the University of Michigan consumer sentiment survey website click on http://www.sca.isr.umich.edu/.

Chapter 3

46

demand function is a linear function then we can write the demand functions for airlines A and B on the route in question as follows: DemandA = a0 – a1PA + a2PB DemandB = b0 – b1PB + b2PA

(3.1) (3.2)

where DemandA is the passenger demand on the route in question for airline A, DemandB is the passenger demand on the route in question for airline B, PA is the average airfare that passengers pay for travel on airline A, and PB is the average airfare paid by passengers for travel on airline B for the route in question. The terms a0 and b0 are parameters that represent the organic demand drivers that influence airline demand, or the non-airfare variables that influence airline demand. These are also referred to as the demand intercept parameters. The airline demand functions (3.1) and (3.2) are linear functions because the price variables on the right-hand side of those functions are (a) raised to the power of an imaginary 1, and (b) the price variables are not multiplied by each other (that is, the price variables are separated). The coefficient a1 is positive and represents the impact of a 1 unit increase in the average airfare of airline A on the demand for airline A on the route in question. As a1 is positive then an increase in PA results in a decrease in DemandA. This is also referred to as an own-price effect – in this case the own-price effect for airline A. The coefficient a2 is also positive and is referred to as a cross-price effect for airline A. This measures the impact of a 1 unit increase in the price of airline B on the demand for airline A. As the cross-price effect a2 is positive this means an increase in the average airfare charged by airline B results in an increase in the demand for airline A. The coefficient b1 is positive and represents the impact of a 1 unit increase in the average airfare of airline B on the demand for airline B for the route in question. As b1 is positive then an increase in PB results in a decrease in DemandB, which is the own-price effect for airline B. The coefficient b2 is also positive and is referred to as a cross-price effect for airline B. This measures the impact of a 1 unit increase in the price of airline A on the demand for airline B. As the cross-price effect b2 is positive this means an increase in the average airfare charged by airline A results in an increase in the demand for airline B.

3.5.3 Airline Demand Curves We can graph the demands for airlines A and B at (3.1) and (3.2) with demand on the vertical axis and the average airfare on the horizontal axis. These are referred to as airline demand curves – refer to Figure 3-1 below. PAXA

PAXB

a0+a2PB b0+b2PA

DemandB

DemandA 0

(a0+a2PB)/a1

PA

0

(b0+b2PA)/b1

PB

Fig. 3-1: Airline Demand Curves for Airlines A and B The demand curve on the left in Figure 3-1 is the demand curve for airline A and on the right is the demand curve for airline B for the route in question. Critical to drawing these demand curves correctly is identifying the points where the demand curve meets the vertical and horizontal axes. The points where the demand curves meet the vertical axes are those where PA and PB equal 0 in airline demand functions A and B, respectively. If we substitute these values into our demand functions (3.1) and (3.2) respectively we obtain: DemandA = a0 – a1(0) + a2PB = a0 + a2PB DemandB = b0 – b1(0) + b2PA = b0 + b2PA

(3.3) (3.4)

We can see in (3.3) and (3.4) these are the points where the demand curves meet the vertical axes in Figure 3-1. To obtain the points where the demand curves cut the horizontal axes, we set DemandA = DemandB = 0 in equations (3.1) and (3.2) as follows:

Passenger Demand

47

DemandA = a0 – a1PA + a2PB = 0 DemandB = b0 – b1PB + b2PA = 0

(3.5) (3.6)

We then solve equation (3.5) for PA and equation (3.6) for PB: P୅ = P୆ =

ୟబ ାୟమ ୔ా

(3.7)

ୟభ ୠబ ାୠమ ୔ఽ

(3.8)

ୠభ

Equations (3.7) and (3.8) tell us where the bottom of the demand curves cut the horizontal axes. We can see this in Figure 3-1 above. Let us now look at a numerical example of the general demand curves (3.1) and (3.2) above in the case of Cebu Pacific and Philippine Airlines services between Manila (MNL) and Jakarta (CGK). In the case of Cebu Pacific services on MNL to CGK, the demand will depend on the average airfare that Cebu Pacific charges on the route and the average airfare that is charged by Philippine Airlines on the route. It will also depend on the quality of the Cebu Pacific service on the route relative to the quality of the Philippine Airlines service. The macroeconomic performances of Manilla and Jakarta, as represented by variables such as Gross Domestic Product, household income, population, and wealth in both cities, will also determine the organic demand for Cebu Pacific on the route. Finally, Cebu Pacific demand on MNLCGK will depend on the quantity and quality of Cebu Pacific advertising on the route and the media channels that the airline uses for its advertising. The following is an example of a simple linear demand function for the airline product of Cebu Pacific on a flight operated by an Airbus A320 aircraft between Manilla and the Philippines: ୑୒୐ିେୋ୏ ୑୒୐ିେୋ୏ Demand୑୒୐ିେୋ୏ = 180 – 0.005Pେୣୠ୳ + 0.0025P୔୅ େୣୠ୳

(3.9)

ெே௅ି஼ீ௄ ெே௅ି஼ீ௄ where ‫݀݊ܽ݉݁ܦ‬஼௘௕௨ is the demand for a seat on a Cebu Pacific flight between Manilla and Jakarta, ܲ஼௘௕௨ is ெே௅ି஼ீ௄ is the average airfare charged by Cebu Pacific on the MNL-CGK flight denominated in Philippine Pesos and ܲ௉஺ the average airfare charged by Philippine Airlines on the MNL-CGK flight denominated in Philippine Pesos. We can see ெே௅ି஼ீ௄ ெே௅ି஼ீ௄ = DemandA, a0 = 180, a1 = 0.005, a2 = 0.0025, ܲ஼௘௕௨ that (3.9) is the same as equation (3.1) with ‫݀݊ܽ݉݁ܦ‬஼௘௕௨ ெே௅ି஼ீ௄ = PB. = PA , and ܲ௉஺ We can use the information in equation (3.9) to construct a demand schedule for Cebu Pacific on MNL-CGK. The demand schedule sets out the changing demand for the flight as we change the average airfare charged by Cebu Pacific, and we assume that the average airfare charged by Philippine Airlines is unchanged. If we set the Philippine Airlines average airfare at 20,000 Philippine Pesos, then we obtain the demand schedule presented in Table 3-1 below for Cebu Pacific on the Manila to Jakarta route.

୑୒୐ିେୋ୏ (Philippine Peso) Pେୣୠ୳ 0 5000 10000 15000 20000 25000 30000 35000 40000

Demand୑୒୐ିେୋ୏ େୣୠ୳ 230 205 180 155 130 105 80 55 30

Table 3-1: Demand Schedule for Cebu Pacific on Manila-Jakarta The demand schedule in Table 3-1 indicates that the vertical intercept is 230 and that demand falls by 25 passengers each time Cebu lifts its average airfare by 5,000 Philippine Peso. The Cebu Pacific demand curve that corresponds with the demand schedule presented in Table 3-1 is drawn below in Figure 3-2. Now let us assume that the Philippine Airline’s demand function is: ୑୒୐ିେୋ୏ ୑୒୐ିେୋ୏ Demand୑୒୐ିେୋ୏ = 180 – 0.004P୔୅ + 0.002Pେୣୠ୳ ୔୅

(3.10)

Chapter 3

48

Cebu Demand MNL-CGK 250 200 150 100 50 ୑୒୐ିେୋ୏ Demand୑୒୐ିେୋ୏ = f(P୔୅ = 20,000) େୣୠ୳

0 0

5,000

10,000

15,000

20,000

25,000

30,000

35,000

40,000

Cebu Average Airfare (PHP) Fig. 3-2: Cebu Demand Curve on Manila-Jakarta If we draw the demand curve for (3.10) assuming that the Cebu Pacific average airfare is fixed at 11,000 Philippine Pesos we obtain the Philippine Airlines demand curve drawn in Figure 3-3 below.

Philippine Airline Demand MNL-CGK 250 200 150 100 50 ‫ିۺۼۻ‬۱۵۹ ۲‫ିۺۼۻ܌ܖ܉ܕ܍‬۱۵۹ = ܎(‫۾‬۱‫ܝ܊܍‬ = ૚૚, ૙૙૙) ‫ۯ۾‬

0 0

5,000

10,000

15,000

20,000

25,000

30,000

35,000

40,000

Philippine Airlines Airfare (PHP) Fig. 3-3: Philippine Airlines Demand Curve on Manila-Jakarta We see that Philippine Airline’s demand curve in Figure 3-3 slopes downward from left to right like Cebu Pacific, but it has a lower vertical intercept and a higher horizontal intercept. Let us now examine what happens when one airline improves its product offering. This may include the airline offering better food and beverages on the flight, making improvements to its airport lounges at the origin or destination airports, improving on-time performance and reducing the queuing time at check-in and baggage collection. This results in an increase in the intercept term in the airline demand equation. For example, if airline A in equation (3.1) invests in its product so that its product offering has improved, its demand function will now be: Demand = a3 – a1PA + a2PB

(3.11)

where a3 > a0. Figure 3-4 below presents the general case of the demand curves for airlines A and B when there is an improvement in the product offered by airline A. We can see in Figure 3-4 that this results in the demand curve of airline A shifting to the right while the demand curve of airline B remains unchanged. Let us now assume in our numerical example that Cebu Pacific makes an improvement to its product offering, resulting in the Cebu Pacific demand function intercept increasing in (3.9) so that the demand function becomes: ୑୒୐ିେୋ୏ ୑୒୐ିେୋ୏ Demand୑୒୐ିେୋ୏ = 250 – 0.005Pେୣୠ୳ + 0.0025P୔୅ େୣୠ୳

(3.12)

Passenger Demand

PAXA

49

PAXB

a3+a2PB

b0+b2PA

a0+a2PB

0

DemandB

۲‫܌ܖ܉ܕ܍‬૛‫ۯ‬

۲‫܌ܖ܉ܕ܍‬૚‫ۯ‬

(a0+a2PB)/a1 (a3+a2PB)/a1

PA 0

(b0+b2PA)/b1

PB

Fig. 3-4: Airline Demand Curves with an Improvement in the Product of Airline A The new Cebu Pacific demand curve after the product improvement is show in Figure 3-5 below. Cebu Demand MNL-CGK 350 Old Demand

300

New Demand

250 200 150 100 50 0 0

5,000

10,000

15,000

20,000

25,000

30,000

35,000

40,000

Cebu Average Airfare (PHP) Fig. 3-5: Cebu Demand Curve on Manila-Jakarta after Product Improvement Figure 3-5 indicates that the new Cebu Pacific demand curve sits to the right and above the old demand curve, with demand higher by 70 passengers at each average airfare level.

3.6 Market Demand 3.6.1 Market Demand Function Market demand represents the sum of the demand of every airline that services a particular route. In the case of our airline-specific demand functions (3.1) and (3.2) the market demand function is found by adding (3.1) and (3.2) together in the following way: Market Demand = DemandA + DemandB = a0 + b0 – (a1 - b2)PA – (b1 - a2)PB

(3.13)

We can see in the case of the market demand function at (3.13) that demand depends on the average airfare of both airlines. If we set these average airfares equal so that we have a single average airfare PA = PB = P, which is the market average airfare, then market demand at (3.13) becomes: Market Demand = a0 + b0 – [(a1 - b2) + (b1 - a2)]P The market demand curve that corresponds to equation (3.14) is draw in Figure 3-6 below.

(3.14)

Chapter 3

50

PAX a0+b0

Demand P

0

(a0+b0)/(a1–b2+b1-a2)

Fig. 3-6: Market Demand Curve for Air Travel Figure 3-6 indicates a linear market demand function, with vertical and horizonal intercepts that depend on the organic demand drivers, changes in the quality of the airline products, and marketing and advertising expenditure of the two airlines. To demonstrate the market demand function numerically we refer back to the Cebu Pacific and Philippine Airlines example. Adding the demand functions of Cebu Pacific and Philippine Airlines at equations (3.9) and (3.10) together yields the market demand function for the MNK-CGK route: Demand୑୒୐ିେୋ୏ = Demand୑୒୐ିେୋ୏ + Demand୑୒୐ିେୋ୏ = 360 െ 0.0045P୑୒୐ିେୋ୏ ୔୅ େୣୠ୳ If we graph the market demand curve that corresponds with this demand function, we obtain Figure 3-7 below. Demand MNL-CGK

400 350 300 250 200 150 100 50 0

0

5,000 10,000 15,000 20,000 25,000 30,000 35,000 40,000 45,000 50,000 55,000

Market Average Airfare (PHP) Fig. 3-7: Market Demand Curve for Air Travel – Manila to Jakarta We can determine the demand for the MNL-CGK market by substituting a value for the market average airfare into the market demand function. This is left as an exercise for the interested reader.

3.6.2 Market Inverse Demand Rather than writing market demand as a function of the average airfare, we sometimes wish to write the average airfare as a function of market demand. This is referred to as the inverse demand curve for air travel. It is referred to as the inverse demand curve because we find it by finding the inverse of the market demand curve. To see how this is done, consider our market demand function at equation (3.14). We wish to re-write this function so that P is the subject and market demand is the independent variable. To do this, we first simplify equation (3.14) by using PAX to represent market demand and we then simplify the notation we use for the coefficient attached to price and the intercept as follows: PAX = E0 + E1P

(3.15)

Passenger Demand

51

where E0 { a0 + b0 and E1 { – [(a1 - b2) + (b1 - a2)]. To find the inverse demand function we rearrange (3.15) so that P is the subject as follows: P=

୔୅ଡ଼ିஒబ ஒభ

=

୔୅ଡ଼ ஒభ

െ

ஒబ

(3.16)

ஒభ

Equation (3.16) is the inverse demand function. To draw the inverse demand curve, we draw a two-dimensional diagram with P on the vertical axis and PAX on the horizontal axis. This generates the curve drawn in Figure 3-8 below. P െ

Ⱦ଴ Ⱦଵ

Inverse Demand 0

E0

PAX

Fig. 3-8: Market Inverse Demand Curve for Air Travel The point where the inverse demand curve meets the vertical axis is called the reservation price and is equal to -E0/E1. This is effectively the highest price that any single passenger is willing to pay on a flight. The inverse demand curve is an important curve if we wish to know the average airfare that would be needed to serve a given level of demand. A numerical example of finding the inverse demand curve from the market demand curve is given in Quiz 3-2 below.

Quiz 3-2 Organic, Market and Airline Level Demand 1.

(a) (b) (c) (d) (e) (f) (g) (h) (i) (j) (k) (l) (m) (n) (o) (p) 2.

Consider the market for flying from London Gatwick Airport to Ibiza. This route is serviced by three carriers – easyJet, British Airways and Norwegian Air International. The flight takes around two and a half hours. In the list of demand drivers below, identify the variables that are more likely to be organic demand drivers, market demand drivers, airline level demand drivers or a variable that is unlikely to affect demand on this route. The income of consumers who live in England in a location that is proximate to any of the London airports. The population of consumers who live in England in a location that is proximate to any of the London airports. The airfares offered by British Airways relative to easyJet and Norwegian Air International on London Gatwick to Ibiza. The safety record of British Airways relative to easyJet and Norwegian Air International. The weather in Ibiza. The price of accommodation in Ibiza. The cost of a cruise to Ibiza which departs from London. The price of accommodation in the Greek Islands. Airfares offered by British Airways on flights between London Heathrow and Berlin. The Great British Pound versus Euro exchange rate. The Great British Pound versus US dollar exchange rate. The Egyptian Pound versus the Turkish Lira exchange rate. British Airways decides to fly slower between London Gatwick and Ibiza to save on fuel costs. easyJet decides to charge people on flights between London Gatwick and Ibiza to use the on-board toilet. Norwegian Air International decides to stop in Oslo to pick up passengers on the way to Ibiza. Gatwick Airport decides to raise airport charges while London Heathrow, London City and Luton Airports keep airport charges unchanged. Consider the city pair Buenos Aires (EZE), Argentina to Santiago (SCL), Chile. This is serviced by two airlines – LATAM (LA) and Sky Airways (H2). LATAM is a full-service carrier that has many hubs around South America, while Sky Airways is a Chilean low-cost carrier with a SCL hub. The flying time for the route is estimated to be around 2 hours and 20 minutes. The LATAM airlines flight is LA 468 and the Sky Airways flight is H2 508. The aircraft used by LATAM is an Airbus A320-200 and that used by Sky Airlines is an Airbus A320neo. The demand functions estimated for the two airlines on EZE-SCL are as follows:

Chapter 3

52

୉୞୉ିୗେ୐ ୉୞୉ିୗେ୐ Demand୉୞୉ିୗେ୐ ୐୅୘୅୑ = 300 െ 0.3P୐୅୘୅୑ + 0.15Pୗ୏ଢ଼ ୉୞୉ିୗେ୐ ୉୞୉ିୗେ୐ Demand୉୞୉ିୗେ୐ = 190 െ 0.4Pୗ୏ଢ଼ + 0.2P୐୅୘୅୑ ୗ୏ଢ଼

Use this information to answer questions (a) through to (g) below. (a) What is the demand for LATAM Airlines on the Buenos Aires to Santiago route at average airfares of ா௓ாିௌ஼௅ ா௓ாିௌ஼௅ = $225 and ܲௌ௄௒ = $215? ܲ௅஺்஺ெ (b) Draw the LATAM and SKY Airways demand curves. Ensure that you properly label the vertical and horizontal axes and that you identify where the curves meet the vertical and horizontal axes. ா௓ாିௌ஼௅ ா௓ாିௌ஼௅ (c) Interpret the coefficient attached to ܲ௅஺்஺ெ in both demand functions and the coefficient attached to ܲௌ௄௒ in both demand functions. (d) What types of factors influence the intercepts 300 and 190 in the airline specific demand functions? (e) Present a schematic representation of the shift in the LATAM demand curve if LATAM improves the quality of its air travel product. (f) Derive the market demand function and draw the market demand curve. (g) Find the inverse demand function and draw the inverse demand curve.

3.7 The Airfare Elasticity of Air Travel Demand 3.7.1 Demand Elasticity Versus Demand Sensitivity In sections 3.5 and 3.6 we investigated the airline specific and market demand functions for air travel. These functions expressed the demand for air travel as a function of the average airfare. The specific demand functions that we used for illustrative purposes are linear demand functions. In these linear demand functions, the coefficient attached to the average airfare identifies the sensitivity of demand to a change in the average airfare because it measures the impact on air travel demand in terms of the number of passengers travelling of a 1-unit change in the average airfare. In this section we analyse in detail another measure of the extent to which the demand for air travel changes in response to a change in the average airfare. This measure is called the elasticity of air travel demand to a change in the average airfare. It examines the percentage change in air travel demand in response to a percentage change in the average airfare. This differs from the sensitivity of air travel demand to a change in the average airfare, which measures the change in demand (not the percentage change) in response to a 1-unit change in the average airfare (not a percentage change in the average airfare).

3.7.2 Market Fare Elasticity of Air Travel Demand The market airfare elasticity of air travel demand measures the percentage reduction in market air travel demand in response to a 1% increase in the market average airfare, which we can write as: Market Airfare Elasticity of Air Travel Demand = Knowing that the percentage change in demand is ஼௛௔௡௚௘ ௜௡ ஺௜௥௙௔௥௘

as:

஺௜௥௙௔௥௘

% େ୦ୟ୬୥ୣ ୅୧୰ ୘୰ୟ୴ୣ୪ ୈୣ୫ୟ୬ୢ

(3.17)

% େ୦ୟ୬୥ୣ ୧୬ ୅୴ୣ୰ୟ୥ୣ ୅୧୰୤ୟ୰ୣ

஼௛௔௡௚௘ ௜௡ ஽௘௠௔௡ௗ ஽௘௠௔௡ௗ

and the percentage change in the average airfare is

then we can write the market elasticity of air travel demand to a change in the average airfare at (3.17)

Market Airfare Elasticity of Air Travel Demand =

େ୦ୟ୬୥ୣ ୅୧୰ ୘୰ୟ୴ୣ୪ ୈୣ୫ୟ୬ୢ ୈୣ୫ୟ୬ୢ

÷

େ୦ୟ୬୥ୣ ୧୬ ୅୴ୣ୰ୟ୥ୣ ୅୧୰୤ୟ୰ୣ ୅୴ୣ୰ୟ୥ୣ ୅୧୰୤ୟ୰ୣ

By inverting the right-hand side and then multiplying, this elasticity term can be written as: Market Airfare Elasticity of Air Travel Demand =

େ୦ୟ୬୥ୣ ୅୧୰ ୘୰ୟ୴ୣ୪ ୈୣ୫ୟ୬ୢ େ୦ୟ୬୥ୣ ୧୬ ୅୴ୣ୰ୟ୥ୣ ୅୧୰୤ୟ୰ୣ

×

୅୴ୣ୰ୟ୥ୣ ୅୧୰୤ୟ୰ୣ ୅୧୰ ୘୰ୟ୴ୣ୪ ୈୣ୫ୟ୬ୢ

(3.18)

This formula for the market airfare elasticity of air travel demand is called the point price elasticity of demand formula. The right-hand side of (3.18) has two component parts: େ୦ୟ୬୥ୣ ୅୧୰ ୘୰ୟ୴ୣ୪ ୈୣ୫ୟ୬ୢ େ୦ୟ୬୥ୣ ୧୬ ୅୴ୣ୰ୟ୥ୣ ୅୧୰୤ୟ୰ୣ ୅୴ୣ୰ୟ୥ୣ ୅୧୰୤ୟ୰ୣ ୈୣ୫ୟ୬ୢ

= slope of the market air travel demand curve

= inverse of the slope of the ray from the origin to the market demand curve

Let us illustrate these two component parts in the case of a linear demand curve for air travel – refer to Figure 3-9 below.

Passenger Demand

53

PAX

Slope of this line =

େ୦ୟ୬୥ୣ ୧୬ ୈୣ୫ୟ୬ୢ େ୦ୟ୬୥ୣ ୧୬ ୅୧୰୤ୟ୰ୣ

A

PAX0

Slope of this line =

୔୅ଡ଼బ ୔బ

Demand 0

P

P0

Fig. 3-9: Airfare Elasticity of Air Travel Demand on the Market Demand Curve Using Figure 3-9, the market average airfare elasticity of air travel demand at point A on the demand curve is equal to the slope of the firm market demand line divided by the slope of the dashed line drawn from the origin to point A. The dashed line is called a ray. It is defined as a line from the origin to a particular point on the demand curve, in our case point A in Figure 3-9. As the slope of the ray is different for different points on the demand curve then this tells us that the airfare elasticity of air travel demand is different for different points on the demand curve. This is an important property of the linear or straight-line demand curve. We demonstrate this property in Figure 3-10 below. PAX

Slope of this line = PAX2

୔୅ଡ଼మ ୔మ

C

Slope of this line =

୔୅ଡ଼భ ୔భ

B

PAX1

Demand 0

P2

P

P1

Fig. 3-10: Airfare Elasticity of Air Travel Demand Varies along the Market Demand Curve Figure 3-10 indicates that the slope of the ray from the origin to point C, equal to ray from the origin to point B, which is

௉஺௑భ ௉భ

௉஺௑మ ௉మ

, is steeper than the slope of the

. As the fare elasticity of air travel demand is equal to the slope of the

demand curve multiplied by the inverse of the slope of the ray line, then the airfare elasticity of air travel demand is greater in absolute terms at point B than at point C. We can show this using equation (3.18) as follows:

Chapter 3

54

Airfare Elasticity of Air Travel Demand at B = ቂ Airfare Elasticity of Air Travel Demand at C = ቂ

୔୅ଡ଼మ ି୔୅ଡ଼భ ୔మ ି୔భ ୔୅ଡ଼మ ି୔୅ଡ଼భ ୔మ ି୔భ

ቃ× ቃ×

୔భ

(3.19)

୔୅ଡ଼భ ୔మ

(3.20)

୔୅ଡ଼మ

We can see that the elasticity is greater in absolute terms at (3.19) than at (3.20) because elasticity expressions that the slope of the market demand curve component,

௉஺௑మ ି௉஺௑భ ௉భ ି௉బ

௉భ ௉஺௑భ

>

௉మ ௉஺௑మ

. Notice in these

, is the same in both equations.

Let us now demonstrate how to compute a numerical value for the elasticity using the market demand function for Manila to Jakarta from section 3.6, which we restate now for your convenience: Demand୑୒୐ିେୋ୏ = 360 െ 0.0045P୑୒୐ିେୋ୏ This is the market demand curve in Figure 3-7. To use this market demand equation to find the airfare elasticity of air travel demand for the route MNL-CGK we first find the slope of the demand function. This is equal to the coefficient attached to the average airfare in the market demand equation, which is: Slope of Demand =

େ୦ୟ୬୥ୣ ୅୧୰ ୘୰ୟ୴ୣ୪ ୈୣ୫ୟ୬ୢ େ୦ୟ୬୥ୣ ୧୬ ୅୴ୣ୰ୟ୥ୣ ୅୧୰୤ୟ୰ୣ

= -0.0045

Notice that this slope does not depend on the average airfare. It is the same regardless of the average airfare. We next multiply the slope of the demand curve by the average airfare divided by the level of demand as follows: Airfare Elasticity of Travel Demand = െ0.0045 ×

୅୧୰୤ୟ୰ୣ ୈୣ୫ୟ୬ୢ

=

ି଴.଴଴ସହ୔౉ొైషిృే ୈୣ୫ୟ୬ୢ౉ొైషిృే

At this point you will notice that the airfare elasticity of air travel demand depends on the market average airfare, PMNLCGK and the level of air travel demand, DemandMNL-CGK. We next substitute market demand as a function of the average airfare into the denominator of our elasticity expression as follows: Airfare Elasticity of Travel Demand =

ି଴.଴଴ସହ୔౉ొైషిృే

(3.21)

ଷ଺଴ି଴.଴଴ସହ୔౉ొైషిృే

We can see in (3.21) that the price elasticity of demand is a complex non-linear function of the average airfare, PMNLCGK . To obtain a particular numerical value for the elasticity we must plug into equation (3.21) a particular value for the average airfare. Let us suppose that PMNL-CGK = 30,000. Substituting this into (3.21) allows us to find the airfare elasticity of air travel demand for the city pair MNL-CGK at an average airfare of 30,000 Philippine Peso: Airfare Elasticity of Travel Demand (PMNL-CGK = 30,000) =

ି଴.଴଴ସହ(ଷ଴,଴଴଴) ଷ଺଴ି଴.଴଴ସହ(ଷ଴,଴଴଴)

= െ0.6

How do we interpret the airfare elasticity -0.6? It means that a 1% increase in the average airfare from a value of PMNL= 30,000 approximately results in a 0.6% reduction in market demand on the route MNL-CGK.13 Let us now find the value of the airfare elasticity of air travel demand at an average airfare of PMNL-CGK = 50,000. In this case the air travel elasticity estimate is:

CGK

Airfare Elasticity of Travel Demand (PMNL-CGK = 50,000) =

ି଴.଴଴ସହ(ହ଴,଴଴଴) ଷ଺଴ି଴.଴଴ସହ(ହ଴,଴଴଴)

= െ1.67

This elasticity indicates that a 1% increase in the average airfare leads to an approximate 1.67% reduction in air travel demand. What this demonstrates is that for a linear demand curve, as the average airfare is increased the airfare elasticity of demand for air travel increases in absolute terms. This means that demand becomes more sensitive to price or more elastic as the average airfare increases. We can define the following three categories of the airfare elasticity of air travel demand. The first is inelastic air travel demand, in which case a 1% increase in the average airfare leads to a reduction in the demand for air travel that is less than 1%. In this case air travel demand is relatively insensitive to a change in the average airfare. For example, if the elasticity of air travel demand to a change in the average airfare is -0.5, this means that a 1% increase in the average airfare leads to a 0.5% reduction in air travel demand. The second is elastic air travel demand, in which case a 1% increase in the average airfare leads to a reduction in the demand for air travel that is more than 1%. For example, if the elasticity of air travel demand to a change in the average airfare is -1.5 this means that a 1% increase in the average airfare leads to a 1.5% reduction in air travel demand. The third is unit elastic demand, in which case a 1% increase in the average airfare leads to a 1% reduction in the demand for air travel. When air travel demand is highly responsive to 13

I use the word approximately in this interpretation because the elasticity is constructed in a way that it can only be interpreted for a very small increase in price, as opposed to a 1% increase in price.

Passenger Demand

55

the average airfare, we say that air travel demand is elastic and when it is relatively unresponsive, we say that air travel demand is inelastic to the average airfare.

3.7.3 Drivers of the Market Fare Elasticity of Air Travel Demand Any variable that shifts the market demand curve for air travel will influence the airfare elasticity of air travel demand because it affects the slope of the ray from the origin to a point on the market demand curve. This includes such variables as the income and wealth of passengers, the exchange rate, the price of substitutes and complements, product improvements of airlines and marketing and advertising expenditure of airlines. To see how a shift in the demand curve affects the airfare elasticity of air travel demand, consider Figure 3-11 below. PAX

C

PAX2

B

PAX1

A

D1

D2 P

P1

0

Fig. 3-11: Airfare Elasticity of Air Travel Demand and Shifts in the Demand Curve In Figure 3-11 there is a parallel outward shift in the market demand curve for air travel from D1 to D2. The parallel shift means that the slope of the demand curve has not changed – remember this is an important component of the airfare elasticity of air travel demand. For the same average airfare at P1 there is an increase in the demand for passengers from PAX1 to PAX2. This can be read-off the market demand curve as a change from point A to point B. The ray from the origin to point A is flatter than the ray from the origin to point B. This means that the airfare elasticity of demand is bigger in absolute terms at point A than point B. This in turn means that demand is more elastic at point A than point B. Let us show how this is the case through a bit of simple geometry. The elasticity at point A can be found in the following way: Elasticity at Point A = Slope u

୅୧୰୤ୟ୰ୣ ୟ୲ ୅ ୈୣ୫ୟ୬ୢ ୟ୲ ୅

=

஼ି௉஺௑భ

=

஼ି௉஺௑భ

଴௉భ

×

଴௉భ ଴௉஺௑భ

=

஼ି௉஺௑భ

=

஼ି௉஺௑భ

଴௉஺௑భ

The elasticity at Point B is equal to: Elasticity at Point B = Slope u

୅୧୰୤ୟ୰ୣ ୟ୲ ୆ ୈୣ୫ୟ୬ୢ ୟ୲ ୆

଴௉భ

×

଴௉భ ଴௉஺௑మ

଴௉஺௑మ

As 0PAX2 > 0PAX1 then the elasticity at A is greater in absolute terms than the elasticity at B, which means that demand is more elastic at A than at B. The shift in the market demand curve to the right results in air travel demand that is more inelastic to the average airfare. This in turn means that when the economy is stronger, air travel demand is typically more inelastic to the average airfare, because a stronger economy results in a shift in the travel demand curve to the right. Similarly, when the economy is weaker and the air travel demand curve shifts in and to the left, this is associated with more elastic air travel demand to the average airfare.

Chapter 3

56

3.7.4 Cross-Mode Market Elasticity of Air Travel Demand The cross-mode market elasticity of air travel demand measures the impact on market air travel demand of a change in the price of a mode of transport that is a substitute to air travel. This includes changes in the price of travel by land, such as the price of using your own car, a rental car, a bus, or a train. It may also include a change in the price of an ocean trip on a ferry or an ocean liner. A change in the price of a substitute mode of transport causes a shift in the market air travel demand curve. An increase in the price of a substitute mode of transport causes the market demand curve for air travel to shift to the right, as indicated in Figure 3.11. In this case the demand for air transport increases for each level of the average airfare. A decrease in the price of a substitute mode of transport causes the market demand curve for air transport to shift to the left, causing lower air travel demand for each level of the average airfare. To identify the cross-mode market elasticity of air travel demand we need to specify the market air travel demand function with the price of a substitute mode of transport included as a right-hand side variable. We do this by adding a term to the market air travel demand function presented at equation (3.15) as follows: PAX = E0 + E1PAir + E2PLand

(3.22)

where PLand is the price of a land mode of transport and PAir is the market average airfare. We could have also added the price of a sea mode of transport as well, but we have omitted such a variable to keep things simple. The elasticity of air travel demand PAX with respect to a change in the price of land transport is equal to the coefficient attached to the PLand variable in equation (3.22), which we can think of as the slope term with respect to that price variable, multiplied by PLand and divided by the quantity of air travel demanded. We can write this as: Cross-Mode Elasticity of Air Travel Demand =

ஒమ ୔ై౗౤ౚ ୔୅ଡ଼

=

ஒమ ୔ై౗౤ౚ ஒబ ାஒభ ୔ఽ౟౨ ାஒమ ୔ై౗౤ౚ

(3.23)

The cross-mode market elasticity of air travel demand is expected to be positive because an increase in the price of land travel leads to an increase in air travel demand or E2 is positive. Like the airfare elasticity of air travel demand, the crossmode elasticity can be elastic, which means it is greater than 1, or it can be inelastic, which means it is less than 1 but greater than zero. When it is elastic, this means that air travel demand is sensitive to a change in the price of land travel. This will typically be the case for city pairs where the distance travelled is relatively short, in which case land transport is a better substitute for air transport. Land transport is a better substitute for air transport for short distances because land transport involves a much longer travel time than air transport. City pairs that involve relatively short distances include Paris to London, Los Angeles to Las Vegas, Sydney to Canberra, Singapore to Kuala Lumpur and Auckland to Wellington. Air transport on these city pairs will face strong competition from land transport. The longer is the journey that a passenger is required to take the greater is the opportunity cost, or implicit cost, to the passenger. Passengers do not like long journey times because the passenger could be doing something more productive with that time. For city pairs that involve relatively long distances over land, such as Sydney to Perth, New York to Las Vegas, Paris to Moscow, and Guangzhou to Ulan Bator in Mongolia, land transport is a relatively weak substitute for air transport because the opportunity cost associated with land transport is too high.

3.7.5 Cross-Destination Market Elasticities of Air Travel Demand The cross-destination market elasticity of air travel demand describes the extent to which passengers will substitute from one destination, A to another destination, B, both travelling by air, because of a change in the airfare to destination A relative to the airfare to destination B. The more alike are the characteristics of destinations A and B the greater the absolute value of the cross-destination market elasticity of air travel demand. For example, suppose that a passenger who is a resident of Perth is considering flying from Perth to Bali. The passenger notices that the airfare to Bali has increased. Rather than fly from Perth to Bali the passenger is considering a new destination, Phuket, which has similar sun, leisure, and cocktail pleasures as Bali. The passenger is considering Phuket over Bali because the price of a holiday to Bali has increased relative to the price of a holiday to Phuket. Consider another example. Suppose a resident from London is considering a holiday to Ibiza. The resident has learnt that airfares to Ibiza on Ryanair have increased while those to the Greek Islands have decreased. The resident of London decides to holiday in the Greek Islands instead of Ibiza, resulting in a reduction in the market demand for London to Ibiza services and an increase in the market demand for London to Greek Island services. Let us now consider the cross-destination elasticity more technically. Let us suppose that the market demand for air travel from origin A to destination B is described by the following linear demand function: PAXA-B = a0 + a1PA-B + a2PA-C

(3.24)

where PA-B is the airfare paid by the passenger for travel between A and B and PA-C is the airfare paid by the passenger for air travel between A and substitute destination C. We have drawn this demand curve in Figure 3-12 below.

Passenger Demand

57

PAXA-B a0 + a2Pଵ୅ିେ

a0 + a2P଴୅ିେ

PAXA-B,1

F

PAXA-B,0

E

DA-B,0 0

PA-B

DA-B,1 PA-B

Fig. 3-12: Demand Function for Travel from Origin A to Destination B The cross-destination market elasticity of demand measures the impact on the demand for travel between A and B caused by a change in the airfare between A and C. Using equation (3.24) the cross-destination elasticity, or the elasticity of AB demand with respect to the A to C airfare PA-C is measured in the following way: ο୔୅ଡ଼ఽషా ο୔ఽషి

×

୔ఽషి ୔୅ଡ଼ఽషా

=

ୟమ ×୔ఽషి ୔୅ଡ଼ఽషా

=

ୟమ ×୔ఽషి ୟబ ାୟభ ୔ఽషా ାୟమ ୔ఽషి

(3.25)

As the coefficient a2 is positive, then the elasticity is positive meaning that an increase in the average airfare between A and C leads to an increase in the demand for air travel between A and B. We can also describe the cross-destination elasticity using the details presented in Figure 3-12. An increase in the airfare associated with travel between A and C, equal to ܲ଴஺ି஼ to ܲଵ஺ି஼ , results in a shift in the demand curve for travel between A and B to the right from DA-B,0 to DA-B,1. For a given A-B airfare PA-B, this results in an increase in the number of passengers carried by air on A-B by PAXA-B,0 to PAXA-B,1 or distance FE. The elasticity of demand for air travel between A-B with respect to a change in the airfare PA-C is therefore: Pଵ୅ିେ FE × Pଵ୅ିେ െ P଴୅ିେ P ୅ି୆ F The elasticity of A-B demand with respect to a change in the airfare of A-C is therefore the distance FE divided by the change in the airfare on route A-C, multiplied by the new airfare on A-C divided by the new demand for A-B, which is distance ܲ ஺ି஻ F. Let us now present a numerical example of the cross-destination air elasticity. The demand function for air travel between London (LON) and Ibiza (IBZ) is: DLON-IBZ = 4,250,000 - 30,000uPLON-IBZ + 10,000uPLON-JMK where PLON-IBZ is the average airfare between London and Ibiza, and PLON-JMK is the average airfare between London and Mykonos in the Greek Islands. We can see in this demand function that there is a positive coefficient attached to the PLON-JMK variable and a negative coefficient attached to the PLON-IBZ variable. The positive coefficient reflects the fact that the Greek Island destination Mykonos is a substitute for the Spanish destination Ibiza. The elasticity of air travel between London and Ibiza to a change in the average airfare for travel between London and Mykonos is: ElasticityLON-IBZ,LON-JMK =

ଵ଴,଴଴଴×୔ైోొషె౉ే ସ,ଶହ଴,଴଴଴ିଷ଴,଴଴଴௉ಽೀಿష಺ಳೋ ାଵ଴,଴଴଴௉ಽೀಿష಻ಾ಼

Chapter 3

58

Let us now evaluate the elasticity at values for the LON-JMK and LON-IBZ airfares. If the average airfare on LON-IBZ services is £90 and the average airfare on LON-JMK is £70 then the elasticity of LON-IBZ air travel with respect to air fares on LON-JMK is: ElasticityLON-IBZ,LON-JMK =

ଵ଴,଴଴଴×଻଴ ସ,ଶହ଴,଴଴଴ିଷ଴,଴଴଴×ଽ଴ାଵ଴,଴଴଴×଻଴

= 0.31

This numerical estimate means that a 1% increase in average airfares for travel between London and Mykonos will result in a 0.31% increase in air travel between London and Ibiza.

3.7.6 Complement Market Elasticity of Air Travel Demand In section 3.3 we analysed the goods and services that are complements to air travel. A feature of these goods and services is that an increase in their price will lead to a reduction in air travel demand because it coincides with the market air travel demand curve shifting in and to the left. Let us demonstrate how this works. Suppose that the demand for air travel is denoted DemandA and this is influenced by the price of air travel, PA and the price of all complements to air travel, which we will call PC. The demand for air travel is assumed to be influenced by the total cost of travel, not just the total cost of air travel. The total cost of travel is equal to: PT = PA + PC In the case of a linear air travel demand function for a particular market, we can write this demand function as a function of the total cost of travel as follows: DemandA = a0 + a1u(PA + PC) For this air travel demand function, a change in the average airfare has the same impact on air travel demand as a change in the price of the complements through the slope parameter a1. The elasticity of air travel demand with respect to a change in the average airfare using this demand function is thus: Elasticity୅,୅ =

ୗ୪୭୮ୣ ୭୤ ୈୣ୫ୟ୬ୢ ୤୳୬ୡ୲୧୭୬×୔ఽ ୈୣ୫ୟ୬ୢఽ

=

ୟభ ×୔ఽ ୟబ ାୟభ ×(୔ఽ ା୔ి )

The elasticity of air travel demand with respect to a change in the price of the complements is: Elasticity୅,େ =

ୗ୪୭୮ୣ ୭୤ ୈୣ୫ୟ୬ୢ ୤୳୬ୡ୲୧୭୬×୔ి ୈୣ୫ୟ୬ୢఽ

=

ୟభ ×୔ి ୟబ ାୟభ ×(୔ఽ ା୔ి )

The elasticity of air travel demand with respect to the total price of travel is: Elasticity୅ =

ୗ୪୭୮ୣ ୭୤ ୈୣ୫ୟ୬ୢ ୤୳୬ୡ୲୧୭୬×(୔ఽ ା୔ి ) ୈୣ୫ୟ୬ୢఽ

=

ୟభ ×(୔ఽ ା୔ి ) ୟబ ାୟభ ×(୔ఽ ା୔ి )

We can see that the three elasticities depend heavily on the level of the price. The demand for air travel is more elastic to the total price than to the airfare alone because the airfare is a fraction of the total travel price. The lower is the airfare as a proportion of the total travel price, that is, the lower is PA/(PA + PC) the less elastic is air travel demand to the airfare compared to the elasticity of air travel demand to the total travel price. In fact, it is conceivable that air travel demand is elastic with respect to the total travel price but inelastic with respect to the average airfare. Let us look at an example of this in the numerical illustration below. There were 4.821m passenger movements between Sydney and Brisbane (BNE) by air in calendar 2019. The average return airfare paid by these passengers was $300. The average passenger paid $700 for other travel expenses, including accommodation, food and beverages, land transport and entertainment. The demand function for air travel between Sydney and Brisbane is explained by the following linear function (which is defined in millions): DemandSYD-BNE = 10.847 – 0.00603 u (P୅ୗଢ଼ୈି୆୒୉ + Pୌଢ଼ୈି୆୒୉ ) where ܲ஺ௌ௒஽ି஻ோ is the airfare on Sydney-Brisbane and ܲ஼ௌ௒஽ି஻ோ is the price paid for all other goods and services consumed with travel between Sydney and Brisbane. The elasticities of air travel demand with respect to the airfare, the price of other goods and services and the total travel price are respectively: Elasticityୗଢ଼ୈି୆୒୉,୅ = Elasticityୗଢ଼ୈି୆୒୉,େ =

ୗ୪୭୮ୣ ୭୤ ୈୣ୫ୟ୬ୢ ୤୳୬ୡ୲୧୭୬×୔౏ౕీషాొు ఽ ୈୣ୫ୟ୬ୢఽ ୗ୪୭୮ୣ ୭୤ ୈୣ୫ୟ୬ୢ ୤୳୬ୡ୲୧୭୬×୔౏ౕీషాొు ి ୈୣ୫ୟ୬ୢఽ

= =

ି଴.଴଴଺଴ଷ×୔౏ౕీషాొు ఽ

ଵ଴.଼ସ଻ି଴.଴଴଺଴ଷ×൫୔౏ౕీషాొు ା୔౏ౕీషాొు ൯ ఽ ి ି଴.଴଴଺଴ଷ×୔౏ౕీషాొు ి

ଵ଴.଼ସ଻ି଴.଴଴଺଴ଷ×൫୔౏ౕీషాొు ା୔౏ౕీషాొు ൯ ఽ ి

Passenger Demand

Elasticityୗଢ଼ୈି୆୒୉ =

ୗ୪୭୮ୣ ୭୤ ୈୣ୫ୟ୬ୢ ୤୳୬ୡ୲୧୭୬×ቀ୔౏ౕీషాొు ା୔౏ౕీషాొు ቁ ఽ ి ୈୣ୫ୟ୬ୢఽ

59

=

ି଴.଴଴଺଴ଷ×ቀ୔౏ౕీషాొు ା୔౏ౕీషాొు ቁ ఽ ి ଵ଴.଼ସ଻ି଴.଴଴଺଴ଷ×൫୔౏ౕీషాొు ା୔౏ౕీషాొు ൯ ఽ ి

If we substitute the price levels ܲ஺ௌ௒஽ି஻ோ = $300 and ܲ஼ௌ௒஽ି஻ோ = $700 into these expressions, we obtain: ି଴.଴଴଺଴ଷ×ଷ଴଴

Elasticityୗଢ଼ୈି୆୒୉,୅ =

ଵ଴.଼ସ଻ି଴.଴଴଺଴ଷ×ଵ଴଴଴

Elasticityୗଢ଼ୈି୆୒୉,େ =

ଵ଴.଼ସ଻ି଴.଴଴଺଴ଷ×ଵ଴଴଴

Elasticityୗଢ଼ୈି୆୒୉ =

ି଴.଴଴଺଴ଷ×଻଴଴

ି଴.଴଴଺଴ଷ×ଵ଴଴଴ ଵ଴.଼ସ଻ି଴.଴଴଺଴ଷ×ଵ଴଴଴

= െ0.406 = െ0.947

= െ1.353

These elasticity estimates are interpreted as follows: x a 1% increase in the average airfare leads to a 0.41% decrease in air travel demand for services between Sydney and Brisbane, holding the price of complements fixed; x a 1% increase in the price of non-airfare travel expenses leads to a 0.95% decrease in air travel demand for services between Sydney and Brisbane, holding the average airfare fixed; and x a 1% increase in the total travel price leads to a 1.35% decrease in air travel demand for services between Sydney and Brisbane. These elasticity estimates tell us that the demand for air travel between Sydney and Brisbane is inelastic to the average airfare, inelastic to the price of non-airfare goods and services, but elastic to the total travel price. The inelasticity of air travel demand for Sydney-Brisbane services to the average airfare reflects the fact that the average airfare is only a fraction, 30%, of the total travel spend.

3.7.7 Airline-Specific Fare Elasticities of Demand In section 3.5 we presented airline specific demand curves at equations (3.1) and (3.2). In this section, we examine how we find the airline specific airfare elasticities of demand from these demand functions. There are four airfare elasticities of demand that we can estimate from the airline specific demand functions A and B at equations (3.1) and (3.2). These elasticities are: x the elasticity of demand for airline A’s services with respect to airline A’s price – this is called the own airfare elasticity of demand for airline A; x the elasticity of demand for airline A’s services with respect to airline B’s price – we call this the cross-price elasticity of demand for airline A; x the elasticity of demand for airline B’s services with respect to airline B’s price – this is the own airfare elasticity of demand for airline B; and x the elasticity of demand for airline B’s services with respect to airline A’s price – we call this the cross-price elasticity of airline B. The first two elasticities are obtained from equation (3.1) in the following way: ElasticityAA =

ୗ୪୭୮ୣ ୭୤ ୅ ୈୣ୫ୟ୬ୢ ୲୭ ୅ ୔୰୧ୡୣ×୔ఽ

ElasticityAB =

ୗ୪୭୮ୣ ୭୤ ୅ ୈୣ୫ୟ୬ୢ ୲୭ ୆ ୔୰୧ୡୣ×୔ా

ୈୣ୫ୟ୬ୢఽ

ୈୣ୫ୟ୬ୢఽ

= =

ିୟభ ×୔ఽ ୈୣ୫ୟ୬ୢఽ ୟమ ×୔ా ୈୣ୫ୟ୬ୢఽ

= =

ିୟభ ×୔ఽ ୟబ ିୟభ ୔ఽ ାୟమ ୔ా ୟమ ×୔ా ୟబ ିୟభ ୔ఽ ାୟమ ୔ా

The elasticity term ElasticityAA is the own-price elasticity of airline A’s demand and is negative in sign, indicating that an increase in the average airfare of airline A causes a reduction in airline A demand. The elasticity term ElasticityAB is the cross-price elasticity of airline A’s demand and is positive in sign, indicating that an increase in the average airfare of airline B causes an increase in airline A’s demand. The remaining two elasticities are obtained from equation (3.2) using similar methods as those obtained from equation (3.1): ElasticityBB =

ୗ୪୭୮ୣ ୭୤ ୆ ୈୣ୫ୟ୬ୢ ୲୭ ୆ ୔୰୧ୡୣ×୔ా

ElasticityBA =

ୗ୪୭୮ୣ ୭୤ ୆ ୈୣ୫ୟ୬ୢ ୲୭ ୅ ୔୰୧ୡୣ×୔ఽ

ୈୣ୫ୟ୬ୢా

ୈୣ୫ୟ୬ୢా

= =

ିୠభ ×୔ా ୈୣ୫ୟ୬ୢా ୠమ ×୔ఽ ୈୣ୫ୟ୬ୢా

= =

ିୠభ ×୔ా ୠబ ିୠభ ୔ా ାୠమ ୔ఽ ୠమ ×୔ఽ ୠబ ିୠభ ୔ా ାୠమ ୔ఽ

Let us now examine how these elasticity expressions are estimated by considering a numerical example, which is based on the numerical airline demand functions presented in section 3.5 in the case of Cebu Pacific services between Manila and Jakarta. Using the Cebu Pacific demand function described by equation (3.9) the airfare elasticity of demand for Cebu Pacific services on Manila to Jakarta with respect to the average airfare charged by Cebu Pacific on the route is:

Chapter 3

60

ElasticityCebu, Cebu =

ି଴.଴଴ହ×୔౉ొైషిృే ి౛ౘ౫

ଵ଼଴ି଴.଴଴ହ୔౉ొైషిృే ା଴.଴଴ଶହ୔౉ొైషిృే ౌఽ ి౛ౘ౫

The elasticity of demand for Cebu Pacific services on Manila to Jakarta with respect to the average airfare of Philippine Airlines is: ElasticityCebu, PA =

଴.଴଴ଶହ×୔౉ొైషిృే ౌఽ

ଵ଼଴ି଴.଴଴ହ୔౉ొైషిృే ା଴.଴଴ଶହ୔౉ొైషిృే ౌఽ ి౛ౘ౫

Let us now evaluate these elasticities when the Cebu Pacific airfare is 30,000 Philippine Pesos and the Philippine Airlines average airfare on Manila-Jakarta is 32,000 Philippine Pesos. Substituting these two airfares into our elasticity expressions and we obtain the following own and cross-price elasticities: ElasticityCebu, Cebu = ElasticityCebu, PA =

ି଴.଴଴ହ×ଷ଴,଴଴଴ ଵ଼଴ି଴.଴଴ହ×ଷ଴,଴଴଴ା଴.଴଴ଶହ×ଷଶ,଴଴଴ ଴.଴଴ଶହ×ଷଶ,଴଴଴

ଵ଼଴ି଴.଴଴ହ×ଷ଴,଴଴଴ା଴.଴଴ଶହ×ଷଶ,଴଴଴

= െ1.36364

= 0.727273

These elasticities indicate that a 1% increase in the Cebu Pacific average airfare on the Manila to Jakarta route leads to a 1.36% decrease in the demand for Cebu Pacific services on the route, holding the airfare of Philippine Airways constant. A 1% increase in the Philippine Airlines average airfare on the Manila to Jakarta route leads to a 0.73% increase in the demand for Cebu Pacific services on the route, holding the airfare of Cebu Pacific constant.

Quiz 3-3 Airfare Elasticity of Air Travel Demand 1. (a) (b) (c) (d)

Market demand in the Australian domestic air travel market is described by the following function DemandAUSAirfare , where DemandAUS-DOM is the number of domestic one-way passengers in millions and DOM = 97 - 0.18P PAirfare is the average one-way airfare. Use this information to answer the following questions. Draw the demand curve in a graph with PAXAUS-DOM on the vertical axis and PAirfare on the horizontal axis. Identify the points on the demand curve at an average airfare of $500 and an average airfare of $150. Evaluate the market airfare elasticity of domestic Australian air travel demand at the price points in (b). Interpret the elasticities. Now suppose that the Coronavirus has an impact on the demand for domestic air travel in Australia, resulting in a new demand function that takes the following form DemandAUS-DOM = 45 - 0.18PAirfare. Estimate the market airfare elasticity of Australian domestic air travel demand using this new demand function at an average airfare of $150. Explain your results.

2. (a) (b) (c) (d)

Indicate whether the following forces make air travel demand more, or less elastic to the average airfare. An increase in household income. An increase in the population. An increase in the price of accommodation. A reduction in the cost of travelling by car.

3.

The demand for air travel between Sydney and Canberra (CBR) is described by the following function DemandAIR + 1,000PCAR, where DemandSYD-CBR is measured in terms of the numbers of SYD-CBR = 1,365,500 - 3,879P AIR passengers, P is the average airfare on the SYD-CBR route and PCAR is the average cost of using a car to drive between Sydney and Canberra. Canberra is the capital of the Australian Capital Territory in Australia and is located around 270km South-West of Sydney. Use this information to answer the following questions. Draw the demand curve in a graph with PAXSYD-CBR on the vertical axis and PAIR on the horizontal axis. Identify in the graph how PCAR affects the intercept terms. Identify the points on the demand curve at an average airfare of $150 and an average airfare of $75. Assume PCAR = $50. Evaluate the airfare elasticity of air travel demand at the price points in (b). Interpret the elasticity. Evaluate the cross-mode elasticity of air travel demand at PAIR = $120 and PCAR = $50. Interpret your answer. What would happen to both the airfare elasticity and the cross-mode elasticity if the demand curve shifts in and to the left? Explain your answer.

(a) (b) (c) (d) (e) 4.

The demand for air travel between Perth and Bali is described by the following linear demand function DemandPER-DPS = 1,900,000 - 2,400PPER-DPS + 400PPER-HKT, where DemandPER-DPS is the number of passengers travelling by air between Perth and Bali (Denpasar or DPS), PPER-DPS is the average airfare for Perth services to Bali and PPER-HKT is the average airfare for Perth services to the substitute destination Phuket (where HKT is the airport code for Phuket Airport). Use this information to answer the following questions. (a) Interpret the coefficients associated with PPER-DPS and PPER-HKT in the PER-DPS demand function. (b) Estimated the level of demand when PPER-DPS = $500 and PPER-HKT = $700.

Passenger Demand

61

(c) Determine the elasticity of PER-DPS demand with respect to the own airfare and the substitute destination airfare. Find these elasticities at the prices in (b). Interpret your results. (d) Assuming PPER-HKT = $700 at what level of PPER-DPS is the elasticity of PER-DPS demand to PPER-DPS equal to 1? 5. (a) (b) (c) (d)

We now re-specify the market demand function for air travel in the domestic Australian market from question 1 in this Quiz to include the price of complements to air travel as follows DemandAUS-DOM = 241 - 0.18(PA + PC). Use this demand function to answer the following questions. What types of prices would be included in PC? Draw the demand curve in a graph with PAXAUS-DOM on the vertical axis and PA on the horizontal axis ensuring you make clear how PC is a part of the vertical and horizonal intercepts. Use your demand curve to demonstrate what happens if there is a decrease in the price of complements. Identify the points on the demand curve at an average airfare of $150 and an average airfare of $300 assuming PC = $800. Evaluate the elasticity of domestic Australian air travel with respect to the average airfare, the price of complements and the total travel price at prices of PA = $200 and PC = $800. Interpret your answers.

6.

Return to the airline specific demand functions from Quiz 3-2 question 2, repeated below for your convenience, ୉୞୉ିୗେ୐ ୉୞୉ିୗେ୐ ୉୞୉ିୗେ୐ Demand୉୞୉ିୗେ୐ and Demand୉୞୉ିୗେ୐ = 190 െ 0.4Pୗ୏ଢ଼ + ୐୅୘୅୑ = 300 െ 0.3P୐୅୘୅୑ + 0.15Pୗ୏ଢ଼ ୗ୏ଢ଼ ୉୞୉ିୗେ୐ 0.2P୐୅୘୅୑ . (a) Find all potential airline specific airfare elasticities of demand. ா௓ாିௌ஼௅ ா௓ாିௌ஼௅ = $225 and ܲௌ௄௒ = $200. Interpret your results. (b) Find numerical values for the elasticities at ܲ௅஺்஺ெ 7.

Describe the elasticity which is relevant to the scenarios (a) to (k) given below. Choose from the following elasticity choices: Elasticity Choices i. the own-price elasticity of airline specific air travel demand; ii. the cross-price elasticity of airline specific air travel demand; iii. the market airfare elasticity of air travel demand; iv. the cross-mode elasticity of air travel demand; v. the cross-destination airfare elasticity of air travel demand; vi. the elasticity of air travel demand to the price of complements; and vii. the elasticity of air travel demand to the total travel price. Questions (a) Jetstar increases its average airfare, and this results in a reduction in Jetstar demand. (b) Air Asia increases its average airfare on the route Singapore (SIN) to Denpasar (DPS) and this leads to a reduction in Scoot demand on this route. (c) Airfares increase on Copenhagen (CPH) to Frankfurt (FRA) which leads to a reduction in the market demand for CPH-FRA services. (d) Accommodation prices increase in Perth leading to a reduction in air travel demand between Singapore and Perth. (e) Average airfares on Jetstar services between Sydney (SYD) and Honolulu (HNL) increase, resulting in an increase in demand for air travel between SYD and Fiji (NAN). (f) The cost of a holiday to the Greek Islands from Sweden has increased, resulting in more Swedes travelling by air to the Canary Islands. (g) A resident of Paris wishes to travel to London by air on an Air France flight but has decided to take the Eurostar high speed train because the Eurostar high speed train has reduced its price. (h) The Australian dollar depreciates against the New Zealand dollar resulting in more Australians taking skiing holidays in Victoria and NSW rather than travelling to Queenstown. (i) The Australian dollar depreciates against the US dollar but remains stable against the Japanese Yen, resulting in Australians switching their trip from the U.S. Disneyland to a trip to the Disneyland in Japan. (j) An Australian travelling to Europe flies from Sydney to London Heathrow on Qantas and is considering flying from London Heathrow to Berlin on British Airways. British Airways, however, has increased its airfares on Berlin flights from London Heathrow. The Australian traveller decides to fly to Hamburg instead of Berlin on British Airways from London Heathrow where the airfares have not changed. (k) A Japanese resident of Tokyo who wishes to travel to Osaka decides to switch from travel on a Peach Aviation flight to travel on the Nozomi fast train because Peach Aviation has increased its average airfare.

Chapter 3

62

3.8 Demand Over Time We often wish to understand why the demand for air travel changes over time. To do this we need to understand the types of movements over time that can occur. There are several different types of movements in the demand for air travel over time that we will explore in this section, including trend movements in demand, cyclical movements in demand, seasonal movements in demand, and structural movements in demand.14

3.8.1 Drivers of the Trend The trend in air travel demand is the long-term direction of demand. To illustrate the intuitive meaning of the trend in demand, we show a graph (Figure 3-13) of the passengers carried by Emirates Airline in every year since 1990.15 Emirates PAX ('000) 70,005 60,005

58,600

50,005 40,005 30,005 20,005 10,005 5

765 2019 2018 2017 2016 2015 2014 2013 2012 2011 2010 2009 2008 2007 2006 2005 2004 2003 2002 2001 2000 1999 1998 1997 1996 1995 1994 1993 1992 1991 1990

Source: Airline Intelligence and Research Database 2021

Fig. 3-13: Emirates Airlines Passengers Carried The trend in the number of passengers carried by Emirates is its long-term direction. The long-term direction can be established by answering the following simple question: Where is the line graph headed over a long horizon? The answer to this question, at least intuitively, is guided by the direction of the dashed line in Figure 3-13, which is determined in Microsoft Excel by using the “Trendline” feature of the program. What drives the trend in air passenger demand? The trend movement is driven by variables that also trend over time. These include such variables as the population, an economy’s income or production, and consumer wealth. Population and income tend to follow a relatively stable upward trend over time, while wealth follows an upward trend but there is more volatility associated with this upward trend.

3.8.2 Cycle The cycle of a variable represents the stable, predictable patterns of movement around the trend in the variable. To see what we mean by the cyclical movements over time in a variable, consider the RPKs of Rex Express Airlines, which are presented at a half yearly frequency between the 6 months to June 30, 2005 and the 6 months to December 31, 2018 in Figure 3-14 below. Regional Express Airlines is a small airline that predominantly operates intrastate air travel services across several Australian states. We can see in Figure 3-14 that Regional Express half-yearly RPKs cycle above and below a mean level of 224m. Regional Express RPKs cycle rather than exhibit a trend because the movement over time always returns to the fixed mean of 224m. This is unlike a trending variable which increases or decreases over time without returning to a fixed average value. Cyclical movements in a variable are often caused by cyclical movements in economic activity, the growth in the population, and movements in variables that move in cycles, such as exchange rates, interest rates, asset returns, consumer confidence and business confidence.

14

There is an excellent summary of basic time series concepts provided for free by the Australian Bureau of Statistics Time Series 2019. 15 The Emirates financial year starts on 1 April and ends on 31 March. The year 2019 on the horizontal axis in Figure 3-13 is the date 1 April 2018 through to 31 March 2019.

Passenger Demand

63

Regional Express RPK (m) 295 275 255 235 224 215 195 175 Dec-18 Jun-18 Dec-17 Jun-17 Dec-16 Jun-16 Dec-15 Jun-15 Dec-14 Jun-14 Dec-13 Jun-13 Dec-12 Jun-12 Dec-11 Jun-11 Dec-10 Jun-10 Dec-09 Jun-09 Dec-08 Jun-08 Dec-07 Jun-07 Dec-06 Jun-06 Dec-05 Jun-05 Source: Airline Intelligence and Research Database 2021

Fig. 3-14: Cycle in Rex Express Airlines RPKs

3.8.3 Seasonality Seasonality represents the peaks and troughs in data that is less than annual frequency. Frequency that is less than annual may include data that is daily, weekly, monthly, quarterly, or half-yearly. Seasonality could be attributable to many forces including but not limited to the climate (summer, winter, spring, and autumn), the weather (sunny, rainy, cloudy, windy), the timing of school holidays, the timing of public holidays, festive seasons (Christmas, Easter, Chinese New Year, Mecca pilgrimage), and major events. For example, Figure 3-15 below presents the seasonality in the seat factor for the total network of Vueling Airlines, which is a low-cost carrier domiciled in Spain and a part of the International Airlines Group. Vueling Seat Factor

95% 90% 85% 80% 75% 70% 65% 60% 55% 50%

May-17

Jan-17

Sep-16

May-16

Jan-16

Sep-15

May-15

Jan-15

Sep-14

May-14

Jan-14

Sep-13

May-13

Jan-13

Sep-12

May-12

Fig. 3-15: Vueling Airlines Total System Seat Factor

Jan-12

Sep-11

May-11

Jan-11

Sep-10

May-10

Jan-10

Sep-09

May-09

Jan-09

Sep-08

May-08

Jan-08

Sep-07

May-07

Source: Airline Intelligence and Research Database 2021

Chapter 3

64

The seasonal peaks and troughs are described respectively by the upper and lower circles. The upper circles are the August peak months, and the lower circles represent the trough month of January. August is likely to be the peak because it is the strongest travel month in the European summer. January is the trough because it is the middle of winter and represents the post New Year lull in air travel.

3.8.4 Structural Events The airline industry is subject to many different types of favourable and unfavourable shocks or structural events. The following are examples of some of the major events that have occurred since 1987 (1) the Coronavirus and the SARS virus (World Health Organisation 2019), (2) the terrorism events of September 11 (Encyclopedia Britannica 2019a), (3) The Bali bombings in October 2002 and 2005 (Encyclopedia Britannica 2019b), (4) Fiji coup d’états in 1987, 2000, 2006 (Tran 2006), (5) Pilot’s strike in the late 1980’s in Australia (Kelly, Metherell and McAsey 2020), (6) Major airline collapse such as Ansett and Air Berlin (Singh 2019 and Whyte and Sumers 2017), (7) Floods such as the Kerala floods in India 2018, (Kerala Floods 2018), (8) Tsunami such as the Indonesia, Thailand and Sri-Lanka Tsunamis in 2004, (Roos 2020), (9) Olympic Games, and (10) Cricket and soccer world cups. As indicated in Figure 3-16 below, the typical impact of adverse shocks traverses through many phases. The initial phase is a relatively weak phase which is referred to as the news phase, where information about the shock is still being processed. The next impact is referred to as a pre-peak shoulder impact, where the impact gathers momentum but it is not yet at its peak. The next phase is the peak, where the impact on demand is at its greatest. The impact then travels through a post-peak shoulder phase, which is followed by a phase that involves servicing pent-up demand. Reduction in Tourism

Pent up Demand Returns News Phase Adverse Impact complete 0 Time Start of Shock

Shoulder Impact Peak Shock Impact

Shoulder Impact

Fig. 3-16: A Typical Timeline of Adverse Shocks In Figure 3-17, we present an example of the impact of the SARs virus on the number of people travelling to Australia from Asia and reconcile this with the phases presented in Figure 3-16. The news phase began in early 2003, at which point demand began to drop. Demand reached a low point in April and May 2003, which we define as the peak adverse impact of the SARS virus. Demand then begins to rise, although remains weak, through the post-peak shoulder. Demand then returns to normal and in fact overshoots normal as pent-up demand is met, that is, passengers who postponed their trip during the outbreak of the virus decide to continue their journey once the virus outbreak has been declared over.

Passenger Demand

65

Inbound Tourists from Asia to Australasia

450,000

Pent-Up Demand

430,000 410,000

News

390,000 370,000

Pre-peak Shoulder

Post-peak Shoulder

350,000 330,000

Peak Jan-2004 Dec-2003 Nov-2003 Oct-2003 Sep-2003 Aug-2003 Jul-2003 Jun-2003 May-2003 Apr-2003 Mar-2003 Feb-2003 Jan-2003 Dec-2002 Nov-2002 Oct-2002 Sep-2002 Aug-2002 Jul-2002 Jun-2002 May-2002 Apr-2002 Mar-2002 Feb-2002 Jan-2002

Source: Australian Bureau of Statistics, catalogue number 3401.0, Table 3. Accessed October 26, 2021.

Fig. 3-17: Impact of the SARS Virus for Australian Inbound Tourism from Asia

3.9 Demand Across City Pairs and the Gravity Model In this section we attempt to understand why passenger demand differs across city pairs. For example, why does the demand for Johannesburg to Cape Town differ from the demand for Bangalore to Mumbai, or Cusco to Lima, or Mexico City to Monterrey? The aviation website ‘routes online’ presents the passenger numbers for the world’s biggest city pairs, which can be used to provide insight into the forces that cause passenger demand to differ across routes. Table 32 below reproduces a table that is constructed by routes on-line. The busiest city pair by passengers carried in world aviation in 2018 is the South Korean city pair Seoul to Jeju. Seoul is the capital of South Korea with a population of just under 10m (World Population Review Seoul 2021) and household income per capita of US$17,408 (CEIC Data Seoul 2021). Jeju is an island that is 450km from Seoul, located in the Korea Strait. Jeju has a population of around 700,000 (Population State Jeju 2021) and has an average high temperature that is above 20 degrees for 6 months of the year (Climates to Travel Jeju 2021). The city pair Seoul-Jeju is therefore a predominantly leisure-purpose city pair, the demand for which will be driven by residents of Seoul who wish a short getaway leisure holiday to an island. The city pair is currently serviced by a variety of carriers who fly over 8 services per day. The carriers include Air Seoul, Asiana, Jeju Air, Korean Air, Jin Air and T’Way Air. The major drivers of the passenger traffic on this city pair are the size of the population of Seoul and the high income per capita of the residents of Seoul. Combined with the short distance that aircraft must travel on this route and therefore the short journey time, along with the large number of carriers operating services on the route, we have the essential ingredients for high passenger numbers. It is also the case that competition faced by airlines on this route from ferry transport is unlikely to be significant because the ferry takes 4 and a half hours compared to a little over 1 hour by air. The second busiest city pair in the world is the Japanese domestic city pair Sapporo (CTS) to Haneda (HND), Tokyo. Sapporo is the fifth most populous city in Japan with a population of 2.7m (Macrotrends Sapporo 2021). Haneda is one of the major airports in Tokyo, along with Narita Airport. Greater Tokyo has a population of 37m (World Population Review Tokyo 2021) with GDP per capita of around US$48,649 in 2014 (Liu 2019). Air travel between the two cities takes one and a half hours while train travel takes around 8 hours, therefore mode of transport competition is relatively weak. The third busiest city pair in world aviation is Melbourne-Sydney. Both Sydney and Melbourne have populations of over 5 million (Regional Populations Australian Bureau of Statistics 2021), with average incomes per capita of over US$50,000 (Australian Bureau of Statistics Census 2021), and the two cities are just 705 kilometres apart. Sydney and Melbourne are both major population and business centres in Australia, thus generating a high percentage of businesspurpose traffic. Given the distance between the cities, car and rail travel takes around 10 hours and there is no possibility of sea travel. This implies weak mode of transport competition. There are many Sydney residents born in Melbourne and Melbourne residents born in Sydney, implying significant visiting friends and relative traffic between the two cities. It is also the case that Sydney Airport operates more international services than Melbourne Airport, which mean the Melbourne-Sydney service is likely to support domestic traffic that wishes to connect to international services.

Chapter 3

66 Rank

City Pair

PAX (2018)

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58

Seoul (GMP) - Jeju (CJU) Sapporo (CTS) - Tokyo (HND) Melbourne (MEL) - Sydney (SYD) Tokyo (HND) - Fukuoka (FUK) Mumbai (BOM) - Delhi (DEL) Ho Chi Minh City (SGN) - Hanoi (HAN) Beijing (PEK) - Shanghai (SHA) Taipei (TPE) - Hong Kong (HKG) Okinawa (OKA) - Tokyo (HND) Surabaya (SUB) - Jakarta (CGK) Denpasar Bali (DPS) - Jakarta (CGK) Jeddah (JED) - Riyadh (RUH) Osaka (ITM) - Tokyo (HND) Beijing (PEK) - Chengdu (CTU) Guangzhou (CAN) - Beijing (PEK) Cancun (CUN) - Mexico City (MEX) Beijing (PEK) - Shenzhen (SZX) Brisbane (BNE) - Sydney (SYD) Jakarta (CGK) - Singapore (SIN) Shanghai (SHA) - Guangzhou (CAN) Shanghai (SHA) - Shenzhen (SZX) Bengaluru (BLR) - Delhi (DEL) Ujung Pandang (UPG) - Jakarta (CGK) Jakarta (CGK) - Medan (KNO) Cape Town (CPT) - Johannesburg (JNB) Kuala Lumpur (KUL) - Singapore (SIN) Rio De Janeiro (SDU) - Sao Paulo (CGH) Shanghai (PVG) - Hong Kong (HKG) Los Angeles (LAX) - San Francisco (SFO) Bogota (BOG) - Medellin (MDE) Mumbai (BOM) - Bengaluru (BLR) Los Angeles (LAX) - New York (JFK) Bangkok (DMK) - Phuket (HKT) Melbourne (MEL) - Brisbane (BNE) Manila (MNL) - Cebu (CEB) Hong Kong (HKG) - Bangkok (BKK) Mexico City (MEX) - Monterrey (MTY) Delhi (DEL) - Kolkata (CCU) Da Nang (DAD) - Ho Chi Minh City (SGN) Bangkok (DMK) - Chiang Mai (CNX) Osaka (KIX) - Seoul (ICN) Lima (LIM) - Cuzco (CUZ) New York (LGA) - Chicago (ORD) Kuala Lumpur (KUL) - Jakarta (CGK) Guadalajara (GDL) - Mexico City (MEX) Jeju (CJU) - Busan (PUS) Izmir (ADB) - Istanbul (IST) Seoul (ICN) - Hong Kong (HKG) Guangzhou (CAN) - Chengdu (CTU) Hong Kong (HKG) - Manila (MNL) Beijing (PEK) - Hong Kong (HKG) Delhi (DEL) - Hyderabad (HYD) Las Vegas (LAS) - Los Angeles (LAX) Johannesburg (JNB) - Durban (DUR) Hong Kong (HKG) - Singapore (SIN) Atlanta (ATL) - Orlando (MCO) Vancouver (YVR) - Toronto (YYZ) Chengdu (CTU) - Shenzhen (SZX)

14,107,414 9,698,639 9,245,392 8,762,547 7,392,155 6,867,114 6,518,997 6,476,268 5,829,712 5,649,046 5,535,108 5,526,110 5,131,757 5,092,442 5,076,229 4,885,602 4,853,038 4,815,609 4,812,342 4,724,514 4,679,294 4,542,638 4,530,428 4,512,830 4,508,214 4,290,463 4,234,631 4,053,909 3,971,922 3,930,332 3,814,494 3,635,655 3,612,373 3,565,266 3,519,525 3,490,988 3,474,971 3,431,999 3,256,718 3,212,124 3,210,813 3,201,414 3,198,700 3,170193 3,170,177 3,106,224 3,084,250 3,081,942 3,044,038 3,008,842 3,000,786 2,980,730 2,961,553 2,953,851 2,923,578 2,918,980 2,912,026 2,873,291

Source: Casey 2019

Table 3-2: Top 100 City Pairs by Passengers Carried

Average base fare (US$) 91 147 195 187 86 104 184 168 307 83 107 164 149 217 180 117 317 175 216 153 217 90 126 143 172 149 88 306 193 90 68 492 49 175 90 223 142 78 67 39 148 91 171 110 117 81 86 238 133 283 450 82 214 129 384 201 385 223

Distance (km) 449 819 705 880 1138 1160 1075 802 1553 690 983 851 402 1558 1880 1284 1957 748 882 1175 1209 1708 1432 1386 1271 296 365 1247 541 216 834 3973 692 1376 566 1690 713 1313 605 568 859 583 1176 1128 459 291 332 2068 1221 1145 1991 1265 380 478 2565 645 3344 1316

Passenger Demand Rank 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100

City Pair Beijing (PEK) - Hangzhou (HGH) Tokyo (NRT) - Taipei (TPE) New York (JFK) - London (LHR) Bogota (BOG) - Cartagena (CTG) Hanoi (HAN) - Da Nang (DAD) Los Angeles (LAX) - Chicago (ORD) Izmir (ADB) - Istanbul (SAW) Kota Kinabalu (BKI) - Kuala Lumpur (KUL) Delhi (DEL) - Pune (PNQ) Delhi (DEL) - Chennai (MAA) Tokyo (HND) - Kagoshima (KOJ) Palembang (PLM) - Jakarta (CGK) Bogota (BOG) - Cali (CLO) Sydney (SYD) - Gold Coast (OOL) Chiang Mai (CNX) - Bangkok (BKK) Dubai (DXB) - London (LHR) Jakarta (CGK) - Semarang (SRG) Beijing (PEK) - Xi An (XIY) Davao (DVO) - Manila (MNL) Bangkok (BKK) - Singapore (SIN) Hangzhou (HGH) - Guangzhou (CAN) Fort Lauderdale (FLL) - Atlanta (ATL) Beijing (PEK) - Chongqing (CKG) Antalya (AYT) - Istanbul (SAW) Jakarta (CGK) - Yogyakarta (JOG) Bangkok (DMK) - Hat Yai (HDY) Jinghong (JHG) - Kunming (KMG) Melbourne (MEL) - Adelaide (ADL) Mumbai (BOM) - Goa (GOI) Kuala Lumpur (KUL) - Penang (PEN) Seattle (SEA) - Los Angeles (LAX) Montreal (YUL) - Toronto (YYZ) Wellington (WLG) - Auckland (AKL) Tokyo (NRT) - Seoul (ICN) Hyderabad (HYD) - Bengaluru (BLR) Jakarta (CGK) - Pontianak (PNK) Madrid (MAD) - Barcelona (BCN) Chongqing (CKG) - Shenzhen (SZX) Mumbai (BOM) - Dubai (DXB) Christchurch (CHC) - Auckland (AKL) Atlanta (ATL) - New York (LGA) Ahmedabad (AMD) - Delhi (DEL)

PAX (2018) 2,856,019 2,848,925 2,846,321 2,825,672 2,810,618 2,810,368 2,799,413 2,790,835 2,787,960 2,777,822 2,768,091 2,746,929 2,722,122 2,722,095 2,694,612 2,665,441 2,650,823 2,639,625 2,630,909 2,597,199 2,594,232 2,572,815 2,554,282 2,545,643 2,517,527 2,513,414 2,502,690 2,493,589 2,492,268 2,489,660 2,485,576 2,485,533 2,481,522 2,466,953 2,466,522 2,463,262 2,448,664 2,428,313 2,394,060 2,385,766 2,378,490 2,363,818

67 Average base fare (US$) 292 270 829 115 67 286 47 123 91 87 187 64 112 120 109 704 71 152 106 203 173 162 224 46 80 96 120 149 66 82 217 262 186 230 29 72 202 169 238 191 221 71

Distance (km) 1127 2181 5539 655 626 2802 343 1622 1156 1759 935 413 278 681 597 5498 422 933 964 1416 1017 937 1463 462 455 776 386 641 423 324 1537 503 480 1257 454 731 483 1062 1925 745 1225 752

Source: Casey 2019

Table 3-2 (cont): Top 100 City Pairs by Passengers Carried These three examples of the biggest city pairs in world aviation tell us much about the potential drivers of the differences in passenger numbers across city pairs. These drivers are the population of the two cities in the city pair, the income per capita of the two cities in the city pair, the distance between the cities, and the number of people living in city A that were born in city B and vice versa. A model that is often used to capture the difference in demand across city pairs using some of these drivers is the gravity model. The gravity model was built originally to understand why trade that country A has with country B differs from trade between country C and country D. The model was built by Walter Isard in the 1950’s (Isard 1954). In its original form the model was specified in the following way: ಊ

Fij =

ಊ

୑౟ భ ୑ౠ మ ಊ

ୈ౟ౠయ

(3.26)

This rather complicated expression says the following. The term Fij is the trade that takes place between country i and country j. Mi is some measure of activity of country i, such as the GDP of country i or the population of country i, and Mj is the GDP or population of country j. The term in the denominator Dij is the distance between countries i and j. The term E1 is the elasticity of freight volumes with respect to the GDP or population of country i. The term E2 is the elasticity

Chapter 3

68

of freight volumes with respect to the GDP or population of country j. The term E3 is the elasticity of freight volumes with respect to the distance between countries i and j. A similar model to (3.26) can be used to explain why there is a difference in the number of passengers across city pairs at a given point in time. Consider the following gravity model for passengers by city pair: ಊ

PAXij =

ಊ

୔୓୔౟ భ ୔୓୔ౠ మ

(3.27)

ಊ

ୈ౟ౠయ

This simple gravity model says that the number of passengers that travel between cities i and j, depends on the population of city i, the population of city j, and the distance between cities i and j. For example, the number of passengers between Seoul and Jeju depends on the population of Seoul and Jeju as well as the distance between them, which is 449km. The job of researchers who use the gravity model in practise is to estimate the value of the elasticities, E1, E2 and E3 in equation (3.27). For example, an estimated version of (3.27) is the following gravity model: PAXij =

୔୓୔బ.లళ ୔୓୔బ.లళ ౟ ౠ

(3.28)

ୈబ.ళ ౟ౠ

In this case the population elasticities are 0.67 each and the distance elasticity is -0.7. The population elasticity implies that if two city pairs, A and B, are the same distance apart, but city pair A has a population that is 10% greater than the population of city pair B then city pair A will have 6.7% greater passenger demand than city pair B. The distance elasticity means that if two city pairs, A and B, have identical populations, but the city pair A distance is 10% longer than the city pair B distance, then city pair A will have 7% lower passenger demand than city pair B. To use the gravity model to estimate or predict the number of passengers that we would expect to travel by air on a city pair we simply plug the population for the cities in the city pair, and the city pair distance, into equation (3.28). If we do this for the Sydney to Melbourne route, for example, where POPSYD = 5m, POPMEL = 5m and DSYD-MEL = 705km, then we obtain the following estimate of demand on the city pair using (3.28): PAXSYD-MEL =

(ହ,଴଴଴,଴଴଴)బ.లళ (ହ,଴଴଴,଴଴଴)బ.లళ ଻଴ହబ.ళ

= 9.6m

This indicates that the gravity model (3.28) predicts passengers carried of 9.6m per annum for the Sydney to Melbourne city pair.

3.10 Demand and Network Effects 3.10.1 What is a Network Effect? An airline’s network is the array of city pairs over which the airline flies. The wider an airline’s network the greater is the number of city pairs flown by the airline. Table 3-3 below presents the network size for key airlines in the global aviation market. The biggest airlines in global aviation, such as the U.S. airlines United, American and Delta, have incredibly big networks with United Airlines flying to 235 domestic cities and 127 international cities at the time of writing,16 American Airlines flying to 237 domestic cities and 126 international cities,17 and Delta Air Lines flying to 226 domestic cities and 97 international cities.18 Passengers value airlines that fly over a wide network of city pairs. This is because passengers prefer a one-stop shop when it comes to buying airline tickets. In other words, when booking flights passengers would rather book on a single website and fly on a single airline. This reduces the time it takes for the passenger to deal with airline booking and reservation systems. Rather than booking complex flight itineraries that involves several city pairs and stopovers across two or more airlines with two or more flight bookings, the passenger need only make a single booking and fly on one airline if the airline has a sufficiently wide network. Booking and flying with a single airline is also likely to result in a more seamless travel experience if a passenger is required to stopover at several airports. For example, a passenger flying with United Airlines from Baltimore (BWI), Maryland in the U.S.A. to Detroit (DTW), Michigan in the U.S.A. will do so by stopping off in Chicago (ORD) for an hour. As the passenger flies with United for the whole journey, the arrival gate at Chicago is likely to be relatively close to the departure gate at Chicago so that the passenger need not be concerned about rushing to a departure gate that is a long distance from the arrival gate. The airline knows that a number of passengers arriving at ORD from BWI are destined for DTW, which enables the airline to schedule a departure time for ORD-DWT that is relatively close to the arrival time 16

“United Airlines Network”, Flight Connections, accessed October 1, 2021. https://www.flightconnections.com/route-map-unitedairlines-ua. 17 American Airlines Network, Flight Connections, accessed October 1, 2021. https://www.flightconnections.com/route-mapamerican-airlines-aa. 18 “Delta Air Lines Network”, Flight Connections, accessed October 1, 2021. https://www.flightconnections.com/route-map-deltadl.

Passenger Demand

69

Airline Australia

Domestic Network

International Network

Qantas Mainline

59

28

Jetstar Australia

19

15

Virgin Australia

40

15

Singapore Airlines

0

66

Jetstar Asia

0

22

Malaysia Airlines

30

45

Air Asia

18

59

Japan Airlines

44

48

All Nippon Airways

53

53

British Airways

17

85

easyJet

16

145

Southwest

88

15

United

235

127

American

237

126

Delta

226

97

Thai International

4

59

Thai Air Asia

23

41

Air China China Southern

124 149

75 78

China Eastern

154

79

19

209

Ryanair

5

223

Aer Lingus

6

98

31

75

1

166

Singapore

Malaysia

Japan

United Kingdom

USA

Thailand

China

Germany Lufthansa Ireland

France Air France Netherlands KLM Source: Flight Connections 2021

Table 3-3: Size of Airline Networks

Chapter 3

70

Airline New Zealand

Domestic Network

International Network

Air New Zealand

20

33

Air Canada

59

153

Westjet

43

67

Philippine Airlines

31

43

Cebu Pacific

26

25

22

45

LATAM

17

127

Azul Airlines

105

9

Avianca

28

49

Copa

2

78

Aeroméxico

43

41

Volaris

40

26

Korea Air

13

103

Asiana

10

66

Emirates

1

143

Etihad

2

80

1

177

Air India

56

44

Spicejet

53

9

Indigo

63

23

South African Airways

22

38

Mango Airlines

8

1

1

64

6

119

Canada

Philippines

Vietnam Vietnam Airlines South American

Mexico

Korea

United Arab Emirates

Qatar Qatar Airlines India

South Africa

Israel El Al Austria Austrian Airways Source: Flight Connections 2021

Table 3-3 (cont): Size of Airline Networks

Passenger Demand

71

Airline Switzerland

Domestic Network

International Network

Swiss Air

2

119

SAS

5

128

Norwegian

15

100

2

145

Iberia

34

106

Vueling

28

91

7

84

Scandinavia

Hungary Wizz Air Spain

Portugal Tap Source: Flight Connections 2021

Table 3-3 (cont): Size of Airline Networks of BWI-ORD. This allows the airline to reduce the passenger’s transit time at stopovers, which in turn reduces the total journey time of the passenger. The passenger’s baggage is carried by a single airline, which means the details associated with that baggage is collected and stored by a single airline and does not need to be communicated between United and other airlines, thus reducing the chance that the passenger’s baggage is lost. If the flight leaving BWI for ORD is delayed, then the transfer of information about the delay takes place within United’s communication systems and does not need to be communicated with other airline systems. This will enable the airline to make better and faster decisions about how the passenger is to be conveyed from ORD to DTW, including delaying the original flight or placing the passenger on a later flight. The value to an airline of offering passengers a wider network is stronger airline specific demand. An airline may be able to steal or take demand from another airline if it is able to offer passengers a wider network of city pairs. In the context of the airline specific demand functions introduced in section 3.5, a wider network will shift those airline specific demand curves to the right or increase the intercept term in those demand functions.

3.10.2 Networks, Indirect Services and Hub Points A direct service is defined as a flight in which a passenger who lives in city A who wishes to travel to city B can fly on a single flight between A and B. It is often the case that there are not enough passengers for an airline to fly directly between two city pairs such as A and B. This may be because cities A and B are both very small, they are a very long distance apart, and there are limited trade or family connections between the two cities. When this is the case, it may be possible for an airline to carry the passenger between A and B but not on a single flight. The airline may be able to carry the passenger from A to B on two or more flights. For example, if the flight from A to C is popular and the flight from C to B is popular, then the airline can carry the passenger from A to C and then from C to B. This is referred to as an indirect service between A and B. The airport C in this case is called the hub airport or the aggregator airport. An airline can fly several services into and from its hub airport to connect passengers between two cities that do not have direct services. An airline can do this if it has a relatively wide network of city pairs on which it flies. For example, the domestic airlines in Australia Qantas, Jetstar, Virgin Australia, and Rex Express, do not operate direct services between the capital city of the state of Tasmania, Hobart (HBA) and the capital city of the state of Western Australia, Perth (PER) at the time of writing. The airlines do operate services, however, between HBA and the capital city of the state of Victoria, Melbourne (MEL), and between MEL and PER, because these services are popular and profitable. This means that the airlines can take the passenger from HBA to MEL and then from MEL to PER. This is an indirect service because the airlines have carried the passenger between two city pairs (HBA and PER) on more than one flight (HBA-MEL and then MEL-PER). In this case we call Melbourne the hub or aggregation point. Passengers will usually need to fly on an indirect service if they wish to fly between a relatively small city and a relatively big city, two relatively small cities, a big city and a relatively small city, and medium size cities that are a long distance apart. In these cases, the population catchments are too small, and therefore demand is too small, for the airline to profitably operate a direct service. Or the population catchments are relatively big but the distance between the population centres is too great to generate enough demand to profitably operate a direct service.

Chapter 3

72

For example, a resident of Cardiff Wales wishes to visit a friend in Frankfurt Germany. Cardiff has a small population catchment of around 500,000 according to macrotrends19 while Frankfurt has around 40m people living within 200kms of the airport.20 There is not enough demand across the two population centres to allow airlines to profitably operate a direct service between the two cities. However, the Dutch airline KLM, which is a part of the Air France / KLM Group, operates a service from Cardiff Airport (CWL) to Amsterdam Airport (AMS) and from AMS to Frankfurt Airport (FRA). The Dutch carrier KLM uses Amsterdam as a hub to enable flights that carry passengers who live in Cardiff to Frankfurt and passengers who live in Frankfurt to Cardiff. While there is not enough demand for KLM to operate CWL-FRA directly there is enough demand to operate CWL-AMS services and enough demand to operate AMS-FRA services, enabling the airline to connect passengers in Cardiff to Frankfurt. Consider another example, this time in Asia. Suppose a passenger from the central Vietnamese city of Hue, which has a population of around 400,000,21 wishes to fly to Singapore. There are no direct flights between Hue Airport (HUI) and Changi Airport in Singapore (SIN) because demand is not strong enough for this city pair to justify a direct flight. While Singapore is a relatively big city with a population of almost 6m,22 Hue is a relatively small city. It is also the case that there is a relatively long distance between the two cities at 1,713km, which means that the relatively high airfare that is required to be paid for a flight on this city pair may not be affordable to many of the people living in Hue. To fly between the two cities the passenger must first fly from HUI to Ho Chi Minh city (SGN) and then from SGN to SIN on Vietnam Airlines. Vietnam Airlines can offer this indirect service because of its network of flights throughout Vietnam and Asia and because it has a hub at Ho Chi Minh city airport.

3.10.3 Before, Beyond and Trunk Routes When a passenger flies on an indirect service this usually involves a combination of flying between relatively small cities and relatively big cities. This can include the following combinations: 1. a passenger flies from a relatively small city to a relatively big city, and then between two relatively big cities; 2. a passenger flies from a relatively big city to a relatively big city, and then from a relatively big city to a relatively small city; and 3. a passenger flies on more than two sectors, or with more than one stop, involving 2 or more relatively small cities and 2 or more relatively big cities, for example, a passenger flies from a relatively small city to a relatively big city, then from a relatively big city to another relatively big city, and then from a relatively big city to a relatively small city. Let us now look at some illustrations of these types of indirect journeys by air. As an example of trip type 1., a passenger flies from Milwaukee, Wisconsin in the U.S., which is a relatively small city, to a large city like Los Angeles, via a relatively large city like Denver. For example, Frontier Airlines, which is an ultra-low-cost carrier that has a hub in Denver, Colorado U.S.A., will carry passengers from General Mitchell International Airport in Milwaukee (MKE) to Denver International Airport (DEN), and then from DEN to Los Angeles International airport (LAX), which is a journey that can be summarised as MKE-DEN-LAX – see Figure 3-18 below.

Source: Great Circle Mapper (http://www.gcmap.com/)

Fig. 3-18: Flight from Milwaukee to Los Angeles Via Denver 19

“Cardiff Wales Population,” Macrotrends, accessed October 1, 2021. https://www.macrotrends.net/cities/22843/cardiff/population. “Location Advantages at Frankfurt Airport,” Frankfurt Airport., accessed October 1, 2021. https://www.frankfurt-airport.com/ en/article-migration/b2b/airlines_tourism/location-advantages/advantages.html. To access this search for “Frankfurt airport population catchment area”. 21 “Hue Vietnam Population,”, Macrotrends, accessed October 1, 2021. https://www.macrotrends.net/cities/23366/hue/population. 22 “Singapore Population”, Worldometers, accessed October 1, 2021. https://www.worldometers.info/world-population/singaporepopulation/. 20

Passenger Demand

73

The passenger has flown from a relatively small city, Milwaukee to a relatively large city, Denver, and from a relatively large city, Denver to a another relatively large city, Los Angeles. Now consider an illustration of the second type of trip. A passenger living in Adelaide which is the capital city of South Australia with a population of around 1.3m, wishes to travel to the small sea-side town of Coffs Harbour, New South Wales, Australia, which has a population of around 26,000. To do this the passenger must first fly from Adelaide to Sydney, New South Wales which has a population of around 5m, and then fly from Sydney to Coffs Harbour. The itinerary for this traveller is therefore Adelaide (ADL) to Sydney (SYD) and then SYD to Coffs Harbour (CFS), which is a journey that can be summarised as ADL-SYD-CFS. This trip involves travel from a relatively large city to another large city and then to a small city – refer to Figure 3-19 below.

Source: Great Circle Mapper (http://www.gcmap.com/)

Fig. 3-19: Flight from Adelaide to Coffs-Harbour Via Sydney This itinerary can be operated by both large Australian domestic carriers Qantas and Virgin Australia. Let us now consider an example of travel type 3. A passenger who lives in Edinburgh, Scotland wishes to fly to Nagasaki in Japan, both relatively small cities. To enable this, Lufthansa Airlines, which is the national carrier of Germany, will fly the passenger from Edinburgh (EDI) to Frankfurt (FRA), then from FRA to Haneda (HND), Tokyo, Japan and then from HND to Nagasaki (NGS), Kyushu Island, Japan with the last leg flown by All Nippon Airways. The passenger has therefore flown from a relatively small city in Edinburgh, to a relatively big city in Frankfurt, then from Frankfurt to a relatively big city, Tokyo and then from Tokyo to a relatively small city in Nagasaki, with a journey that can be summarised as EDI-FRA-HND-NGS – refer to Figure 3-20 below.

Source: Great Circle Mapper (http://www.gcmap.com/)

Fig. 3-20: Flight from Edinburgh to Nagasaki via Frankfurt and Tokyo

Chapter 3

74

When a trip involves two or more stopovers this increases the chance that more than one airline must be involved in operating the flight. In the above Edinburgh to Nagasaki trip illustration, this required two sets of airlines to complete the trip - Lufthansa and All Nippon Airways, because Lufthansa is not permitted to operate services within Japan. When more than one airline is required to fly the passenger between the home port and the destination port, the airline that issues the ticket, in our case Lufthansa, must organise the passenger to fly on another carrier when required, in this case All Nippon Airways for the domestic Japan leg of the passenger’s journey. This will involve the airline finding a codeshare partner, an alliance partner or an interline partner to operate the last leg or two of the flight. We will learn more about these three partnerships in Chapter 10. In our three indirect service combinations presented above, the passenger flies on before legs, trunk legs and beyond legs. A before leg is a service from a relatively small city to a relatively big city that occurs prior to a service between two relatively big cities. In our Milwaukee to Los Angeles illustration given above in Figure 3-18, the passenger travels on the before leg Milwaukee to Denver, which involves travel between a relatively small city and a relatively big city, prior to travelling from Denver to Los Angeles, which involves travel between two relatively big cities. In our Edinburgh to Nagasaki illustration presented in Figure 3-20, the before leg is Edinburgh to Frankfurt, which is a trip from a relatively small city to a relatively big city that occurs prior to the trip Frankfurt to Haneda, which is a trip between two relatively big cities. A trunk leg involves travel on a relatively thick route, or a route with high levels of demand, which usually involves travel between two relatively big cities. In our Milwaukee to Los Angeles example, the trunk route is Denver to Los Angeles. In our Adelaide to Coffs Harbour example (Figure 3-19), the trunk route is Adelaide to Sydney. In our Edinburgh to Nagasaki illustration the trunk route is Frankfurt to Haneda. In each of these cases, the trunk route involves travel between city pairs that have relatively high levels of demand and between two relatively big airports and population catchments. A beyond leg is a service from a relatively big city to a relatively small city that occurs after travel between two big cities. In our Adelaide to Coffs Harbour illustration in Figure 3-19, the beyond leg is Sydney to Coffs Harbour, which is travel from a relatively big city to a relatively small city that occurs after travel between relatively big cities. In the Edinburgh to Nagasaki illustration, the beyond leg is Haneda to Nagasaki, because this involves a trip between a relatively big city and a relatively small city which occurs after the Frankfurt to Haneda trip, which occurs between two big cities.

Quiz 3-4 Changes in Demand Over Time and Network Effects 1. (a) (b) (c) (d) (e) (f) (g) (h) (i) 2. (a) (b) (c) (d) (e) (f) (g)

Indicate whether the following forces explain the trend, cycle, season, structural events, or irregularity in air travel demand over time. Gross Domestic Product. Aggregate share price indices. Exchange rates. Interest rates. An act of terror at an airport. An act of terror in the air as an aircraft is flying. Volcanic ash in air space. Unemployment rate. Population growth. Indicate whether the following scenarios involve network effects and hubs and whether a leg is a before, trunk or beyond leg. If you do not know the 3-digit airport code, you will need to use the internet to research the code. A passenger who lives in Melbourne wishes to travel to Sydney to visit a friend on a flight between Melbourne and Sydney. A passenger who lives in Port Macquarie (PQQ), which is a small city in the northern part of the state of New South Wales, Australia, needs to travel to Melbourne for a conference. The passenger flies into Sydney from Port Macquarie before catching a connecting flight to Melbourne. A politician who lives in Canberra (CBR) needs to travel to London (LHR) to meet other politicians. The passenger flies into Sydney from Canberra before catching a connecting flight to London via Dubai (DBX). A passenger who has Danish heritage wishes to fly back to Copenhagen (CPH) from Sydney. The passenger flies British Airways from Sydney into London Heathrow and then catches a connecting flight from London Heathrow into CPH. A passenger from Singapore who wishes to visit a friend in Perth flies on a Singapore Airlines flight from Singapore to Perth. A passenger from Brisbane wishes to fly to Lord Howe Island (LDH) for a peaceful holiday by travelling via Sydney. A passenger flies from Singapore to Jeju Island (CJU) for a holiday. The passenger flies from Singapore to Seoul and then from Seoul to Jeju.

Passenger Demand

75

(h) A resident of Brunei travels from Bandar Seri Begawan, Brunei (BWN) to Manila, Nino Aquino International Airport (MNL) on Royal Brunei Airlines. (i) A South African resident living in Cape Town wishes to travel to Cairo. The resident travels from Cape Town International Airport (CPT) to Bole International Airport, Addis Ababa (ADD) on Ethiopian Airlines and then travels on EgyptAir between ADD and Cairo International Airport (CAI). (j) An Argentinian resident wishes to travel from Buenos Aires, Argentina to Caracas, Venezuela. There is no direct flight between the cities, but there is an indirect flight via Panama City. The resident flies from Ezeiza International Airport (EZE) to Tocumen International Airport, Panama (PTY) and then from PTY to Simón Bolivar International Airport (CCS), Venezuela on the hub carrier Copa Airlines, which uses Panama City as a hub. (k) A resident of New Orleans travels to Mexico City. The resident travels from Louis Armstrong New Orleans International Airport (MSY) to Dallas/Fort Worth International Airport (DFW) and then from DFW to Mexico City International Airport (MEX). Both legs of the flight are operated by American Airlines. (l) A resident of the city of Campinas in São Paulo, Brazil wishes to fly to the capital of Denmark, Copenhagen. To make the journey the resident flies from Viracopos International Airport (VCP) to RIOgaleão – Tom Jobin Airport (GIG) in Rio de Janeiro on a flight operated by GOL Airlines. The passenger then flies from GIG to Lisbon Portela Airport (LIS) on Tap Air Portugal and then from LIS to Copenhagen Airport (CPH).

CHAPTER 4 SHORT RUN AIRLINE REVENUE

One of my duties at Qantas when I was Chief Economist was to build models of airline revenue and use those models as the basis for building forecasts. I did this as head of the Economics Department of the airline, but it was really to provide support for the Pricing and Revenue Management and Revenue Forecasting Departments. The one thing I learnt, amongst many, about building models of airline revenue was just how difficult it is to build models that could accurately explain revenue. This difficulty comes about because airline revenue, like demand, is affected by so many different variables, ranging from macroeconomic variables such as GDP, equity indices, the population, employment, and the exchange rate, to adverse shocks such as viruses, ash cloud at altitude, and earthquakes. These variables are volatile and difficult to predict, hence making it difficult to model and to forecast airline revenue with any accuracy. The most important challenge that an analyst confronts when modelling short run airline revenue is in building a relationship between the average airfare and the passenger seat factor. What spiked my interest in this relationship between airfares and passenger load, is a conversation I had with a Qantas pilot back in 2014. When Qantas’ earnings were at their weakest level in history, my colleague the Qantas pilot asked me why earnings were diabolical despite the planes being completely full of passengers. My answer to him was that sometimes a full plane is not the most profitable plane. I continued to say that sometimes it is better for revenue and profits to operate the plane at an 80% seat factor rather than a 100% seat factor because the benefits of a higher average airfare for revenue and profits dominates the benefit of selling more seats. In this chapter you will be confronted with some mathematics – a bit of algebra and some calculus. The level of algebra and calculus you need to understand is intermediate level mathematics, which is around the same level of mathematics that many of you confronted during your last year of high school. If you are reasonably competent at this level of mathematics, then the only thing stopping you from understanding the mathematics that I present in this chapter is the topic to which the mathematics is applied. I am hoping that I can explain the topic to you in a way that this will not pose a problem. It is up to you to decide how much of the mathematics you wish to understand. I realise that some people have a love-hate relationship with mathematics, and the mathematics is not completely necessary. As such, it should not be forced upon you. My brain is wired analytically, which means I enjoy the mathematics and I can see the power of it in understanding the complexity associated with airline revenue. The most important part of the revenue analysis is the intuition, so if you are not wired analytically then try and follow the mathematics but prioritise the logic and intuition. In this chapter we examine in some detail the relationship between revenue, the average airfare, and the passenger seat factor. This relationship between the three variables is a short run relationship because it is examined for a given number of seats, which can only be varied over a medium run horizon. When we allow the number of seats and the capacity of the airline to change, then we are investigating a medium run revenue relationship. You will have to wait until the next chapter, chapter 5, to see the relationship between capacity and revenue over the medium run. Where I believe it to be useful, I will build my revenue numerical examples in Microsoft Excel so this chapter will give you the opportunity to learn some new Microsoft Excel techniques. These techniques are exceptionally valuable if you are to work in airline or airport management. I remember back when I worked at Qantas that the analysts that were the most sought after were those with the highest level of Microsoft Excel skills.

4.1 Airline Time Horizons – Short, Medium and Long Runs Airlines operate to three timeframes – the short run, the medium run, and the long run. Each of these runs or timeframes is defined according to the constraints that the airlines face and how these constraints are relaxed over time. It is important that we talk about these runs or timeframes because the levers or variables that the airline can manipulate to change revenue differs across the various timeframes.

4.1.1 Short Run Decision Variables The passenger seat factor (PSF) is a short run decision variable. It represents the percentage of seats on the plane that are occupied by passengers as discussed in chapter 2. The airline varies the PSF by varying the number of seats it is prepared to sell at the relatively cheap fares and the relatively expensive fares. It can sell more seats on the flight, thus increasing the PSF, if it offers more seats at cheaper fares. Conversely, the airline can operate to a lower PSF if it offers fewer seats at lower fares and more seats at higher fares.

Short Run Airline Revenue

77

The average airfare is also a short run decision variable. It is the passenger revenue from a flight divided by the number of passengers on that flight as discussed in chapter 2. The average airfare, like the PSF, will depend on decisions by airline yield management analysts to allocate more seats at cheaper fares and fewer seats at higher fares or vice-versa. If the airline allocates more seats to cheap fares this lowers the average airfare, and if it offers fewer cheap seats than it raises the average airfare. Marketing is also a short run decision variable. Airlines use marketing to promote their brand and to sell more seats to passengers. It is often combined with sales prices when an airline wishes to promote cheap seats on a route or across a network. Marketing not only depends on how much to spend but also on the instruments that the airline chooses to use, such as billboards, TV, radio, newspapers, magazines, emails, and social media. Customer service in the air, at the airport and after the plane has landed are also operational levers that an airline can alter over a short horizon. This may be as simple as a request by management for cabin crew and check-in staff to smile more, or to be more attentive to customer needs. Or it may be more involved such as training associated with cabin crew dealing with disruptive or aggressive passengers. Many customer service changes can be made relatively quickly. Changes to key loyalty scheme parameters and variables are also considered short run decision variables of the airline. What variety of goods and services can loyalty members buy to earn points? How many points are earned for a given amount of money spent on a credit card? What goods and services can the loyalty member buy by redeeming points? All these critical loyalty scheme parameters can be altered relatively quickly. Airlines will often decide to cancel services over a short horizon when it is difficult to fill the plane with passengers. Cancelling services in this way is often referred to as slow flying. The airline will usually only slow fly when they can off-load those affected passengers onto adjoining flights on the same city pair. This will be the case when the airline operates regular services. Cancelling services or slow flying is therefore a short run decision variable of the airline.

4.1.2 Medium Run Decision Variables It takes some time for airlines to change the frequency of its services. This is largely because the time of departure of a service and the aircraft to be used on that service have usually been advertised on an airline’s website up to a year before departure. Airlines set their schedules, including the time and date of a service and the aircraft to be used, up to a year before departure to allow passengers time to plan their trips and book flights and to lock in landing and take-off slots at airports. Changing the frequency of services is therefore a medium run decision variable of the airline. The route network is a measure of the variety of city pairs on which an airline operates. Adding and deleting city pairs is a time-consuming process because the airline must build business cases for doing so. In addition, when servicing a port for the first time, the airline is required to employ people at the airport and to employ a sales team at the new destination, all of which takes time. Changing the route network is therefore a medium run decision variable of the airline. It takes time for airlines to find airline partners and develop relationships. These relationships may be as simple as selling seats to passengers on flights operated by the partner airlines, such as interline and codeshare arrangements. They could also be more complex such as setting up an arrangement in which airline partners share revenue, one airline may buy an equity stake in another airline, two airlines may merge, or one airline may acquire another airline. These types of relationships usually develop over a medium run horizon, as they require time to determine the benefits and the costs of the relationship. The topic of airline relationships will be examined in more detail in chapter 10. Altering the labour force of the airline also takes time. It takes time for airlines to place job ads for the labour that it is in search of, to engage an employment agency, to conduct interviews and find the right person for the job. The person offered the job may not be able to start for some time because they must give their current employee a period of notice. Similarly, it takes time for airlines to shed labour. They must decide on the number of labour units to shed, which employees will be told to leave, and exiting employees will be offered a certain notice period. The airline will also need to engage with labour unions to discuss the rationale for shedding labour, which is likely to draw an adverse reaction from unions, further increasing the time it takes to downsize the labour force. Changing the stock of labour of an airline is therefore a medium run decision variable.

4.1.3 Long Run Decision Variables It takes time for airlines to choose the right aircraft type and the right configuration of seats on that aircraft. Once an airline has decided on the aircraft and configuration that it wants, it must then sit in a production queue with aircraft manufacturers. It is also the case that orders are often delayed, which in turn adds to the length of time it takes to change the stock of fleet. For these reasons, changing the number of operating fleet units of the airline takes a long period of time. It may also take time before the airline finds an institution(s) or organisation(s) that will contribute to the funding of aircraft investments. These institutions or organisations include commercial and investment banks, which need to agree on key details of any lending arrangement. Reaching such an agreement can take a long period of time. Banks will take their time in this process because the amount of money they are likely to lend to airlines could be in the hundreds of millions of dollars reflecting the price of new aircraft, while many airlines are considered a risky investment because of their volatile earnings.

Chapter 4

78

Over a long horizon, airlines will also make decisions about their use of fossil fuel versus renewable fuels. This decision will take a long period of time because the renewable fuel must be compatible with current engine technology, and/or engine technology must be altered to permit the use of renewable fuels. Airlines buy other airlines (horizontally integrate), and they buy hotels, catering and ground handling businesses, and shares in oil companies or refineries (vertically integrate). They also buy into completely different industry sectors (they diversify). Decisions associated with large investments such as these take time. Airline analysts must put together a business case for these investments, the business case must often be taken to the airline board for approval, and it takes time for the counterparties to reach a mutually beneficial agreement. It also takes time for airlines to enter markets or the industry, and to exit markets. The airline must receive an Air Operator’s Certificate from local regulators, which effectively indicates that those regulators believe the airline is safe to fly in the local market. Before an airline can exit the market, it must attempt to service all passengers that have been booked on flights, and administrators of the airline must return at least some money to creditors, including passengers, and negotiate a way out of bankruptcy if possible – all of this can take a long period of time.

4.2 Observations About the Passenger Seat Factor Figure 4-1 below presents the passenger seat factor in 2017 for 219 airline flying segments around the world.23 It shows that over calendar 2017 there were a wide range of passenger seat factor outcomes for different airlines, with the highest passenger seat factor over a calendar year recorded by the European low-cost carrier (LCC) Ryanair at 96.0% and the lowest recorded by the regional Australian carrier Regional Express (Rex) Airlines at 60.0%. The average passenger seat factor for 2017 across the world was found to be 81.3%. Figure 4-1 indicates that there would therefore appear to be significant differences in airline views about the level of the passenger seat factor that is likely to generate maximum profit. Should airlines operate to passenger seat factors that are above 90% as is the case for the aggressive LCCs Ryanair, SpiceJet, Wizz Air, easyJet and Transavia? Or should they be operating to passenger seat factors in the 70’s and low 80s like airlines such as Thai Airways, Alaska, Turkish Airlines, Korean Air, Air India, and Eva Air? What factors are likely to be driving these decisions? Why do LCCs operate to a much higher average passenger seat factor than full-service airlines (FSAs)? And why do regional carriers tend to operate to a much lower passenger seat factor? As indicated in Table 4-1 below, the LCC average passenger seat factor in calendar 2017 was almost 5 percentage points higher than the average passenger seat factor for FSAs, and 15 percentage points higher than the passenger seat factor for regional airlines. In the following sections we will attempt to answer some of these interesting and important aviation questions. Airline Type

Number of Flying Segments

Average Passenger Seat Factor

Maximum Passenger Seat Factor

Minimum Passenger Seat Factor

FSA LCC REG (Regional)

175 39 4

80.7% 85.4% 70.1%

89.8% 96.0% 80.8%

66.1% 76.5% 59.9%

Standard Deviation Passenger Seat Factor 4.3% 4.2% 8.6%

Source: Airline Intelligence and Research Database 2021

Table 4.1: Global Passenger Seat Factors in 2017 by Airline Type

4.3 Short Run Revenue Definitional Relationships Over a short time-horizon, the airline faces constraints over its ability to sell seats to passengers. This is because over the short run the airline has a fixed number of seats that it can sell. These seats or capacity have been locked into place for up to 12 months before the departure of the flight.24 Faced with this constraint, we wish to understand the impact on revenue if the airline chooses to fly a relatively full plane at a low average fare, or a plane with many vacant seats at a higher average airfare. Another way of saying this is that we wish to analyse the impact on revenue if the airline targets a higher or lower passenger seat factor. Our first step in response to this challenge is to set-up the definition of an airline’s passenger revenue. This is equal to the number of passengers carried on the plane multiplied by the average airfare paid by those passengers. We can write this passenger revenue definition in the following way:

23

A flying segment for the purpose of Figure 4-1 is defined as an aggregate of domestic, international, or regional flying, or it may involve geographic aggregates such as South-East Asia, China services, Japan services, European services, and Americas services. Estimates are also presented at the total network level in Figure 4-1. 24 Airlines can and will cancel services over a short period of time, which can have a minor impact on capacity in the short run. The capacity adjustments that we are talking about in this section are relatively substantial changes that require passengers to be notified 6 to 12 months in advance.

Fig. 4-1: Global Passenger Seat Factors

0%

10%

20%

30%

40%

50%

60%

59.9% 70%

80%

90%

100%

2017 Passenger Seat Factor

Source: Airline Intelligence and Research Database 2021

96.0%

79 Short Run Airline Revenue

Spicejet domestic Wizz Air Total System Spicejet international TigerAir Australia Total System Norwegian Air Shuttle Total System Indigo Airlines Domestic Air France/KLM Total System Delta Air Lines Latin America Delta Air Lines Pacific Hawaiin Airlines Total System Delta Air Lines Domestic Delta Air Lines International Nok Air Total System LATAM Airlines Total System Air New Zealand North America/UK Jetstar Australia Domestic Jetstar Group Total System Delta Air Lines Atlantic Iberia Total System International Airlines Group Latin America and Carribean Southwest Airlines Total System West Jet Total System Air Canada Pacific Thai Airways Europe Jetstar Asia Aegean Air Total System Air Canada Domestic China Eastern Domestic Japan Airlines Europe American Airlines Pacific Air France/KLM Africa and Middle East China Southern Domestic Jet Airways Total System United Airlines Total System Air Canada Total System Pegasus Airlines Total System International Airlines Group Europe British Airways Total System Air New Zealand Tasman Aero Mexico Domestic Air New Zealand Domestic Swiss Airlines Total System Aer Lingus Total System China Eastern Total System Lufthansa Group Total System Finnair Europe Delta Air Lines Regional GOL Airlines Domestic Alaska Airlines Regional Japan Airlines Korea GOL Airlines Total System Volaris International Korean Air Total System United Airlines Pacific Asiana Airlines International American Airlines Latin America Air India International Air China International Scandinavian Airlines System Intercontinental Thai Airways Regional Japan Airlines Total System Cathay Pacific China Scoot Total System All Nippon Airways International GOL Airlines International South African Airlines Total System Oman Air Total System Scandinavian Airlines System Europe All Nippon Airways Total System Japan Airlines Domestic All Nippon Airways China Scandinavian Airlines System Domestic Rex Express Total System

Chapter 4

80

Passenger Revenue = Average Fare u Passengers

(4.1)

We can write equation (4.1) in a different way by utilising the fact that the number of passengers is equal to the PSF multiplied by the number of seats: Passengers = PSF u Seats

(4.2)

In the short run, the number of seats is fixed. Let us define this fixed number of seats to be ܵҧ (the bar at the top of a variable is often used to denote a variable that is fixed in economics). If we substitute the fixed seats variable into the passenger formula (4.2) then we obtain: Passengers = PSF u Sത

(4.3)

If we substitute (4.3) into the passenger revenue expression at (4.1) then we obtain an expression for short run passenger revenue that depends on the PSF and the fixed number of seats: Passenger Revenue = Average Fare u PSF u Sത

(4.4)

Equation (4.4) says the passenger revenue that an airline earns in the short run depends on the average airfare, the passenger seat factor and the fixed number of seats that are on offer. Because the number of seats is fixed, this places an upper limit on the number of passengers that can be flown and a lower limit on the average airfare. This in turn places a constraint on the amount of passenger revenue that the airline can earn in the short run. Equation (4.4) is called a definitional relationship. It must hold true on all occasions. We wish to add a behavioural relationship to equation (4.4) that will make this passenger revenue equation more complicated. It will take (4.4) from being a definitional relationship to being a behavioural relationship. The behavioural relationship that we wish to add to (4.4) is a relationship between the average fare component on the right-hand side of (4.4) and the PSF. In the next section I will show you how to build a behavioural relationship between the PSF and the average airfare. This behavioural relationship is the basis for much of the analysis to be conducted in this chapter so make sure your attention energy is high when you read through the next section.

4.4 Relationship between Average Airfares and the Passenger Seat Factor 4.4.1 The Logic There is an inverse relationship between the PSF and the average airfare. A completely full plane will have a lower average airfare than a plane that is only half full, other things being equal. This inverse relationship between the PSF and the average airfare is an extremely important relationship for the airline, especially for determining the average airfare and PSF strategy of the airline that maximises profit in the short run (which is something we will talk more about in chapter 7). Most importantly, the negative relationship tells us that a plane that is completely sold out, or the PSF is 100%, may not necessarily maximise passenger revenue or profit over a plane that is, say, three quarters full, because the full plane leads to an average airfare that is too low. To illustrate, consider a particular Jetstar flight from Sydney to Melbourne on an A320 with 180 seats in a single class configuration. The airline sells seats at 11 different prices (we call these price points) and targets a seat factor of 95%, which represents 171 seats. When it targets 171 seat sales it generates the following sales into the relevant price points. Price Points Sales to PAX PAX Revenue

$67 40 $2,680

$72 35 $2,520

$78 30 $2,340

$85 24 $2,040

$93 18 $1,674

$102 9 $918

$133 5 $665

$149 4 $596

$171 3 $513

$199 2 $398

$292 1 $292

Total 171 $14,636

The average airfare and revenue that the airline derives from targeting a 95% PSF and selling tickets to passengers at the prices given above is found by dividing total passenger revenue, $14,636 (found in the last column and last row) by the total passengers, 171 (found in the last column and middle row), which gives $85.59. By targeting this relatively high seat factor, the airline risks selling tickets to passengers that buy a low-price ticket but are willing and able to pay higher airfares. If the airline targets an 85% seat factor, which involves selling to 153 passengers, it can generate the following set of sales: Price Points Sales to PAX PAX Revenue

$67 32 $2,144

$72 25 $1,800

$78 21 $1,638

$85 18 $1,530

$93 15 $1,395

$102 12 $1,224

$133 10 $1,330

$149 8 $1,192

$171 5 $855

$199 4 $796

$292 3 $876

Total 153 $14,780

Short Run Airline Revenue

81

For these sales, the average airfare outcome for the airline is $96.60, found by dividing passenger revenue of $14,780 by 153 passengers. By restricting the number of seats for sale, the airline secures a better outcome for the average airfare and passenger revenue.

4.4.2 Linear Relationship between the Average Airfare and the Seat Factor Economists use simple tools to describe the negative relationship between the seat factor and the average airfare. The first simple tool that we will use is called the average airfare function. We present a straight-line example of the average airfare function in Figure 4-2 below. Average Airfare € 1,100 € 1,000 € 900 € 800 € 700 € 600 € 500 € 400 € 300 99% 96% 93% 90% 87% 84% 81% 78% 75% 72% 69% 66% 63% 60% 57% 54% 51% 48% 45% 42% 39% 36% 33% 30% 27% 24% 21% 18% 15% 12% 9% 6% 3% 0%

Passenger Seat Factor Fig. 4-2: Straight Line or Linear Airfare Function – Air France Example The information contained in Figure 4-2 is derived from the following details about an Air France flight: An Air France flight departs on the 17th of May 2017 from Charles De Gualle (CDG) Airport in Paris bound for Frankfurt (FRA) Airport in Germany. Air France opens bookings for the flight on 18th May 2016. The flight is serviced by a 200 seat Airbus A320-200 aircraft with a single (economy class) configuration. The airline knows that it can fill 75% of the plane, that is, sell 150 seats on the plane, at an average airfare of €500. It also knows that only 1 person is likely to be willing to pay €1,000 for the flight.

We wish to find an equation that describes the downward-sloping straight line in Figure 4.2 based on the information about the Air France flight presented above. This downward-sloping straight line is a linear equation. We can write this linear equation generally as: Average Airfare = a + b u PSF

(4.5)

We say generally here because the parameters of the straight-line equation in (4.5), a and b, have not been given specific numerical values. The a term in equation (4.5), called the intercept, is essentially the average airfare that the airline would receive if it carried a single passenger. It is sometimes called the reservation airfare or the maximum airfare that any single passenger in the market would be willing to pay. In Figure 4-2 above, it is the point where the downwardsloping line meets the vertical axis, which is around €1,000 as read off the graph in our example. The b term in equation (4.5) is the change in the average airfare that is generated by a change in the PSF. It is the slope of the downward-sloping line in Figure 4.2. The slope is less than zero because the line slopes downward from left to right, which means a higher PSF is associated with a lower average airfare. Let us demonstrate how we find the a and the b terms using our Air France example. Once we know the a and the b terms then we can find the full average airfare function (4.5). If we consider the Air France example once again, we know that at a PSF of 75% the average airfare is €500, and at a PSF of 0.5% (one passenger flies) the average airfare = €1,000. These represent two points on the airfare function. We can write these points as A = (Average Airfare0, Seat Factor0) = (500, 0.75), and B = (Average Airfare1, Seat Factor1) = (1,000, 0.005). Using these two points we can find the slope of the straight average airfare line. The slope is the vertical rise over the horizontal run, which for our example is the difference between the airfares divided by the difference between the corresponding seat factors, which calculates to be:

Chapter 4

82

Slope =

୅୴ୣ୰ୟ୥ୣ ୅୧୰୤ୟ୰ୣభ ି୅୴ୣ୰ୟ୥ୣ ୅୧୰୤ୟ୰ୣబ ୔ୗ୊భ ି୔ୗ୊బ

=

ହ଴଴ିଵ଴଴଴ ଴.଻ହି଴.଴଴ହ

= െ671

This slope estimate means that if the airline wishes to increase the seat factor by 1 percentage point, say from 70% to 71%, then this will require a €6.71 reduction in the average airfare. Knowing the slope and using the two points A and B we can also find the intercept of the average airfare function. We know that the two points A and B given above must satisfy the average airfare function, which means: 500 = a – 671 u 0.75 and 1000 = a – 671 u 0.005 We can rearrange either of these expressions to generate a value for the average airfare function parameter a. This estimate of the reservation airfare is: a = 500 + 0.75 u 671 = 1,003 Combining the information for the estimated value of the slope term, b = -671 and the estimated value of the reservation airfare, a = 1,003 then the average airfare function for the flight is: Average Airfare = 1,003 – 671 u PSF We can find the average airfare that corresponds with a given PSF by simply plugging the PSF into this expression. For example, if the plane is only half-full, so that the PSF is 50%, then the average airfare that corresponds to a half-full plane is found by plugging 0.5 into the estimated average airfare function for the PSF as follows: Average Airfare (PSF = 0.5) = 1,003 – 671 u 0.5 = 667.5 At a PSF of 50% the average airfare expected is €667.5.

4.4.3 Numerical Example Aeroméxico is a FSA based out of Mexico City and is the national flag carrier of Mexico (Aeroméxico 2019). The airline flies to more than 90 destinations mostly in South America but also to North and Central America, the Caribbean, Europe, and Asia. Aeroméxico operates a service from Mexico City (MEX) to Santiago (SCL) on the 21st of March departing at 10:15am and arriving at 9:20pm on a Boeing 787-800 Dreamliner aircraft with 243 seats. Based on prior experience and using historical data, the analysts for Aeroméxico know that a 50% passenger seat factor coincides with an average airfare of US$1,500, and an 80% passenger seat factor coincides with an average airfare of US$1,000. We wish to use these two points to build a straight-line relationship between Aeroméxico average airfares and the seat factor for the MEX-SCL flight. Since the relationship is linear or a straight line then the slope of the straight line is the same for every combination of average airfare and seat factor points. This means that we can compute the slope of the line using our two points for the average airfare and the PSF as follows: Slope =

ଵ,ହ଴଴ିଵ,଴଴଴ ଴.ହି଴.଼

= െ1,666.67

The intercept of the relationship (a) between the average airfare and the passenger seat factor can then be found using the slope estimate and the coordinates for the point {Average Airfare, PSF} = (1500, 0.5) as follows: 1,500 = a – 1,666.67 u 0.5 This expression can be rearranged to obtain the intercept, which is a = 2,333.33. This means that the relationship between the average airfare and the passenger seat factor for the Aeroméxico flight using the estimates of a and b is: Average Airfare = 2,333.33 – 1,666.67 u PSF The average airfare on the Aeroméxico flight from Mexico City to Santiago that we would expect at a PSF of 75% is: Average Airfare = 2,333.33 – 1,666.67 u 0.75 = $1,083.33 What is the average airfare for a full plane? When the PSF is 1 the average airfare is:

Short Run Airline Revenue

83

Average Airfare = 2,333.33 – 1,666.67 u 1 = $666,67 Several of these points are described in Figure 4-3 below, where we have schematically plotted the relationship between the average airfare and the seat factor for the Aeroméxico flight between Mexico City and Santiago. Average Airfare $2,333

$1,500

$1,083

$667

0.5

0.75

1

Seat Factor

Fig. 4-3: Straight Line or Linear Airfare for the Aeroméxico Example

4.5 Estimating the Parameters of the Linear Function In the Aeroméxico example presented in section 4.4 above, we refer to an analyst consulting historical data on the average airfare and the PSF to capture the parameters in the linear average airfare equation. How might this be done if you are an analyst working in yield management for an airline? For every flight, airlines will typically keep historical data within their corporate data warehouse on the average airfare generated from the flight (the passenger revenue generated divided by the number of passengers carried) and the PSF from the flight. If you have many years of this data, it is possible to estimate a relationship between the average airfare and the PSF. It is not just a matter, however, of trying to fit a line through the average airfare and the PSF data. It is more difficult than this because the movement in the average airfare over time will not just reflect the movements in the PSF over time, but it will also reflect the movements in underlying demand, and unit cost. What this means is that when we analyse the movements in the average airfare and the PSF over time, these two variables may not necessarily move in opposite directions as expected. This is because economic growth and higher unit cost may force the airline to increase both the average airfare and the passenger seat factor at the same time, thus giving the impression that the average airfare and the PSF move positively together. To illustrate this point, consider Figure 4-4 below. In Figure 4-4 we present a starting relationship between the average airfare and the PSF as the firm line that is downward sloping, closest to the origin and has the point A located on it. On that line is a PSF and average airfare combination of 75% and $150 as described in the figure at point A. If the airline were to increase the average airfare along this line, this would coincide with a lower PSF. As underlying demand and unit cost increase over time the relationship between the average airfare and the PSF shifts out over time, as indicated by the dashed lines that sit further to the right from the origin. As the average airfare function shifts out over time, the airline will choose a point on a new average airfare/PSF curve. In Figure 4.4, the new point that the airline chooses on the dashed line that is immediately to the right of the firm line is the point {$160, 75%}, indicated by point B in the graph. At point B, the stronger underlying demand and unit cost has the effect of raising the average airfare, without any movement in the PSF. On the second dashed line the airline chooses a point that is {$155, 80%}, indicated by C. At this point the average airfare is lower and the PSF is higher. And for subsequent dashed lines in future years the airline chooses {$165, 82%} at point D and {$155, 90%} at point E. The historical time path of the average airfare/PSF combination is described by the free-hand drawn line. This free-hand line shows that the historical time-path of the average airfare and the PSF indicates that the two variables can both move in the same, positive direction, as is the case between points C and D, or they can move in opposite directions, as is the case between points B and C, and D and E.

Chapter 4

84

Average Airfare

D $165 $160

B C

$155 $150

E

A

75%

80%

82%

90%

Airfare/Seat Factor time path

PSF

Fig. 4-4: Average Airfare and Passenger Seat Factor - Straight Line Shifting Out Over Time To properly identify a negative relationship between the average airfare and the PSF it is necessary to use statistical methods that take into consideration the positive impact that underlying demand and unit cost has on average airfares. In the language of econometricians, the impact of underlying demand and unit cost on airfares must be purged from the data to isolate the relationship between the average airfare and the PSF. The method that I use for this purging is multivariate regression analysis. This is not something that I will be discussing in this book, but it is certainly a very useful technique to identify relationships between multiple aviation variables. While I will not be discussing the technique of multivariate regression analysis in this book, I will be demonstrating how the basic output of multivariate regression analysis can be used to understand important economic theories of air transport. Let us look at a real-life example of the relationship between the average airfare, the PSF, underlying demand and unit cost. In Figure 4.5 below I present the movements between FY06 and FY18 in the average airfare and the PSF for Regional Express (Rex) Airlines, which is a small regional carrier in Australia (Rex Express 2019). Rex Express Average Fare (A$) Fare

$260

PSF

Passenger Seat Factor 69% 68%

$240

67% 66%

$220

65% 64%

$200

63% $180

62% 61%

$160

60% $140

59% FY06 FY07 FY08 FY09 FY10 FY11 FY12 FY13 FY14 FY15 FY16 FY17 FY18

Source: Airline Intelligence and Research Database 2021

Fig. 4-5: Rex Express Airlines Average Airfare Versus the Passenger Seat Factor FY06 to FY18 I have picked Rex Express Airline because its average sector length is reasonably stable over time at around the 380km mark, and so movements in average distance flown does not influence the relationship between the average airfare and

Short Run Airline Revenue

85

the PSF. This allows us to simulate, as closely as possible, the relationship between the average airfare and the PSF on a particular route. We can see in Figure 4-5 that the average airfare increases over time in trend terms (the firm line) and the PSF falls over time in trend terms (the dashed line), with a steep fall post FY11. There are periods over which the two variables move in the same direction, for example FY06 and FY07 (where they move in the same positive direction) and FY09, FY10, FY13, FY14 (where they move in the same negative direction) and there are periods over which they move in opposite directions. The fact that the two variables move in the same direction and in opposite directions tells us that there are more variables influencing the change in the average airfare over time aside from changes in the PSF over time. To account for these potential other variables, I have estimated a multivariate linear regression relationship between average airfares, Australian real GDP Growth (which is used to represent underlying demand) and Rex Express operating cost per passenger as follows: Average Airfaret = 258.4281 – 379.369 u PSFt + 1625.749 u GDP Growtht + 0.7926 u Cost Per Passenger This is the summary output of an estimated multivariate regression equation. As mentioned above, this book will not go into detail about how we obtain the coefficient estimates, whether the regression is estimated correctly, and whether the assumptions that underpin the way the coefficients are estimated are satisfied or not. All I want you to know in this textbook is how you use the equation and how you interpret the coefficients. We can see in this equation that the coefficient attached to the passenger seat factor is negative at -379.369. This is the slope of our downward-sloping lines in Figure 4-4. The coefficients attached to the economic growth and cost per passenger variables are positive, meaning that increases in these variables lead Rex Express Airlines to increase average airfares. These are the variables that cause the average airfare function to shift to the right over time, which are the dashed lines in Figure 4-4. We can make this graph two dimensional by assuming values for GDP growth and cost per passenger and substituting them into our regression equation. At values of GDP growth of 0.0291 and cost per passenger of $205.77 the average airfare equation is: Average Airfaret = 258.4281 – 379.369 u PSFt + 1625.749 u 0.0291 + 0.7926 u 205.77 If we simplify this expression, we obtain: Average Airfaret = 468.83 – 379.369 u PSFt This expression simply says that the average airfare falls in a linear way as the PSF increases. We can use Microsoft Excel to draw this straight-line relationship – refer to Figure 4-6 below. Rex Express Average Airfare (A$) $500 $450 $400 $350 $300 $250 $200 $150 $100 $50 $0

99% 96% 93% 90% 87% 84% 81% 78% 75% 72% 69% 66% 63% 60% 57% 54% 51% 48% 45% 42% 39% 36% 33% 30% 27% 24% 21% 18% 15% 12% 9% 6% 3% 0%

Rex Express Passenger Seat Factor Fig. 4-6: Rex Express Airlines Average Airfare Function in FY18 To build this graph in Microsoft Excel, use the following steps. Step 1, go into a blank worksheet in Excel. In cell A1 type in the title “Passenger Seat Factor”. Step 2, in cell A2 type in the number 0 and in cell A3 type in the formula “=A2+0.01”. Step 3, copy the formula in cell A3 down the column until you arrive at the number 1. In column A you should have a series of numbers from 0 to 1 in steps of 0.01. Step 4, in the column to the right of the passenger seat factor, column B, plug the following formula into cell B2 “= 468.83-379.369*A2”. Step 5, copy this formula in B2 down column B until you arrive at the average airfare at a passenger seat factor of 1. The entries in column B are your average

Chapter 4

86

airfares for different passenger seat factors. Step 6, create a graph of the average airfare function using the column B data as your vertical axis data and the column A data as your horizontal axis data. This is done by clicking on “Insert” then choose the “line chart” option in the “charts” menu. This should give you a figure that looks like Figure 4-6 with a bit of luck. We can see that the average airfare function in Figure 4-6 is downward sloping, starting at a reservation airfare of $468.83 and ending with a 100% seat factor at an average airfare of $89.46. The slope term means that a 1 percentage point higher PSF (for example the seat factor increases from 60% to 61%) requires a $3.79 reduction in the average airfare. By increasing the rate of economic growth and cost per passenger we can see how the downward-sloping line shifts to the right in Figure 4-6. We will leave this for the reader to do as an exercise.

Quiz 4-1. Short Run Passenger Revenue and the PSF 1. Which of the following statements is true in relation to observations about the PSF? (a) The low-cost carrier average PSF is higher than the average regional carrier PSF which is higher than the average full-service PSF. (b) The low-cost carrier average PSF is higher than the average full-service carrier PSF which is higher than the average regional PSF. (c) The regional carrier average PSF is higher than the average low-cost carrier PSF which is higher than the average full-service PSF. (d) The full-service carrier average PSF is higher than the average low-cost carrier PSF which is higher than the average regional carrier PSF. 2. Put forward an argument as to why the low-cost carrier average PSF may be higher than the full-service carrier average PSF. Or if you believe the opposite is true, put forward an argument to support this case. 3. JetBlue flight B6 215 from Boston (BOS) to Dallas Fort Worth (DFW) leaves at 7:39am and arrives at 11:13am. The aircraft that is used is a 150 seat Airbus A320 aircraft. The airline knows that if it fills the flight, it can do so at an average airfare of $120 but if it fills three quarters of the flight the average airfare rises to $145. (a) Use this information to find an expression for a linear or straight-line average airfare function as a function of the seat factor of the form Average Airfare = a + b u PSF. Hint: You will need to find the slope term b and the intercept term, a. Find the slope term first. The slope is found by using the formula: b=

PSFଵ െ PSF଴ Average Airfareଵ െ Average Airfare଴

The intercept is then found using the formulae: a = (Average Airfare1 – b u PSF1) or a = (Average Airfare0 – b u PSF0) (b) Build a graph of the average airfare function in (a) above with average airfare on the vertical axis and passenger seat factor on the horizontal axis. Hint: Try to do this in Microsoft Excel. The steps to do this are as follows. Step 1. In column A of your worksheet, create a series of values 0, 0.005, 0.01, 0.015, 0.02, ……., 1. This should fill-out 200 cells. These are assumed to be given values of the PSF. Step 2. You wish to find values for the average airfare that corresponds to each value of the 200 values for the PSF. Create these values in column B. In cell B1 type-in the formula “=a + b*A1”, where the a and the b terms are based on your answer for (a) above. Copy this down column B for the rest of the cells and you should have a column of values for the PSF in column A and a column of values for the corresponding average airfare in column B. Step 3. Use a line graph to graph the average airfare on the vertical or Y-axis and the PSF on the horizontal or X-axis. This completes the question. 4. When analysing historical data on the average airfare and the PSF for a particular flight, an analyst observes periods over which the PSF and the average airfare increase together. Does this necessarily mean that an increase in the PSF, other things being equal, leads to an increase in the average airfare? Explain your answer.

Short Run Airline Revenue

87

4.6 Short Run Revenue with Linear Average Airfare Function 4.6.1 Numerical Example If the average airfare is described by equation (4.5) or can be approximated by a (linear) equation like (4.5), this means that the short run passenger revenue function is equal to (4.5) times the PSF times the fixed number of seats. We can write this as follows: SR Passenger Revenue = [a + b u PSF] u PSF u S

(4.6)

If we expand the right-hand side of equation (4.6), then we obtain the non-linear expression: SR Passenger Revenue = a u PSF u S + b u PSF2 u S

(4.7)

What does equation (4.7) tell us? If you recall from your high school maths, equation (4.7) says that when there are a fixed number of seats that an airline can fly, passenger revenue will change as the seat factor changes, but it will not change in a straight line. It will change like a concave down parabola. If you do not remember what a concave down parabola looks like, google “concave down parabola” and click on the first image (not the first entry but the first image).25 What a concave down parabola means when it is applied to airline passenger revenue is the following. Suppose an airline listed possible seat factors on a particular flight from 0% up 100% in steps of 0.5%. For each of these possible seat factors, the airline determines the likely outcome for the average airfare and thus passenger revenue. If the airline graphs the relationship between passenger revenue and the seat factor, then it would look like a concave down parabola. This means that if the airline targets a seat factor that exceeds a particular critical level, then the outcome will be lower passenger revenue. To help you further understand the shape of the passenger revenue curve and how it relates to the relationship between the average airfare and the passenger seat factor, consider the Air France example once again. In the Air France example from section 4.4.2, we know that the average airfare function is the following linear function: Average Airfare = 1,003 – 671u PSF To find the short run passenger revenue function, we multiply this average airfare expression by the number of passengers. If you turn back to equation (4.3) you will see that the number of passengers in the short run is just the fixed number of seats multiplied by the PSF. If we multiply the average airfare by the PSF and multiply this by the fixed number of seats, then we obtain the Air France short run passenger revenue function: SR Revenue = 1,003 u PSF u Sത – 671 u PSF2 u Sത We can see that the short run passenger revenue curve consists of two parts (1) 1,003 u PSF u Sത , which keeps going up as the passenger seat factor increases towards 1, and (2) – 671 u PSF2 u Sത , which is a negative number that becomes more negative as the PSF increases towards 1. This negative number comes about because the average airfare falls as the PSF approaches 1. There is a point at which the second negative factor begins to dominate the first positive factor and passenger revenue begins to fall as the seat factor increases. We show this in Figure 4-7 below for the Air France example. I will now show you how we arrive at the number 75% for the seat factor and $500 for the average airfare that generate maximum revenue of $75,000 in Figure 4-7. As indicated above, the Air France short run passenger revenue equation is: SR Passenger Revenue = 1,003 u PSF u Sത – 671 u PSF2 u Sത If you do not like this equation in this form, you can also convert it into a form with Y’s and X’s as follows: Y = 1,003 u X u 200 – 671 u X2 u 200 where Y is passenger revenue, X is the PSF and there are 200 seats on the plane. As it turns out, the number of seats is irrelevant to finding the revenue maximising PSF and average airfare, so we can get rid of the ܵҧ = 200 variable altogether to obtain the simplified expression:

25

If you still need more help with parabolas, there is an excellent website on quadratic functions put together by the University of North Carolina Wilmington that may help you out, http://dl.uncw.edu/digilib/Mathematics/Algebra/mat111hb/PandR/quadratic/quadratic.html.

Chapter 4

88

Average Airfare € 1,100 € 1,000 € 900 € 800 € 700 € 600 € 500 € 400

75%

€ 300

99% 96% 93% 90% 87% 84% 81% 78% 75% 72% 69% 66% 63% 60% 57% 54% 51% 48% 45% 42% 39% 36% 33% 30% 27% 24% 21% 18% 15% 12% 9% 6% 3% 0%

Passenger Seat Factor Revenue € 80,300

$75,000

€ 70,300 € 60,300 € 50,300 € 40,300 € 30,300 € 20,300 € 10,300 € 300 99% 96% 93% 90% 87% 84% 81% 78% 75% 72% 69% 66% 63% 60% 57% 54% 51% 48% 45% 42% 39% 36% 33% 30% 27% 24% 21% 18% 15% 12% 9% 6% 3% 0%

Passenger Seat Factor Fig. 4-7: Short Run Average Airfare and Revenue Functions Y = 1,003 u X – 671 u X2 If you remember back to your year ten maths, the first derivative of a parabola can be used to find the first derivative of this Y function as follows:26 ୢଢ଼ ୢଡ଼

= 1,003 – 2 u 671 u X

To find the maximum turning point we set this first derivative equal to 0: 1,003 – 2 u 671 u X = 0 We then solve this simple equation for X:

26

If you cannot remember back to your year ten mathematics, then the best book I can recommend for differentiating parabolas with an economic context is Chiang and Wainwright 1985, 148-152. If you want something more basic that is without an economics context, then the notes developed by the University of Wisconsin-Madison 2019 are quite good.

Short Run Airline Revenue

X* =

ଵ଴଴ଷ ଶ×଺଻ଵ

89

= 75%

This is the value of X that maximises the Y function. Remember X is the passenger seat factor and so we have solved for PSF* = 75%, which is the PSF that maximises passenger revenue. The average airfare that corresponds with this PSF is found by plugging a PSF of 75% back into the average airfare function as follows: Average Airfare* = 1,003 – 671u 0.75 = $500 This is the average airfare that maximises passenger revenue. The passenger revenue that the airline earns at a PSF of 75% is found by substituting 75% into the short run passenger revenue expression as follows: SR Passenger Revenue* = 1,003 u 0.75 u 200 – 671 u 0.752 u 200 = $75,000 Or alternatively, we simply multiply the passenger revenue maximising average airfare of $500 by the number of passengers at a 75% passenger seat factor which is 0.75 u 200 = 150 passengers, which gives the maximum passenger revenue that can be earned by Air France: SR Passenger Revenue = 500 u 150 = $75,000 If the airline chooses a PSF that is lower or higher than 75% then it will lose passenger revenue. For example, if it chooses a PSF of 85% then its passenger revenue is $73,591, and if it chooses a PSF of 70% then its passenger revenue is $74,698.

4.6.2 A More General Illustration If you are interested, it is also possible to analyse the relationship between the average airfare, the PSF and short run revenue when we do not know the numerical values of a and b in the average airfare function. In this case the short run passenger revenue function is the function given by (4.7). If we differentiate the passenger revenue function (4.7) and set the result equal to zero, we obtain: ୢ ୗୖ ୔ୟୱୱୣ୬୥ୣ୰ ୖୣ୴ୣ୬୳ୣ ୢ ୔ୗ୊

= a + 2 × b × PSF = 0

(4.8)

Solving (4.8) for the PSF yields the PSF that generates maximum passenger revenue: PSF ‫= כ‬

ି௔

(4.9)

ଶ×ୠ

You may vaguely recall this formula as the axis of symmetry formula in secondary school maths. The airfare that maximises passenger revenue is found by substituting (4.9) back into (4.5) as follows: Average Airfare‫ = כ‬a െ

ୠ×ୟ ଶ×ୠ

=

ୟ

(4.10)

ଶ

Equations (4.9) and (4.10) tell us that the average airfare and the PSF that maximise revenue depend on two forces - the reservation average airfare, and the responsiveness of the average airfare to a change in the PSF. Maximum passenger revenue is found by substituting (4.9) into (4.7) as follows: Max. SR Passenger Revenue = -a u

௔ ଶ×ୠ

uS+buቀ

ି௔ ଶ

ቁ uS= ଶ×ୠ

ି௔మ ସ×ୠ

uS

(4.11)

Let us now see how these formulae apply to the Air France example. We know that a = 1,003 and b = -671. If we substitute these parameter values into the formulae (4.9) through to (4.11) then we obtain the revenue maximising PSF, average airfare and level of short run revenue for the Air France example: PSF ‫= כ‬

െܽ െ1,003 = ൎ 75% 2 × ܾ 2 × 671

Average Airfare‫= כ‬

a 1,003 = ൎ $500 2 2

SR Passenger Revenue =

ି௔మ

u S = ସ×ୠ

ିଵ,଴଴ଷమ ିସ×଺଻ଵ

× 200 ൎ $75,000

Chapter 4

90

You should now confirm that these are indeed the figures we obtained for the Air France example.

4.7 Elasticity of the Average Airfare to the Seat Factor – A little more Advanced 4.7.1 Change in Revenue and the Airfare Elasticity Why does short run passenger revenue start to fall when the passenger seat factor is increased beyond 75% in the Air France example of section 4.6? The answer to this question is based on the elasticity of the average airfare to a change in the seat factor. In the analysis to follow I will explain what this is, and how it helps us to understand why passenger revenue could go up or down in response to an airline targeting a higher PSF. Let us start by examining the change in passenger revenue in the short run. Using (4.7) we can write this change in passenger revenue as: 'Passenger Revenue = (Average Airfare u 'PSF + PSF u 'Average Airfare)uSത

(4.12)

Knowing that the average airfare depends negatively on the PSF, we can write (4.12) as: 'Passenger Revenue = 'PSFuAverage Airfareuቀ1 +

ο୅୴ୣ୰ୟ୥ୣ୅୧୰୤ୟ୰ୣ ο୔ୗ୊

×

୔ୗ୊ ୅୴ୣ୰ୟ୥ୣ ୅୧୰୤ୟ୰ୣ

ቁuSത

(4.13)

We can re-write (4.13) in a simplified way by recognising the following elasticity on the right-hand side of (4.13): ο୅୴ୣ୰ୟ୥ୣ୅୧୰୤ୟ୰ୣ ο୔ୗ୊

×

୔ୗ୊ ୅୴ୣ୰ୟ୥ୣ ୅୧୰୤ୟ୰ୣ

= Airfare Elasticity

This is the elasticity of the average airfare to the seat factor. We can use this to re-write (4.13) in words as follows: 'Passenger Revenue = 'PSFuAverage Airfareu(1 + Airfare Elasticity)uSത

(4.14)

Equation (4.14) tells us why passenger revenue goes up or down when the passenger seat factor goes up. Critical to this is the component of the right-hand side in brackets, which is unity plus the airfare elasticity. The airfare elasticity measures the % decrease in the average airfare when there is a 1% increase in the PSF. For example, if the airfare elasticity is -1.5 this means that a 1% increase in the PSF leads to a 1.5% reduction in the average airfare. In this case we say that the average airfare is elastic, or highly responsive, to the PSF. If the airfare elasticity is greater than -1, for example it is -0.5, this means a 1% increase in the seat factor leads to a 0.5% reduction in the average airfare. In this case the airfare is inelastic, or relatively unresponsive, to the seat factor. When 'Passenger Revenue is positive this means that an increase in the PSF leads to an increase in passenger revenue. When 'Passenger Revenue is negative this means that an increase in the PSF leads to a decrease in passenger revenue. To determine if 'Passenger Revenue is positive or negative we turn to each of the components of the righthand side of (4.14). If we do this, the first thing that we identify is that 'PSF > 0, Average Airfare > 0 and Sത > 0. The only component on the right-hand side of (4.14) that can be positive or negative is (1 + Airfare Elasticity). If (1 + Airfare Elasticity) is less than zero, then 'Passenger Revenue < 0 and if (1 + Airfare Elasticity) is greater than zero then ' Passenger Revenue > 0. We can summary this in the following way: If (1 + Airfare Elasticity) < 0 then 'Passenger Revenue < 0 and n PSF Ÿ p Passenger Revenue If (1 + Airfare Elasticity) > 0 then 'Passenger Revenue > 0 and n PSF Ÿ n Passenger Revenue When is (1 + Airfare Elasticity) < 0? The answer to this question is when Airfare Elasticity < -1 (for example the airfare elasticity is -1.5). In this case the airfare is highly elastic to the PSF. This means that an increase in the passenger seat factor leads to a reduction in passenger revenue when the airfare is elastic to the PSF. When is (1 + Airfare Elasticity) > 0? The answer is when Airfare Elasticity > -1 (for example the airfare elasticity is -0.5). This is the case when the airfare is inelastic to the PSF. This means that an increase in the PSF leads to an increase in passenger revenue when the airfare is inelastic to the PSF. The intuition here is that if it takes a considerable reduction in the average airfare to fill the plane when the airline targets a high PSF, or the airfare is elastic to the PSF, then passenger revenue falls in response to a higher PSF. If the seats on the plane can be filled with a relatively minor airfare reduction, then a higher PSF coincides with an increase in passenger revenue.

4.7.2 Calculating the Airfare Elasticity To demonstrate how we calculate the airfare elasticity, let us return to the Air France example once again of section 4.6. To calculate the airfare elasticity, we take the slope of the average airfare function, b, which in the case of the Air France

Short Run Airline Revenue

91

example is -671, multiply this by the PSF and divide the result by the average airfare. If we do this for our Air France example, we obtain the following airfare elasticity: Airfare Elasticity =

ି଺଻ଵ×୔ୗ୊ ୅୴ୣ୰ୟ୥ୣ ୅୧୰୤ୟ୰ୣ

=

ି଺଻ଵ×୔ୗ୊ ଵ଴଴ଷି଺଻ଵ×୔ୟୱୱୣ୬୥ୣ୰ ୗୣୟ୲ ୊ୟୡ୲୭୰

We can see for this elasticity that, as the slope of the airfare function is a constant (-671) or the airfare function is linear, the elasticity varies with the PSF. We draw this in Figure 4-8 below, which shows how the airfare elasticity with respect to the PSF is linked to the curvature of the passenger revenue function. Airfare Elasticity 0.0

-0.5

-1.0

-1.5

-2.0

-2.5 99% 96% 93% 90% 87% 84% 81% 78% 75% 72% 69% 66% 63% 60% 57% 54% 51% 48% 45% 42% 39% 36% 33% 30% 27% 24% 21% 18% 15% 12% 9% 6% 3% 0%

Passenger Seat Factor Revenue € 80,300 € 70,300 € 60,300 € 50,300 € 40,300 € 30,300 € 20,300 € 10,300 € 300 99% 96% 93% 90% 87% 84% 81% 78% 75% 72% 69% 66% 63% 60% 57% 54% 51% 48% 45% 42% 39% 36% 33% 30% 27% 24% 21% 18% 15% 12% 9% 6% 3% 0%

Passenger Seat Factor Fig. 4-8: Airfare Elasticity of Demand and the Revenue Functions For example, at a PSF of 0.7 the airfare elasticity is equal to: Airfare Elasticity =

ି଺଻ଵ×଴.଻ ଵ଴଴ଷି଺଻ଵ×଴.଻

= െ0.88

This result means that a 1% increase in the PSF from 0.7 will lead to a 0.88% decline in the average airfare. Since the airfare elasticity is greater than -1 this PSF increase will result in an increase in passenger revenue. You can see in Figure 4-8 that the airfare elasticity varies as the passenger seat factor increases. In fact, as the PSF increases towards 75% the airfare elasticity falls towards -1 and when the PSF increases beyond 75% the airfare elasticity falls further below -1. At an airfare elasticity that is equal to -1 this is consistent with the top of the revenue function. The intuition for this result is that for low values of the PSF and high values for the average airfare, the airfare

Chapter 4

92

elasticity of the PSF is relatively low in absolute terms. An increase in the PSF leads to a proportionately weaker decrease in the airfare resulting in higher passenger revenue. As the PSF increases to high levels and the average airfare to relatively low levels, the airfare elasticity of the PSF turns elastic. The seat factor increases by proportionately less than the airfare falls resulting in lower passenger revenue.

4.7.3 A Case Study Using Rex Express Airlines Data Let us now use the Rex Express Airlines real life data that we have analysed previously in this chapter to demonstrate some of these concepts. Let us return to the FY18 average airfare function for Rex Express from section 4.5, which is Average Airfare = 468.83 – 379.369 u PSF. We assume that the average number of seats per flight is 34, which is the seat count on the Saab 340 fleet that is used by Rex Express (Rex Fleet 2019). Given this seat count, the passenger revenue function of Rex Express in FY18 is the average airfare expression above multiplied by the PSF times the seat count as follows: Passenger RevenueFY18 = Average AirfareuPSFuSeat Count = 468.83uPSFu34 – 379.369uPSF2u34 This simplifies to: Passenger Revenue = 15,940.22 u PSF – 12,898.55 u PSF2 The elasticity of the average airfare to a change in the passenger seat factor in FY18 is: Airfare EasticityFY18 =

ିଷ଻ଽ.ଷ଺ଽ×୔ୗ୊ ୅୴ୣ୰ୟ୥ୣ ୅୧୰୤ୟ୰ୣ

=

ିଷ଻ଽ.ଷ଺ଽ×୔ୗ୊ ସ଺଼.଼ଷିଷ଻ଽ.ଷ଺ଽ×୔ୗ୊

Let us now draw the airfare elasticity and passenger revenue curves of Rex Express airlines with the help of Microsoft Excel. Refer to Figure 4-9 below. We can see in Figure 4-9 that the point at which the airfare elasticity is -1 (see the top component of the figure) coincides with the top of the passenger revenue function in the bottom figure. The PSF that generates maximum revenue is found by differentiating the revenue function with respect to the PSF, setting the result equal to zero and solving that equation for the PSF. The first derivative of the passenger revenue function with respect to the PSF is: ୢୖୣ୴ୣ୬୳ୣ ୢ௉ௌி

= 15,940.22 – 12,898.55 u 2 u PSF = 0

Solving this equation for the PSF generates the passenger revenue maximising PSF: PSF* =

ଵହ,ଽସ଴.ଶଶ ଵଶ,଼ଽ଼.ହହ×ଶ

= 61.8%

You should be able to see in Figure 4-9 that the uppermost point on the passenger revenue function corresponds with a point on the horizontal axis which is around the 62% mark. The average airfare that is estimated to correspond with this PSF is found by plugging 61.8% back into the average airfare function as follows: Average AirfareFY18 (PSF = 61.87%) = 468.83 – 379.369 u 0.618 = $234.38 Let us now look at what happens if the economy is growing stronger than expected and cost per passenger is higher. You will remember that to derive our FY18 average airfare function in section 4.5 we assumed the Australian economy was growing at 2.91% and the cost per passenger of the airline was $205.77. We now assume that, for example, the economy is growing at 3.5% and the cost per passenger is $220. In this case the average airfare function is: Average Airfare = 258.4281 – 379.369 u PSF + 1625.749 u 0.035 + 0.7926 u 2220 = 489.70 – 379.369 u PSF The impact of these new assumptions is that the average airfare line shifts to the right, as does the revenue function. The shift to the right in the average airfare function is presented in Figure 4-10 below. We can see in Figure 4-10 that the new average airfare function, which is the dashed line, sits further to the right than the original average airfare function. This implies that for each PSF the airline can charge a higher average airfare. For example, at a PSF of 60% the airline can charge $20.87 more per passenger, which is the vertical distance between the firm line and the dashed line in Figure 4-10.

Short Run Airline Revenue

93

Rex Express Airfare Elasticity 0.0 -0.5 -1.0 -1.5 -2.0 -2.5 -3.0 -3.5 -4.0 -4.5 99% 96% 93% 90% 87% 84% 81% 78% 75% 72% 69% 66% 63% 60% 57% 54% 51% 48% 45% 42% 39% 36% 33% 30% 27% 24% 21% 18% 15% 12% 9% 6% 3% 0%

Rex Express Passenger Seat Factor Rex Express Revenue (A$) $6,000 $5,000 $4,000 $3,000 $2,000 $1,000

61.8%

$0

99% 96% 93% 90% 87% 84% 81% 78% 75% 72% 69% 66% 63% 60% 57% 54% 51% 48% 45% 42% 39% 36% 33% 30% 27% 24% 21% 18% 15% 12% 9% 6% 3% 0%

Rex Express Passenger Seat Factor Fig. 4-9: Airfare Elasticity of Demand and the Revenue Functions for Rex Express in FY18 Rex Express Average Airfare (A$) $600 $500

FY18

FY19

$400 $300 $200 $100 $0 99% 96% 93% 90% 87% 84% 81% 78% 75% 72% 69% 66% 63% 60% 57% 54% 51% 48% 45% 42% 39% 36% 33% 30% 27% 24% 21% 18% 15% 12% 9% 6% 3% 0%

Rex Express Passenger Seat Factor Fig. 4-10: Shift in the Rex Express Average Airfare Function to the Right

Chapter 4

94

As indicated in Figure 4-11 below, the Rex airfare elasticity curve shifts to the right, and the passenger seat factor that corresponds with an airfare elasticity of -1 is higher. This means that the PSF that maximises the airline’s revenue is also higher. Rex Express Airfare Elasticity 0.0 -0.5 -1.0 -1.5 -2.0 -2.5 -3.0 -3.5 -4.0 -4.5

FY18

FY19

99% 96% 93% 90% 87% 84% 81% 78% 75% 72% 69% 66% 63% 60% 57% 54% 51% 48% 45% 42% 39% 36% 33% 30% 27% 24% 21% 18% 15% 12% 9% 6% 3% 0%

Rex Express Passenger Seat Factor Rex Express Revenue (A$)

FY18

FY19

$6,000 $5,000 $4,000 $3,000 $2,000 $1,000

61.8%

$0

64.5%

99% 96% 93% 90% 87% 84% 81% 78% 75% 72% 69% 66% 63% 60% 57% 54% 51% 48% 45% 42% 39% 36% 33% 30% 27% 24% 21% 18% 15% 12% 9% 6% 3% 0%

Rex Express Passenger Seat Factor Fig. 4-11: Shift in the Average Airfare Elasticity and Shift to the Right in the Revenue Curve To find the PSF which maximises short run revenue we must first build the new passenger revenue function, which is just the new average airfare multiplied by the PSF multiplied by the fixed number of seats, 34, which gives: Passenger Revenue = 489.7 u PSF – 379.369 u PSF2 We differentiate this new passenger revenue function with respect to the PSF to obtain: ୢ୔ୟୱୱୣ୬୥ୣ୰ ୖୣ୴ୣ୬୳ୣ ୢ୔ୗ୊

= 16,649.8 – 12,898.55 u 2 u PSF = 0

This solves for the PSF as follows: PSF* =

ଵ଺,଺ସଽ.଼ ଶ×ଵଶ,଼ଽ଼.ହହ

= 64.5%

This is the new revenue maximising PSF, which is 3.7% higher than the previous revenue maximising PSF. At this new PSF, the average airfare is: Average Airfare* = 489.78 – 379.369 u 0.645 = $245.0

Short Run Airline Revenue

95

This is the average airfare that maximises revenue. It is higher by $10.67 because of stronger economic growth and higher unit cost. Note that the stronger economy and the higher unit cost have caused both the revenue maximising PSF and the average airfare to rise.

4.8 Constant-Elasticity Average Airfare Function In earlier sections we analysed cases in which there is a straight-line or linear relationship between the average airfare and the PSF. In the linear case a given increase in the PSF has the same dollar impact on the average airfare no matter if the PSF is currently at low levels or high levels. When the average airfare shares a linear relationship with the PSF a key outcome is that passenger revenue can fall if the passenger seat factor is increased beyond a certain level. This is because the airfare elasticity to the passenger seat factor varies as the PSF is increased and turns from inelastic to elastic beyond a certain critical PSF. An alternative relationship between the average airfare and the PSF can give rise to a model that says passenger revenue never falls if the passenger seat factor is increased. This relationship is called a Cobb-Douglas relationship or a constant-elasticity average airfare function.27 In the passenger revenue model where there is a Cobb-Douglas relationship between the average airfare and the PSF, passenger revenue continues to rise as the PSF increases but it increases at a decreasing rate. Let us now show why this is the case. In the airfare function that shares a Cobb-Douglas relationship with the PSF, the average airfare depends on the PSF raised to a certain power as follows: Average Airfare = A u PSF-H

(4.15)

The power term H is an important parameter in this function. It represents the elasticity of the average airfare to a change in the PSF. It measures the % decrease in the average airfare in response to a 1% increase in the PSF. Most importantly, this number does not depend on the level of the PSF. It is a constant. It is this property that prohibits revenue from falling if the PSF is set too high. This is different from the situation in which the average airfare is a linear function of the PSF, in which case the elasticity of the average airfare to the PSF changes from being relatively inelastic to being relatively elastic, thus causing a turning point in the passenger revenue function. The passenger revenue function in the case where the average airfare function is a Cobb-Douglas function of the PSF, is just the average airfare function at (4.15) multiplied by the PSF multiplied by the fixed number of seats. We write this as follows: SR Passenger Revenue = A u PSFଵିக u S

(4.16)

We will now describe what the short run passenger revenue function (4.16) looks like when we graph it with revenue on the vertical axis and the PSF on the horizontal axis. To demonstrate the passenger revenue graph, we will apply equations (4.15) and (4.16) to the Air France example. We further assume in this example that the elasticity of the average airfare to the PSF, -H is equal to -0.5, meaning that the average airfare falls by 0.5% when the PSF is increased by 1%. We also continue to assume that a PSF of 75% gives rise to an average airfare of €500. With these two pieces of information, we can find the value of A in equation (4.16). We do this by substituting into (4.16) an average airfare that is €500, a PSF that is 0.75 and an airfare elasticity that is -H = -0.5. This yields the following expression: 500 = A u 0.75-0.5

(4.17)

If we now re-arrange (4.17) for A we obtain: A = 500 u 0.750.5 = 433 Substituting A = 433 and H = 0.5 into (4.15) yields the average airfare relationship with numerical values for the parameters: Average Airfare = 433 u PSF-0.5

(4.18)

The short run passenger revenue expression at (4.16) is found by multiplying (4.18) by the PSF times the fixed number of seats. This gives: SR Passenger Revenue = 433 u PSF0.5 u 200 = 86,600 u PSF0.5

(4.19)

27 The Cobb-Douglas function was originally designed to examine the relationship between labour, capital, and output in the production process. See Sandelin 1976, 117-123 if you are interested in investigating the origins of the function.

Chapter 4

96

We can now graph both the average airfare function and the short run passenger revenue function with the help of Microsoft Excel. See Figure 4.12 below. Average Airfare (EUR) $5,000 $4,500 $4,000 $3,500 $3,000 $2,500 $2,000 $1,500 $1,000 $500 $0 100% 97% 94% 91% 88% 85% 82% 79% 76% 73% 70% 67% 64% 61% 58% 55% 52% 49% 46% 43% 40% 37% 34% 31% 28% 25% 22% 19% 16% 13% 10% 7% 4% 1%

Seat Factor

Revenue (EUR) $100,000 $90,000 $80,000 $70,000 $60,000 $50,000 $40,000 $30,000 $20,000 $10,000 $0 99% 96% 93% 90% 87% 84% 81% 78% 75% 72% 69% 66% 63% 60% 57% 54% 51% 48% 45% 42% 39% 36% 33% 30% 27% 24% 21% 18% 15% 12% 9% 6% 3% 0%

Seat Factor Fig. 4-12: Airfare and Revenue Functions Cobb Douglas Airfare Function We can see in Figure 4.12 that the key difference between the constant-elasticity and the linear relationship between the average airfare and the PSF is that the average airfare expression is no longer a straight line but bows in towards the origin (the upper component of Figure 4-12). We also note that the average airfare function never touches the vertical and horizontal axes, although it gets close to them. We also see in the bottom segment of the graph that the passenger revenue function increases as the PSF increases but never decreases. It never reaches a maximum turning point. This is because the elasticity of the average airfare to the PSF never falls below -1. In this case it is always greater than -1 and so passenger revenue never falls as the PSF is increased.

Quiz 4-2. Short Run Passenger Revenue and the PSF 1. Let us re-examine the JetBlue flight B6 215 from Boston (BOS) to Dallas Fort Worth (DFW) from Quiz 4.1. The details are repeated below for your convenience. The aircraft that is used is a 150 seat Airbus A320 aircraft. The airline knows that if it fills the flight that it can do so at an average airfare of $120 but if it fills three quarters of the flight the average airfare rises to $145. (a) Find the short run passenger revenue function for JetBlue flight B6 215. You may wish to use your answer for the average airfare function from Quiz 4-1. (b) Create a graph of the short run passenger revenue function, with passenger revenue on the vertical axis and the PSF on the horizontal axis. Hint: create a new column in your Microsoft Excel worksheet from Quiz 4.1, say column C. In column C enter a formula that multiplies the PSF by the average airfare times the number of passenger seats which is 150. For example, in cell C1 enter the formula =A1*B1*150.

Short Run Airline Revenue

97

(c) Identify the point on the short run passenger revenue function where short run passenger revenue is maximised. Find the PSF and the average airfare that corresponds with the maximum point. 2. If the airfare elasticity with respect to the PSF is -1.5, how is this interpreted? (a) A 1% increase in the PSF causes a 1.5% increase in the average airfare. (b) A 1.5% increase in the PSF causes a 1% decrease in the average airfare. (c) A 1.5% increase in the PSF causes a 1% increase in the average airfare. (d) A 1% increase in the PSF causes a 1.5% decrease in the average airfare. 3. Which of the following is true about the relationship between passenger revenue and the airfare elasticity of the PSF. (a) An increase in the PSF will lead to an increase in passenger revenue if the average airfare is elastic to the PSF. (b) An increase in the PSF will lead to a decrease in passenger revenue if the average airfare is inelastic to the PSF. (c) An increase in the PSF will lead to a decrease in passenger revenue if the average airfare is elastic to the PSF. (d) A decrease in the PSF will lead to a decrease in passenger revenue if the average airfare is elastic to the PSF. 4. Which of the following is true about the linear and Cobb-Douglas average airfare functions and the passenger revenue function? (a) The passenger revenue function is a concave down parabola if the average airfare function is a downward sloping linear function of the PSF. (b) The passenger revenue function is a convex parabola if the average airfare function is a downward sloping linear function of the PSF. (c) The passenger revenue function is a concave parabola that never slopes downward if the average airfare function is a Cobb-Douglas function of the PSF, and the elasticity of the average airfare to the PSF is inelastic. (d) The passenger revenue function is a convex parabola if the average airfare function is a Cobb-Douglas function of the PSF and the elasticity of the average airfare to the PSF is elastic. A Cobb-Douglas relationship between the average airfare and the PSF is estimated to be Average Airfare = 500 u PSF. Use this information to answer the following four questions.

0.75

5. What is the elasticity of the average airfare to the PSF? (a) 500 (b) -0.75 (c) 1/-075 (d) 500—0.75 6. What is the average airfare at a PSF of 80%? (a) 500 u (0.8)-0.75 (b) 500 u (80)-0.75 (c) 500 x (0.8)0.75 (d) 500 x (80)0.75 7. What is the passenger revenue function for this average airfare function? (a) Passenger Revenue = 500 u PSF0.75 (b) Passenger Revenue = 500 u PSF0.25 u Seats (c) Passenger Revenue = 500 u PSF-0.25 u Seats (d) Passenger Revenue = 500 u PSF1.25 u Seats 8. Which of the following statements is true about this average airfare function? (a) It gives rise to a passenger revenue function that is decreasing in the PSF. (b) The elasticity of the average airfare to the PSF becomes more inelastic as the PSF increases. (c) The elasticity of the average airfare to the PSF becomes more elastic as the PSF increases. (d) The elasticity of the average airfare to the PSF is invariant to the PSF. 9. An airline economist estimates the following relationship between the average airfare for Turkish Airlines in some quarter t and the PSF in quarter t amongst other variables: Average Airfaret = 165.9 – 13.6uMarch – 8.8uJune + 4.0uSeptember + 0.651uFuel Pricet + 0.168uFuel Pricet-5 – 67.21uPSFt where Average Airfaret = Turkish Airlines revenue per passenger in US dollars in quarter t, March = 1 when the quarter in question is March and 0 otherwise, June = 1 when the quarter in question is June and 0 otherwise, September = 1 when the quarter in question is September and 0 otherwise and Fuel Price = spot price of jet fuel in US dollars per barrel.

98

Chapter 4

(a) Simplify this multivariate relationship so that it is bivariate by assuming March = 0, June = 0, September = 0, Fuel Pricet = 90 and Fuel Pricet-5 = 59. (b) Graph the simplified bivariate relationship between the average airfare and the PSF. (c) Find the average airfare at PSFs of 75% and 85%. (d) Derive the revenue function assuming an average aircraft seat count of 186. Graph the revenue function. What is the estimated revenue at PSFs of 75% and 85%? (e) Find the PSF and average airfare that maximises revenue. What is the maximum revenue? (f) Now assume March = 1, with all other variables as before. Find the average airfare function, the revenue function and the PSF that maximises revenue. Compare these results to the results in (a) through to (e). (g) Now, re-assume that March = 0 and further assume that Fuel Pricet = 100 and Fuel Pricet-5 = 69. Find the average airfare and revenue functions and the revenue maximising PSF. What happens to optimal revenue in response to the increase in jet fuel prices?

CHAPTER 5 MEDIUM RUN AIRLINE REVENUE

When an airline expands capacity, this is positive for revenue because it means selling more tickets to passengers, but at the same time it results in lower passenger yields, which is negative for revenue. The outcome for an airline’s revenue when it expands capacity therefore depends on the balance of these unfavourable price and favourable volume effects. While most airlines understand the favourable volume effect because they have a good idea about how many tickets they are likely to sell if they operate more seats, they have less idea about the unfavourable price effect. It is critical that airlines understand this unfavourable price effect, because if this effect is stronger than the favourable volume effect then revenue will fall if the airline decides to expand capacity. As the airline’s costs unambiguously go up when they expand capacity, as we will learn in chapter 6 shortly, a situation in which revenue falls in response to an increase in capacity means that profit must unambiguously fall. For this reason, it is imperative that airlines understand the circumstances in which revenue might fall if they expand capacity. Revenue will fall in response to an increase in capacity if yields fall by proportionately more than capacity increases. In other words, revenue falls when yields are sensitive to capacity. Yields are sensitive to capacity when it is difficult to fill extra seats by lowering airfares. While it is relatively easy to stimulate price-sensitive leisure travellers to travel by offering them lower airfares, it is not as easy to stimulate business-purpose travellers in the same way. It follows that it is more likely that revenue will fall in response to an increase in capacity if the dominant passenger mix of the airline is business-purpose. Revenue is also more likely to fall in response to an airline increasing capacity if the airline faces aggressive, high-quality competition, and this competition reacts to an increase in capacity by expanding their own capacity. This will be the case when the competition does not want to concede market share, which is true of most airline markets. There are many airline groups around the world that operate both full-service and low-cost brands, such as the Qantas Group, Singapore Airlines Group, International Airlines Group, Air France/KLM Group and Lufthansa Group. The impact of capacity expansion on yield and revenue is more complicated for these airline groups because they potentially face cannibalisation effects. This occurs when an airline group expands the capacity of the low-cost airline within its group, and this leads to a reduction in the yield of the full-service carrier within the group. This chapter may be challenging for some readers because the relationship between revenue and capacity is complex, which means it needs to be treated with slightly more complex mathematics and analytics than chapter 4. To meet this chapter head-on you will need to have your thinking cap on. The relationship between yields and capacity is probably the least understood in the airline business, but it is also the most important. If you understand this relationship, then you will have a skill set that is highly valued by aviation management.

5.1 Medium Run Capacity Over the medium run the airline has enough time to alter the capacity that it offers to the market. It is usually defined as a period that is more than 12 months but no longer than 2 to 3 years. In the case of passenger aircraft, capacity is altered by changing the ASKs that are offered. ASKs can be altered in a variety of different ways, including altering the frequency of services, altering the size of the aircraft or the seat count (also referred to as the aircraft gauge), or increasing the average distance that is flown by flying more services on longer routes. The airline can also change freight capacity, AFTKs, and total capacity, ATKs, over the medium run. Freight and total capacity can be altered in the same way as passenger capacity, that is, changing the frequency of services, the average distance travelled and by altering the size of the aircraft. When an airline changes capacity in the medium run it also changes revenue. An increase in capacity means that the airline can carry more passengers and freight. By selling more services to passengers and freight this leads to an increase in revenue. At the same time, however, the airline must reduce the price it charges passengers and freight if it wishes to fill the extra capacity, other things being equal. This reduction in price dampens revenue. The ultimate impact on revenue of an increase in capacity depends on the extent to which it sells more seats and aircraft space to passengers and freight, which raises revenue, and the negative impact of a lower price, which reduces revenue. This is summarised in Figure 51 below. In Figure 5-1 we see that there are two forces impacting revenue in response to an increase in own capacity – a volume effect and a price effect. The volume effect says that when an airline increases its capacity, as indicated by the n Own Capacity box, it will fly more passengers and freight, which leads to an increase in revenue. This is the left-hand branch of Figure 5-1 starting from the box in the top right-hand corner and is referred to as the volume effect. The second force says that when an airline increases its own capacity this leads to a decrease in passenger and freight yields. This is

Chapter 5

100

n Volume of PAX and FRT

Volume Effect

n Own Capacity Own Yield Elasticity & Price Effect p Yield

'Revenue

Fig. 5-1: Impact on Medium Run Revenue of an Increase in Capacity the branch that extends vertically down from the box in the top right-hand corner of Figure 5-1. A reduction in passenger and freight yields lead to lower medium run revenue and is referred to as the price effect. The extent to which passenger and freight yields decline in response to an increase in own capacity is determined by the Own Yield Elasticity as labelled in Figure 5-1. The reduction in revenue due to the lower passenger and freight yield is balanced against the increase in revenue due to the higher passenger and freight volumes, with the net outcome for revenue depending critically on the extent to which passenger and freight yields fall in response to a given increase in capacity.

5.2 Medium Run Revenue Analytics There is a simple identity that we can use to understand the movement in medium run revenue in response to a change in capacity. To make the explanation as simple as possible, we will concentrate on the passenger side of the airline business. The same principles also hold in the case of the freight side of the airline business. Let us first define medium run passenger revenue analytically. On a particular route, passenger revenue is the product of the number of passengers carried and the average airfare as follows: Passenger Revenue = Passengers Carried u where the component

௉௔௦௦௘௡௚௘௥ ோ௘௩௘௡௨௘ ௉௔௦௦௘௡௚௘௥ ஼௔௥௥௜௘ௗ

୔ୟୱୱୣ୬୥ୣ୰ ୖୣ୴ୣ୬୳ୣ

(5.1)

୔ୟୱୱୣ୬୥ୣ୰ େୟ୰୰୧ୣୢ

on the right-hand side of (5.1) is simply the average airfare. If we multiply and

divide the right-hand side of (5.1) by the passenger average sector length (PASL) we obtain: Passenger Revenue = Passengers Carried u PASL u

୔ୟୱୱୣ୬୥ୣ୰ ୖୣ୴ୣ୬୳ୣ ୔ୟୱୱୣ୬୥ୣ୰ୱ େୟ୰୰୧ୣୢ×୔୅ୗ୐

(5.2)

Note that in moving from (5.1) to (5.2) we have not changed anything – we have just divided the numerator and denominator by the same variable, PASL. Passengers carried times the PASL is just revenue passenger kilometres. This means we can write passenger revenue at (5.2) as follows: Passenger Revenue = Revenue Passenger Kilometres u

୔ୟୱୱୣ୬୥ୣ୰ ୖୣ୴ୣ୬୳ୣ ୖୣ୴ୣ୬୳ୣ ୔ୟୱୱୣ୬୥ୣ୰ ୏୧୪୭୫ୣ୲୰ୣୱ

(5.3)

If we now multiply and divide the right-hand side of (5.3) by the PSF, we can re-write medium run passenger revenue as a function of capacity and passenger revenue per ASK (PRASK) as follows: Passenger Revenue = Available Seat Kilometres u

୔ୟୱୱୣ୬୥ୣ୰ ୖୣ୴ୣ୬୳ୣ ୅୴ୟ୧୪ୟୠ୪ୣ ୗୣୟ୲ ୏୧୪୭୫ୣ୲୰ୣ

(5.4)

It is (5.4) that we will use to analyse the medium run relationship between passenger revenue and airline capacity. We can simplify (5.4) in the following way: Passenger Revenue = Available Seat Kilometres u PRASK

(5.5)

Equation (5.5) is an identity that simply says that medium run passenger revenue is the number of ASKs supplied by the airline multiplied by its PRASK. To model passenger revenue, the next step is to model PRASK on the right-hand side of (5.5) as a function of ASKs. This is the task of the next section.

Medium Run Airline Revenue

101

5.3 Modelling PRASK and Medium Run Revenue 5.3.1 Linear PRASK 5.3.1.1 Model Analytics The key to understanding the impact of an increase in ASKs on the medium run passenger revenue of an airline is in understanding the impact of ASKs on PRASK. In this section we assume that there is a linear relationship between ASK and PRASK that takes the following form: PRASK = a0 + a1 u ASK

(5.6)

The slope term a1 is the key parameter. It is less than zero, meaning that when an airline increases its own passenger capacity this leads to a reduction in PRASK. The reduction in PRASK can come about in one of two ways - the seat factor can fall, and/or passenger fares can fall. The interpretation of the coefficient a1 is as follows: For every 1 unit increase in own airline ASKs there is an a1 unit reduction in the own PRASK of the airline.

Notice in this interpretation that we use the word unit. This is because the interpretation of the size of a1 depends on the unit of measurement of ASKs (which is usually in terms of millions or thousands of ASKs) and the unit of measurement of PRASK (which could be in dollars or cents or another currency unit such as the Thai Bhat, the Japanese Yen, the Russian Ruble, The Turkish Lira, or the Norwegian Krone). The parameter a0 is the level of PRASK that is determined independently of ASKs and is mostly driven by the forces that drive underlying demand for passenger travel as discussed in chapter 3. Substituting equation (5.6) into equation (5.5) yields an expression for medium run revenue of the airline when PRASK is a straight-line function of ASKs: RMR = PRASK u ASK = a0 u ASK + a1 u ASK2

(5.7)

The medium revenue function in this case is a concave down parabola, which we have met before in chapter 4 in the case of the short run passenger revenue function. It is concave down because the a1 term is less than zero, and this is less than zero because PRASK falls when ASKs increase. The concave down medium run revenue function is illustrated below in the Figure 5-2 schematic. RMR

RMR = a0 u ASK + a1 u ASK2

0

ASK*

ASK

Fig. 5-2: Impact on Medium Run Revenue of an Increase in Capacity We can see in Figure 5-2 that revenue is zero when ASKs are zero because zero capacity means the airline does not sell any tickets to passengers and therefore derives zero revenue. As ASKs increase, initially this leads to an increase in revenue because the volume effect outweighs the price effect. When the airline tries to increase ASKs beyond ASK* however, revenue starts to fall. This is because yields are pushed too low by the higher capacity, resulting in the negative yield effect dominating the positive volume effect. 5.3.1.2 Airline Illustration Let us now demonstrate (5.6) and (5.7) using a numerical example. Qantas domestic PRASK over an annual period is represented by the following linear function of its own ASKs:

Chapter 5

102

PRASK = 31.5 - 0.0004 u ASK where PRASK is measured in terms of Australian cents per available seat kilometre and ASK is measured in terms of millions of annual available seat kilometres. The slope parameter a1 = -0.0004 is interpreted in the following way: If Qantas domestic available seat kilometres increase by 1b over the space of a year (or ASKs increase by 1,000) then Qantas domestic PRASK is expected to fall by 1,000 u 0.0004 = 0.4 cents per ASK

If we multiply the Qantas domestic PRASK expression by Qantas domestic ASKs we obtain the following medium run revenue function for Qantas in the domestic Australian market: RMR = PRASK u ASK = 31.5 u ASK - 0.0004 u ASK2 In this case RMR is measured in terms of Australian cents. To convert medium run revenue into Australian dollars we would need to divide this expression by 100. We draw a graph of this concave down parabola in Figure 5-3 below along with the linear PRASK function in the upper part of the figure. Qantas Domestic PRASK (c/ASK) 20 19 18 17 16 15 14 13 45,000

44,000

43,000

42,000

41,000

40,000

39,000

38,000

37,000

36,000

35,000

34,000

33,000

32,000

31,000

30,000

Annual Qantas Domestic ASK (m) Qantas Domestic Revenue (A$ m) $6,250 $6,200 $6,150 $6,100 $6,050 $6,000 $5,950 $5,900 $5,850 $5,800 45,000

44,000

43,000

42,000

41,000

40,000

39,000

38,000

37,000

36,000

35,000

34,000

33,000

32,000

31,000

30,000

Annual Qantas Domestic ASK (m) Fig. 5-3: Qantas Domestic Medium Run PRASK and Revenue Functions We can see that the linear PRASK function in the upper part of Figure 5-3 is downward sloping and linear, which in turn gives rise to the concave down parabola representing medium run Qantas domestic revenue in the bottom half of Figure 5-3. The revenue function reaches a maximum at around 40b ASKs per annum. Before the maximum, revenue goes up as ASKs are increased, and after the maximum, revenue falls as ASKs are increased.

5.3.2 Cobb-Douglas PRASK 5.3.2.1 Model Analytics Another way to represent PRASK as a function of ASKs is via the use of a Cobb-Douglas PRASK function. Another name for this type of function is the double logarithmic function or the constant-elasticity function. It is called a double

Medium Run Airline Revenue

103

logarithmic function because it is converted from a Cobb-Douglas function to a linear function when we take the logarithm of both sides.28 The Cobb-Douglas PRASK function takes the form: PRASK = A u ASKb

(5.8)

In this case the PRASK function is no longer a straight line, although it is still downward sloping because we assume the parameter b is less than zero. This parameter must be less than zero if PRASK is to fall when ASKs are increased. The shape of the Cobb-Douglas PRASK function is described schematically below in Figure 5-4. PRASK

PRASK = A u ASKb 0

ASK

Fig. 5-4: Cobb-Douglas PRASK Function We can see that the Cobb-Douglas PRASK function is downward sloping in Figure 5-4, but it bends in towards the origin. The fact that the PRASK function bends in towards the origin means that it is not linear. The parameter b in (5.8) has an important interpretation in the Cobb-Douglas PRASK function. This interpretation is as follows: A 10% increase in own ASKs leads to a bu10% decrease in PRASK.

Notice that we do not use the units of measurement of ASKs and PRASK in this definition. The only measure that we discuss is the percentage change. Because of this, we interpret b as an elasticity. In fact, it is called the own yield elasticity, or the elasticity of own yield to own ASKs. We use the word own here because we are measuring the response of one airline’s yield to its own ASKs. In later sections in this chapter, we will examine how the yield responds to the ASKs of other, competing airlines in which case we determine a Cross Yield Elasticity (or an elasticity that is not an own elasticity). The medium run revenue function in the case of the Cobb-Douglas PRASK function is found by multiplying (5.8) by ASKs, which gives: RMR = A u ASK1+b

(5.9)

You will notice that when we multiply (5.8) by ASKs to get the medium run revenue function all that we have done is add 1 to the exponent of ASKs in the PRASK function, so that the exponent attached to ASKs is now 1+b. The most important difference between the medium run revenue function (5.9) in the case of the Cobb-Douglas PRASK function, and the medium run revenue function generated by the linear PRASK function at (5.7), is that in the case of the CobbDouglas medium run revenue function, revenue never actually falls as ASKs increase - revenue will slow down as ASKs increase but it will never actually fall. This can be seen in the Figure 5-5 schematic diagram below.

28

The natural logarithm of both sides of (5.8) is loge PRASK = loge A + b Loge ASK. We can write this as prask = a + b × ask, where the lower-case variables represent the natural logarithm of the variable. If you want to learn more about logarithms or need a refresher, see Chiang and Wainwright 1985, 22-23 for an excellent summary. If this is a little too advanced or you need more detail go to the mathsisfun 2019 website.

Chapter 5

104

RMR RMR = A u ASK1+b

ASK

0 Fig. 5-5: Medium Run Function for Cobb-Douglas PRASK Function 5.3.2.2 Airline Illustration

We now assume that the Qantas domestic PRASK function is a Cobb-Douglas function with an own elasticity of yield to ASKs of b = -0.75 and an A parameter equal to 45,000 as follows: PRASK = 45,000 u ASK-0.75 where PRASK is defined in Australian cents and ASKs are defined in millions of ASKs per year. The own yield elasticity b = -0.75 means that a 1% increase in Qantas domestic ASKs leads to a 0.75% reduction in Qantas domestic PRASK. At annual Qantas domestic ASKs of 35b per year the PRASK of the airline is found by substituting ASKs = 35,000 into the Cobb-Douglas PRASK function as follows: PRASK = 45,000 u (35,000)-0.75 = 17.59 Australian cents per ASK We substitute ASK = 35,000 into this function rather than ASK = 35 because ASKs are measured in millions and 35b available seat kilometres represents ASK = 35,000 million. In this case the medium run revenue function is this PRASK expression multiplied by the ASKs of Qantas domestic, which is: RMR = 45,000 u ASK1-0.75 = 45,000 u ASK0.25 We can see in this RMR function that as ASKs increase revenue increases because the exponent term (which is +0.25) is positive. Because the exponent is a fraction, however, it means that revenue goes up at an ever-decreasing rate as ASKs increase, but never reaches a maximum. Readers should now try and use Microsoft Excel to draw a graph of this medium run Qantas domestic revenue function, linking it to the Qantas domestic PRASK function as we have done in Figure 53.

Quiz 5-1. Linear and Cobb-Douglas PRASK 1. (a) (b) (c) (d)

Which of the following is a key characteristic of the linear PRASK function? The coefficient attached to ASKs varies with the level of ASKs. The coefficient attached to ASKs is fixed invariant to the level of ASKs. PRASK is independent of ASKs. The elasticity of PRASK to ASKs is fixed.

The following linear PRASK function is estimated for China Eastern domestic services PRASK = 0.812 – 0.00035uASK, where PRASK is denominated in Chinese Yuan and ASK is denominated in millions of ASKs. Use this information to answer the following two questions. 2. (a) (b) (c) (d)

What is the interpretation of the coefficient attached to ASKs in this equation? A 1 million increase in China Eastern ASKs leads to a reduction in PRASK in Chinese Yuan of 0.35. A 1 billion increase in China Eastern ASKs leads to a reduction in PRASK in Chinese Yuan of 0.35. A 10 billion increase in China Eastern ASKs leads to a reduction in PRASK in Chinese Yuan of 0.35. A 100 billion increase in China Eastern ASKs leads to a reduction in PRASK in Chinese Yuan of 0.35.

Medium Run Airline Revenue

3. (a) (b) (c) (d)

105

What is the medium run revenue function of China Eastern domestic? RMR = 0.812 – 0.00035uASK2 RMR = 0.812uASK – 0.00035uASK RMR = 0.812uASK – 0.00035uASK2 RMR = 0.812uASK2 – 0.00035uASK2

The following linear PRASK function is estimated for Wizz Air PRASK = 12.54 – 0.000891uASK, where PRASK is denominated in Euro cents and ASK is denominated in millions. 4. (a) (b) (c) (d)

What is the interpretation of the coefficient attached to ASKs in this equation? A 1 million increase in Wizz Air ASKs leads to a reduction in PRASK in Euro cents of 0.891. A 10 million increase in Wizz Air ASKs leads to a reduction in PRASK in Euro dollars of 0.891. A 1 billion increase in Wizz Air ASKs leads to a reduction in PRASK in Euro cents of 0.891. A 100 billion increase in Wizz Air ASKs leads to a reduction in PRASK in Euro cents of 0.891.

5. (a) (b) (c) (d)

What is the medium run revenue function of Wizz Airlines in Euro cents? RMR = 12.54 – 0.000891uASK2 RMR = 12.54uASK – 0.000891uASK RMR = 12.54uASK2 – 0.000891uASK2 RMR = 12.54uASK – 0.000891uASK2

6.

The following graph presents PRASK as a linear function of ASKs. Use the points A, B and C in the graph to determine the slope and intercept of the PRASK function. PRASK C

B PRASK = a + buASK 0 7.

A

ASK

Use the information in questions 2 and 3 to graph the PRASK and medium run revenue functions of China Eastern domestic.

Aer Lingus PRASK is described by the following Cobb-Douglas function PRASK = 1,000 u ASK-0.45 where PRASK is measured in Euro cents and ASK is measured in millions per year. Use this information to answer the following three questions. 8. (a) (b) (c) (d)

What is the PRASK of Aer Lingus in Euro cents at ASKs of 30b per annum? PRASK =1,000 u 30,000-0.45 PRASK =1,000 u 3,000-0.45 PRASK =1,000 u 30-0.45 PRASK =1,000 u 3-0.45

9. (a) (b) (c) (d)

What is the interpretation of the yield elasticity? A 10% increase in ASKs leads to a 4.5% increase in PRASK. A 1% increase in ASKs leads to a 4.5% increase in PRASK. A 1% increase in ASKs leads to a 4.5% decrease in PRASK. A 10% increase in ASKs leads to a 4.5% decrease in PRASK.

10. (a) (b) (c) (d)

What is the medium run revenue function of Aer Lingus in Euro cents? RMR = 1,000 u ASK-0.45 RMR = 1,000 u ASK-1.45 RMR = 1,000 u ASK0.55 RMR = 1,000 u ASK0.45

Chapter 5

106

11. What is the difference between the linear and the Cobb-Douglas PRASK functions? (a) The yield elasticity varies with ASKs in the case of the linear PRASK function whereas it is fixed, independent of ASKs in the case of the Cobb-Douglas PRASK function. (b) The revenue function in the case of the linear PRASK function can go down when ASKs increase but the revenue function in the case of the Cobb-Douglas PRASK function never goes down when ASKs are increased if yield is inelastic to ASKs. (c) The interpretation of the coefficient attached to ASK in the linear PRASK function depends on the units of measurement of ASKs and PRASK while the interpretation of the yield elasticity in the Cobb-Douglas PRASK function does not depend on the units of measurement of ASKs and PRASK. (d) All the above.

5.4 Deeper Medium Run Revenue Analytics29 5.4.1 General Analytics Let us now examine what happens to passenger revenue when there is a change in ASKs for a general PRASK function. A general PRASK function means that we do not provide a specific functional form for the PRASK function, such as a linear functional form (like we use in section 5.3.1) or a constant-elasticity functional form (like we use in section 5.3.2). The general PRASK function says that PRASK is functionally related to capacity, but it does not specify the exact way that capacity affects PRASK. We can write the general PRASK function as follows: PRASK = f(ASK) where the “f” denotes the fact that PRASK is a function of ASKs. Because we wish to find the derivative of the “f” function the only condition we place on it is that it is smooth and continuous. The first thing we wish to do in a deeper analytical treatment of the relationship between medium run revenue and ASKs is to find an expression for the change in passenger revenue. This expression will enable us to examine all the forces that may cause passenger revenue to change. The change in passenger revenue expression is found by finding the total differential of our medium run revenue equation (5.5). We can write the total differential of (5.5) in the following way: οPassenger Revenue = ASK × οPRASK + PRASK × οASK

(5.10)

Equation (5.10) just says that passenger revenue can change because of a change in PRASK and because of a change in own ASKs. The first component of the right-hand side of equation (5.10), ‫ × ܭܵܣ‬οܴܲ‫ܭܵܣ‬, describes the price effect in Figure 5-1 while the second component of the right-hand side of equation (5.10), ܴܲ‫ × ܭܵܣ‬ο‫ܭܵܣ‬, describes the volume effect in Figure 5-1. We know from the general function for PRASK, PRASK = f(ASK), that the change in PRASK on the right-hand side of (5.10) depends on the change in ASK as follows: 'PRASK =

ο୔ୖ୅ୗ୏ ο୅ୗ୏

× οASK

ο௉ோ஺ௌ௄

ௗ௙

The expression is just the slope of the general PRASK function, which we can also write as for very small ο஺ௌ௄ ௗ஺ௌ௄ changes in ASKs. If the PRASK function were a linear function of ASKs, then this slope would be a constant term. If the PRASK function were a Cobb-Douglas function, then the slope would depend on the level of ASKs because the slope of the Cobb-Douglas function changes as you move around the curve or as you increase ASKs. Substituting the change in PRASK into (5.10) yields an expression for the change in passenger revenue as a function of the change in ASKs only: οPassenger Revenue = ASK ×

ο୔ୖ୅ୗ୏ ο୅ୗ୏

× οASK + PRASK × οASK

(5.11)

We can simplify (5.11) by finding a common factor of 'ASK in both components of the right-hand side of (5.11) as follows: οPassenger Revenue = ቂASK ×

ο୔ୖ୅ୗ୏ ο୅ୗ୏

+ PRASKቃ × οASK

We can then re-write this expression by pulling the PRASK term in the square parentheses outside of those parentheses:

29 In this section we examine the relationship between passenger revenue and ASKs using a deeper set of analytical tools. I understand this is not everyone’s (Dilmah) cup of tea. I encourage you to read through this section and understand the detail, but if your cup of tea is half full and you choose to bypass this section, I will not hold it against you.

Medium Run Airline Revenue

οPassenger Revenue = ቂ

ο୔ୖ୅ୗ୏ ୅ୗ୏ ο୅ୗ୏ ୔ୖ୅ୗ୏

+ 1ቃ × PRASK × οASK

107

(5.12)

ο௉ோ஺ௌ௄ ஺ௌ௄

on the right-hand side of (5.12) in square parentheses is called the elasticity of own PRASK with The term ο஺ௌ௄ ௉ோ஺ௌ௄ respect to own ASKs. We also refer to this as the own yield elasticity for short. This measures the percentage change in PRASK in response to a 1% increase in own ASKs. It is a negative number because higher own ASKs will reduce own PRASK. Using information about the own yield elasticity we can re-write (5.12) using words in the following way: οPassenger Revenue = (Own Yield Elasticity + 1) × PRASK × οASK

(5.13)

ο௉ோ஺ௌ௄ ஺ௌ௄

. Equation (5.13) says that if we wish to find the change in passenger revenue where Own Yield Elasticity { ο஺ௌ௄ ௉ோ஺ௌ௄ of an airline when there is a given change in the ASKs of that airline, we need only assume a value for the Own Yield Elasticity and the starting PRASK of the airline and substitute these numbers into the right-hand side of equation (5.13). By dividing both sides of (5.13) by ASK we can convert (5.13) into an expression that is in percentage change terms: % Change in Passenger Revenue = [1 + Own Yield Elasticity] u % Change in Own ASK

(5.14)

Equation (5.14) is used when an analyst wishes to estimate the percentage impact on revenue of a certain percentage change in capacity. The only parameter that the analyst needs in this case is information about the own yield elasticity. By determining whether the left-hand side of (5.14) is positive or negative, we can ascertain whether an increase in ASKs leads to higher or lower passenger revenue. Because % Change in Own ASK is positive, as we are assuming that ASKs increase, then the sign of the left-hand side of (5.14) depends on whether the term in square brackets on the righthand side [1 + Own Yield Elasticity] is positive or negative. When this term is positive then an increase in ASKs causes passenger revenue to go up, and when the term is negative an increase in ASKs causes passenger revenue to go down. As the Own Yield Elasticity is negative then it is possible that the term in brackets can be positive or negative. We can summarise this discussion as follows: Passenger revenue decreases in response to an increase in ASKs when the absolute value of the Own Yield Elasticity is greater than 1, which means the own yield is elastic to ASKs. Passenger revenue increases in response to an increase in ASKs when the absolute value of the Own Yield Elasticity is less than 1, which means the own yield is inelastic to ASKs.

What do we mean by elastic and inelastic own yield elasticity? An elastic own yield elasticity means that it is less than -1, for example it is -1.5 or -2.30 In this case the own yield is sensitive to a change in ASKs, which will be the case when the airline flies a predominantly business-purpose demographic. An inelastic own yield elasticity means that it is greater than -1, for example it is -0.5 or -0.75. In this case the own yield is relatively insensitive to a change in ASKs, which will be the case when the airline flies a predominantly leisure-purpose demographic. Let us now demonstrate how these medium run revenue rules are applied in the case of an airline illustration.

5.4.2 An Airline Illustration – Southwest Airlines An airline analyst has built a model of Southwest Airlines PRASM using quarterly data. This model says that the PRASM of Southwest Airlines in each quarter depends on the ASMs of the airline in each quarter in the following way: PRASM = 0.195 – 0.000000591 u ASM where PRASM is measured in US dollars (not cents) and ASM is the number of available seat miles in millions (remember that we are dealing with a North American airline, therefore the passenger yield is measured as PRASM and not PRASK). We wish to understand what happens to Southwest passenger revenue when the airline increases its ASMs by 1b from current values. The current value for PRASM is 13.358 US cents per ASM while the current value for ASMs is 40,570m. To determine the impact on Southwest passenger revenue if it increases ASMs by 1b we use equation (5.11). Let us now examine how we find each of the terms on the right-hand side of (5.11). The first term that we need to find is the response by PRASM to a change in ASMs. This can be found in the Southwest PRASM function as the coefficient attached ο௉ோ஺ௌ௄ = -0.000000591. It is important that you understand how to interpret this number. to ASMs, which is equal to ο஺ௌ௄ Because the PRASM function is a linear equation then the correct interpretation of the number relies on a correct understanding of the units that are used to measure PRASM and ASM. Remember that PRASM is measured in US dollars and ASM is measured in millions of ASMs per quarter. It follows that the coefficient attached to ASMs in the Southwest PRASM equation is interpreted as follows: 30 Remember that a number such as -2 is less than a number such as -1 because a higher negative number is lower. For the same reason, a number such as -0.5 is greater than a number such as -1 because it is a lower negative number.

Chapter 5

108

A 10 b increase in Southwest quarterly ASMs, which represents an increase in ASM by 10,000 (since ASM is measured in millions) causes a 0.00591 US dollar or 0.591 US cents decrease in Southwest PRASM, other things being equal.

To find the change in revenue in response to a change in ASMs of 10,000 the first step is to compute the first component on the right-hand side of equation (5.11). To do this we use information about the slope of the PRASM function presented above multiplied by the original ASMs, which is equal to 40,570, multiplied by the change in ASMs, which is assumed to be 'ASM = 10,000. This yields the following calculation: ASM ×

ο୔ୖ୅ୗ୑ ο୅ୗ୑

× οASM = 40,570u(– 0.000000591)u10,000 = -US$239.77m

This represents the price effect of a change in capacity – it is caused by the fact that PRASM declines in response to a 10b increase in Southwest capacity. This effect has a negative impact on Southwest revenue. We next find the second component on the right-hand side of (5.11), which is the volume effect. To find the volume effect we only need information about the original PRASM of the airline, which is PRASM = 0.13358, and the change in ASMs which gives: PRASM × οASM = 0.13358u10,000 = $1,335.80m The value US$1,335.80m represents the volume effect of a change in Southwest capacity. This says that the airline can sell more seats to passengers for a given fare if it increases capacity by 10b. Substituting the price and volume effects into (5.11) yields the change in medium run passenger revenue of Southwest Airlines in response to a 10b increase in quarterly ASMs: οPassenger Revenue = $1,335.80 െ $239.77 = $1,096.03 This result means that a 10b increase in Southwest ASMs leads to a US$1,096.03m increase in Southwest passenger revenue. Let us now show how we use the short cut formulae (5.13) and (5.14) to perform the same calculation for Southwest Airlines. Equations (5.13) and (5.14) allow us to examine the change in revenue by using more user-friendly parameters. Let us start with equation (5.13). The key component of this expression is the Own Yield Elasticity. The Own Yield Elasticity for Southwest Airlines in the case of a linear PRASM function is found by multiplying the coefficient attached to ASM in the PRASM function by ASMs divided by PRASM as follows: Own Yield Elasticity =

ο୔ୖ୅ୗ୑ ο୅ୗ୑

×

୅ୗ୑ ୔ୖ୅ୗ୑

= Slope of PRASM u

୅ୗ୑ ୔ୖ୅ୗ୑

= -0.000000591 u

ସ଴,ହ଻଴ ଴.ଵଷଷହ଼

= -0.17949

This own yield elasticity means that a 10% increase in Southwest ASMs leads to a 1.8% reduction in Southwest PRASM. If we substitute the estimated Own Yield Elasticity into equation (5.13) together with the change in ASMs, and the value of PRASM before the change in ASMs, then we obtain: οPAX Revenue = (Own Yield Elasticity + 1) × PRASM × οASK = (1 െ 0.17949) × 0.13358 × 10,000 = $1,096.04 You will notice that this generates a similar result to the use of equation (5.11), with the very small difference due to rounding error. Let us now look at the percentage change in Southwest revenue in response to the increase in Southwest ASMs using equation (5.14). This is found using the same estimate of the Own Yield Elasticity. Substituting the relevant parameters into (5.14) yields the percentage change in Southwest passenger revenue: % Change in Passenger Revenue = [1 + Own Yield Elasticity] u % Change in Own ASM = [1 – 0.17949] u ଵ଴,଴଴଴ = 20.2% ସ଴,ହ଻଴

Note that the increase in Southwest capacity is

ଵ଴,଴଴଴ ସ଴,ହ଻଴

= 24.6%. This means that a 24.6% increase in capacity results in

a 20.2% lift in revenue. Revenue does not lift by the full 24.6% increase in capacity because of the 4.4% reduction in PRASM. To demonstrate the reduction in PRASM we perform the following calculation: % Reduction in PRASM = Yield Elasticity u % Increase in ASM = -0.17949 u24.6% = 4.4% ஺ௌெ

You will also notice in this illustration that the Own Yield Elasticity will heavily depend on the value of . In other ௉ோ஺ௌெ words, it will depend heavily on starting values for Southwest capacity and yield. At current ASMs and PRASM, the

Medium Run Airline Revenue

109

Southwest Own Yield Elasticity is highly inelastic at -0.17949 which means increasing ASMs will increase passenger revenue because there is only a small decline in PRASM. If we were to use a much higher value for ASMs resulting in an Own Yield Elasticity which is relatively elastic, then passenger revenue will fall in response to the increase in ASMs. Readers should now develop a scenario for ASMs in which yield is elastic to ASMs and revenue falls in response to an increase in capacity.

5.4.3 Drivers of the Own Yield Elasticity The own yield elasticity is incredibly important in understanding what will happen to passenger revenue when the airline decides to increase capacity. As a result, it is necessary that we understand the key drivers of this elasticity. The own yield elasticity tells us about the extent to which yields must fall to fill the extra seats that an airline plans to operate. It is therefore primarily determined by the extent to which the airline can attract passengers onto its flights by offering a lower price. This is called the airfare elasticity of demand as we described in significant detail in chapter 3. The airfare elasticity of demand primarily depends on whether the route in question is leisure-focused or businessfocused. The more leisure-focused is a route, the more sensitive demand is likely to be to fares. This means that a relatively small yield reduction will coincide with an increase in ASKs, or the own yield elasticity will be low in absolute terms. Conversely, the more business-focussed is a route, the less sensitive demand is likely to be to a drop in fares. This means that a relatively large yield reduction will coincide with an increase in ASKs, or the own yield elasticity is high in absolute terms.

Quiz 5-2. PRASK and Own ASKs Rex Express Airlines competes by itself on the Sydney (SYD) to Merimbula (MIM) route (Merimbula is a small beachside town on the far south coast of New South Wales, Australia). Rex decides to increase its ASKs on the route by 10%. Its Own Yield Elasticity is -0.5. Use this information to answer the following three questions. 1. What is the percentage change in Rex’s revenue on the route that is attributable to the reduction in PRASK? (a) -5% (b) +5% (c) -10% (d) +10% 2. What route? (a) (b) (c) (d)

is the percentage change in Rex’s revenue on the route that is attributable to the change in Rex ASKs on the -5% +5% -10% +10%

3. What is the interpretation of Rex’s Own Yield Elasticity? (a) A 1% increase in Rex PRASK leads to a 0.5% decrease in Rex ASKs. (b) A 1% increase in Rex ASKs leads to a 5% decrease in Rex PRASK. (c) A 1% increase in Rex ASKs leads to a 0.5% decrease in Rex PRASK. (d) A 1% increase in Rex ASKs leads to a 0.5% decrease in Rex revenue. 4. Consider the Qantas domestic PRASK function introduced in section 5.3.1 once again and repeated for your convenience, PRASK = 31.5 - 0.0004 u ASK, where PRASK is denominated in Australian cents and ASKs is measured in millions. Use this information to answer the following questions. (a) Use this Qantas domestic PRASK function to estimate the impact of a change in ASKs on medium run revenue. (b) Split the change in revenue up into price and volume effects. (c) Determine the elasticity of PRASK to ASKs at ASKs = 30,000m. Interpret the elasticity for a 10% increase in ASKs. (d) Find the level of ASKs and PRASK at a PRASK elasticity to ASKs of -1. 5. An airline economist estimates the following revenue equation for Turkish Airlines using quarterly data which is denominated in terms of millions of US dollars, Rt = 8.61 u 10-5uASKt – 4.32u10-13uASKtuASKt, where ASK is defined in units of thousands per quarter t. Use this quadratic revenue function to answer the following questions. (a) Graph Turkish Airlines quarterly revenue as a function of ASKs using Microsoft Excel (revenue on the vertical axis and ASKs on the horizontal axis). Note that 8.61 u 10-5 can be written in Excel as 0.0000861. When building the revenue function in Excel, start with ASKs of 0 and increase this in increments of 100,000 until you obtain a range of values for ASKs from 0 up to 150,000,000.

Chapter 5

110

(b) The current level of Turkish Airlines ASKs is 50,000,000 per quarter. Find the estimated level of revenue per quarter in millions of US dollars at this level of ASKs. Find the revenue per ASK in US cents at this level of ASKs. Remember that in the revenue equation, revenue is measured in millions of US dollars per quarter and ASKs are measured in thousands of ASKs per quarter. (c) Now increase ASKs by 10% to 55,000,000 per quarter. What is the impact on revenue? Decompose the revenue change into a volume effect and a price effect. Explain your answer. 6. Consider the Aer Lingus Cobb Douglas PRASK function once again from questions 8, 9 and 10 in Quiz 5.1 repeated for your convenience, PRASK = 1,000 u ASK-0.45, where PRASK is measured in Euro cents and ASK is measured in millions. Use this information to find the percentage change in Aer Lingus revenue and the Euro dollar increase in Aer Lingus revenue when there is a 10% increase in ASKs (assume base ASKs for Aer Lingus of 30,000m).

5.5 Impact of ASKs on Medium Run Revenue in the Presence of Competitors 5.5.1 Passenger Revenue Effects - Logic In section 5.4 we examined the impact on an airline’s revenue when there is a change in its own ASKs. The implied assumption that we used in this section was (1) that competitors do not change their ASKs in response to a change in own ASKs, or (2) the airline is a monopolist, which means it does not face competition, or (3) competitors do change their ASKs in response to a change in own ASKs, but these changes have no influence on own PRASK. These assumptions would seem to be unrealistic for two reasons. Firstly, it is unlikely that all competitors of an airline will sit idly by and let their market share decline when there is an increase in own ASKs. Secondly, along most routes there are at least two airlines offering services, or in most cases more than two airlines. The aim of this section is to re-examine the impact on medium run revenue of an airline increasing its own capacity under the new assumption that there is a group of competitors who react to this change in own capacity, and this reaction may have an impact on the PRASK and thus revenue of the own airline. When an airline changes its own capacity and competitor airlines react to this by changing their capacity, the impact that these capacity changes, both own and competitor, have on the own yield of the airline are far more complicated. Figure 5-6 below attempts to capture this extra complexity associated with the impact of the reaction by competitors on the outcome for yield in a schematic diagram. n Own ASKs

Competitor Reaction

Volume Effect Own Yield Elasticity 'Passenger Revenue

'Competitor ASKs

Cross Yield Elasticity

'PRASK

Fig. 5-6: Impact of Own ASKs on Revenue with Competitor Reaction Figure 5-6 says that an increase in own ASKs (n Own ASKs box in the top left-hand corner) leads to five different effects on own passenger revenue. The first is the volume effect of higher ASKs on an increase in passenger volumes and thus passenger revenue, as indicated by the arrow that comes out of the box in the top left-hand corner and heads vertically down to the box entitled 'Passenger Revenue (this volume effect is the same effect as in the no-competitors model). The second is the price effect of higher own ASKs on a lower own PRASK, which has a negative impact on passenger revenue, as indicated in Figure 5-6 by the arrow that runs from the box in the top left-hand corner diagonally down and across to the right to the 'PRASK box, with the arrow entitled “Own Yield Elasticity” (this is the same price effect as the no-competitors model). The third is the impact of own ASKs on competitor ASKs, which is indicated in Figure 5-6 by an arrow that runs from the box in the top left-hand corner horizontally to the right to the 'Competitor ASKs box and entitled “competitor reaction”. The fourth is the impact of competitor ASKs on own airline PRASK, which is indicated by the arrow that runs vertically down from the 'Competitor ASKs box to the 'PRASK box which is entitled “Cross Yield Elasticity”. The fifth is the impact of a change in PRASK on passenger revenue, which is indicated by the arrow that runs from the 'PRASK box horizontally to the left to the 'Passenger Revenue box. The first, second and fifth effects are the same as those analysed in Figure 5-1 and section 5.4 in the case in which there are no competitors. The new effects are the third and fourth effects described above. Let us now discuss these new effects in more detail.

Medium Run Airline Revenue

111

The third effect is referred to as the competitor reaction effect. It measures the percentage change in competitor capacity in response to an increase in own capacity. The competitor reaction effect is summarised by a parameter called the competitor reaction elasticity, which we will define more formally below. The competitor reaction elasticity can be positive or negative. If it is positive then competitors increase their capacity in response to an increase in own capacity, and it is negative when competitors reduce their capacity. The competitor reaction elasticity is equal to 1 when competitors increase their capacity by the same percentage as own capacity increases, and it is equal to 0 when competitors do not react at all. A competitor reaction elasticity that is equal to 0.5 means that competitor capacity increases by 5% each time own capacity increases by 10%. A competitor reaction elasticity of -0.5 means that competitor capacity falls by 5% for each 10% increase in own capacity. A competitor reaction elasticity of 1.5 means that competitor capacity increases by 15% each time own capacity increases by 10%. For competitors who wish to preserve their market share, they will increase ASKs by the same percentage as own ASKs, in which case the competitor reaction elasticity is equal to 1. The fourth effect is referred to as the cross-yield effect. It measures the percentage change in own yield in response to an increase in competitor capacity. The cross-yield effect is summarised by a parameter called the cross-yield elasticity, which we will define analytically below. The cross-yield elasticity is always negative. This means that an increase in competitor capacity reduces the own yield. When the elasticity is -0.5, for example, this means that a 10% increase in competitor capacity leads to a 5% reduction in the own yield. If the cross-yield elasticity equals -1 this means that a 10% increase in competitor capacity leads to a 10% reduction in the own yield. And if the cross-yield elasticity is equal to -1.5 this means that a 10% increase in competitor capacity leads to a 15% reduction in own yield.

5.5.2 Passenger Revenue Effects – Analytics Let us now examine the influence of the competitor reaction and cross-yield effects analytically. Unfortunately, the algebra in this section does get a little complicated so this sub-section is really for those students who wish to understand the analytics in a lot more detail. If you are not mathematically minded, then it is perfectly fine for you to understand the logic presented in section 5.5.1 and use the summary formulae that are presented in this section to estimate the impact of a change in ASKs on revenue. To examine the impact of competitors on the relationship between a change in capacity and revenue, we must first specify the general PRASK function as a function of both own ASKs and competitor ASKs as follows: PRASK = f(ASKs, Comp. ASKs)

(5.15)

Equation (5.15) simply says that there is a function that relates the PRASK of an airline to both its own ASKs, or the ASKs of the airline in question, and the ASKs of the set of competitors. We now wish to find the impact on the left-hand side of (5.15) if there is a change in both ASKs and Competitor ASKs on the right-hand side of (5.15). We write the change in PRASK on the left-hand side of (5.15) as a function of the change in own and competitor ASKs in the following way: οPRASK =

ο୔ୖ୅ୗ୏ ο୅ୗ୏

× οASK +

ο୔ୖ୅ୗ୏ οେ୭୫୮୅ୗ୏

×

οେ୭୫୮୅ୗ୏ ο୅ୗ୏

× οASK

(5.16)

Let us look at the component parts of the right-hand side of (5.16). The first component on the right-hand side, ο௉ோ஺ௌ௄ × ο‫ܭܵܣ‬, says that when an airline changes its own ASKs this has a direct impact on own PRASK. The second ο஺ௌ௄

component on the right-hand side of (5.16),

ο௉ோ஺ௌ௄ ο஼௢௠௣஺ௌ௄

×

ο஼௢௠௣஺ௌ௄ ο஺ௌ௄

× ο‫ܭܵܣ‬, says that when an airline changes its own

ASKs, this influences competitor ASKs which in turn affects the own PRASK. The full effect on PRASK therefore has a direct effect, which is the first component, and an indirect effect through competitor ASKs, which is the second component. If we substitute (5.16) into the right-hand side of the change in passenger revenue expression (5.10) and simplify we obtain the following rather complicated expression for the change in passenger revenue: 1+ οPassenger Revenue = ቎

ο୔ୖ୅ୗ୏ ୅ୗ୏

+ቀ

ο୔ୖ୅ୗ୏

ο୅ୗ୏ ୔ୖ୅ୗ୏ οେ୭୫୮୅ୗ୏ οେ୭୫୮୅ୗ୏ ୅ୗ୏

×ቀ

ο୅ୗ୏

×

×

େ୭୫୮୅ୗ୏ ୔ୖ୅ୗ୏

ቁ େ୭୫୮୅ୗ୏

ቁ

቏ × PRASK × οASK

(5.17)

We can see that the change in passenger revenue at (5.17) depends on three key elasticities, which are all inside the square brackets on the right-hand side of equation (5.17). These elasticities are the Own Yield Elasticity which is ο୔ୖ୅ୗ୏ ୅ୗ୏ ο୔ୖ୅ୗ୏ େ୭୫୮୅ୗ୏ , the Cross Yield Elasticity, which is × , and the Competitor Reaction Elasticity, which ο୅ୗ୏ ୔ୖ୅ୗ୏ οେ୭୫୮୅ୗ୏

୅ୗ୏

ο୅ୗ୏

େ୭୫୮୅ୗ୏

is

×

οେ୭୫୮୅ୗ୏

୔ୖ୅ୗ୏

. We can substitute these names for the various elasticities into (5.17) to obtain the change in

passenger revenue in words:

Chapter 5

112

οPax Rev. = [1 + Own Yield Elast. +Cross Yield Elast.× Comp. Reaction Elast. ] × PRASK × οASK (5.18) The three elasticities that are contained in the square brackets on the right-hand side of (5.18), which are key to determining whether passenger revenue goes up or down in response to an increase in ASKs, are interpreted in the ο୔ୖ୅ୗ୏ ୅ୗ୏ is the % change in own PRASK due to a 1% increase in own following way. The Own Yield Elasticity { ASKs. The Cross Yield Elasticity {

ο୅ୗ୏ ୔ୖ୅ୗ୏ ο୔ୖ୅ୗ୏ େ୭୫୮୅ୗ୏

οେ୭୫୮୅ୗ୏

×

୔ୖ୅ୗ୏

is the % change in own PRASK due to a 1% increase in

competitor ASKs. Finally, the Competitor Reaction Elasticity {

οେ୭୫୮୅ୗ୏ ο୅ୗ୏

×

୅ୗ୏ େ୭୫୮୅ୗ୏

is the % change in Competitor

ASKs due to a 1% increase in own ASKs. Equation (5.18) says that if an airline were to increase its ASKs, then we can calculate the approximate impact on passenger revenue by substituting the relevant parameters and variables into the right-hand side of equation (5.18). Let us now have a look at a brief numerical example of how (5.18) can be used to determine the change in revenue in response to an increase in ASKs.

5.5.3 Airline Illustration – Frontier, Southwest, and Spirit There are three carriers which operate direct, non-stop services on the San Francisco (SFO) to Denver (DEN) city pair – Frontier, United and Southwest. We wish to examine the impact on Frontier passenger revenue if it decides to increase the number of non-stop services on the route from 2 daily services to 3 daily services, representing a 33% increase in Frontier capacity. A 33% increase in Frontier capacity represents an increase of 100m ASMs per year based on current ASMs at the time of writing. In response to the 33% increase in Frontier capacity, a combination of Southwest and United Airlines, who both compete against Frontier, increase capacity on the city pair by 10%. The competitor reaction elasticity is therefore: % Change Southwest & United ASMs 10% = = 0.303 % Change in Frontier ASMs 33% The PRASM of Frontier before the increase in ASMs is 11.8 US c/ASM. The own yield elasticity is assumed to be -0.25 and the cross-yield elasticity is assumed to be -0.1. Substituting these key parameters into (5.18) allows us to find the change in Frontier passenger revenue on SFO-DEN in response to a 33% increase in Frontier ASMs: οFrontier Pax Revenue = [1 െ 0.25 െ 0.1 × 0.303] × 0.118 × ο100 = US$8.49m A 33% increase in Frontier ASMs leads to a US$8.49m increase in Frontier passenger revenue, even after taking into consideration the increase in Southwest and United ASMs.

5.5.4 Passenger Revenue Effects – Analytics Percentage Change in Revenue Rather than investigating the change in passenger revenue in response to a change in ASKs we can also investigate the percentage change in passenger revenue in response to a percentage change in ASKs. To obtain the percentage change in passenger revenue we need to modify formula (5.18) by dividing both sides by passenger revenue as follows: %οRevenue = [1 + Own Yield Elast. +Cross Yield Elast.× Comp. Reaction Elast. ] × % οASK (5.19) Equation (5.19) simply says that the percentage change in revenue in response to a given percentage change in own ASKs depends on the Own Yield Elasticity, the Cross Yield Elasticity, and the Competitor Reaction Elasticity. Depending on the sizes of these elasticities own revenue can go up or down in response to a given percentage increase in own ASKs. According to (5.19), passenger revenue increases in response to an increase in own capacity when the following condition holds: 1 + Own Yield Elasticity + Cross Yield Elasticity × Competitor Reaction > 0

(5.20)

The condition at equation (5.20) simply says that passenger revenue increases in response to an increase in own capacity when capacity from both the own airline and competitors do not generate a reduction in PRASK that is proportionately greater than the increase in own capacity. Own revenue falls in response to a given increase in capacity when the following condition holds: 1 + Own Yield Elasticity + Cross Yield Elasticity × Competitor Reaction < 0

(5.21)

In this case the movements in own and competitor capacity cause a fall in own PRASK that is proportionately greater than the increase in own ASKs. PRASK is more likely to decline by more than own ASKs increase, generating a passenger revenue reduction, when air travel demand is relatively inelastic to the average airfare, there is a strong, positive

Medium Run Airline Revenue

113

competitor reaction to the increase in capacity, and the group of competitors produce a product that is similar to that of the own airline, so that the cross-yield elasticity is similar in magnitude to the own yield elasticity. To see how (5.20) can be used, consider our Frontier example on SFO-DEN once again from section 5.5.3. Substituting the relevant parameters into (5.19) gives: % Change in Frontier Passenger Revenue = [1 – 0.25 – 0.1u0.303] u 0.33 = 23.75% This implies that the 33% increase in Frontier ASMs results in a 23.75% increase in revenue. While Frontier’s PRASM falls by 33% – 23.75% = 9.25% this is insufficient to dominate the increase in ASMs and the favourable volume effect this produces, resulting in an increase in revenue.

5.5.5 Airline Illustration – Southwest Airlines and Spirit 5.5.5.1 Long Method Rather than using the formulae presented in sections 5.5.2 and 5.5.4 to find the change in passenger revenue of an airline in response to an increase in ASKs we can simply estimate the change in revenue using a more direct method, which takes slightly longer, hence the reason for calling it the long method. To demonstrate this method, we consider Southwest Airlines competition against Spirit Airlines on the Los Angeles (LAX) to Las Vegas (LAS) route. Southwest (WN) operates 16 direct return services per day on LAX-LAS and Spirit (NK) operates 6 direct return services per day. WN’s PRASM function on the route is estimated to be: PRASMWN = 0.19 - 0.0003 × ASMWN – 0.00015 × ASMNK where PRASMWN is measured in US dollars and ASMWN and ASMNK are the available seat miles of Southwest and Spirit respectively, both measured in millions of annual ASMs. We assume WN’s ASMs per year on the route currently are ASMWN = 200m and NK’s are ASMNK = 100m. WN’s medium run revenue function is the PRASM function multiplied by ASMWN, which gives: ଶ RWN = 0.19ASM୛୒ െ 0.0003ASM୛୒ െ 0.00015ASM୒୏ × ASM୛୒

(5.22)

We wish to find the change in WN medium run revenue at (5.22) in response to a change in WN’s ASMs and see how this relates to the formulae (5.18) and (5.19) above. Let us first find an expression for the total change in revenue of WN. By doing this we are finding the impact of a change in WN ASMs on WN revenue using the ‘long approach’. The long approach starts by totally differentiating the RWN function (5.22), which gives:31 dRWN = [0.19 – 0.00015ASMNK – 0.0003ASMWN]dASMWN - [0.00015dASMNK + 0.0003dASMWN]ASMWN

(5.23)

You will notice in (5.23) that the first component on the right-hand side in square brackets is just WN PRASM multiplied by the change in Southwest ASMs. This is the volume effect of a change in WN ASMs: Volume Effect = [0.19 – 0.00015 × ASMNK – 0.0003× ASMWN] × dASMWN

(5.24)

This tells us how much Southwest revenue goes up if PRASM does not change and the airline decides to expand capacity. The right-hand component of the right-hand side of (5.23) is the price effect, which is: Price Effect = - [0.00015 × dASMNK + 0.0003 × dASMWN] × ASMWN

(5.25)

The price effect (5.25) tells us the impact of changes in WN and NK ASMs on WN PRASM, and the effect of a change in WN PRASM on revenue. We assume that NK increases its ASMs by half of the increase in WN’s increase in ASMs. This means the change in NK ASMs is: dASMNK = 0.5×dASMWN

(5.26)

Substituting (5.26) into (5.23) yields the change in WN revenue as a function of the change in WN ASMs: 31

See Chiang and Wainwright 1985, 185-190 if you need a refresher on how to find the total differential of a function. The total differential of the function y = f(x) is dy = f’(x) dx. The total differential of the function y = f(x, z) is dy = fx dx + fz dz. All you are doing here is finding the total change in the dependent variable as a function of the changes in the independent variables. The changes in the independent variables affect the dependent variable through the derivative of the dependent variable with respect to the relevant independent variables.

Chapter 5

114

dRWN = [0.19 – 0.00015×ASMNK – 0.0003×ASMWN] ×dASMWN - [0.00015×0.5 + 0.0003] ×ASMNK×dASMWN

(5.27)

Let us now substitute the pre-change values of WN and NK ASMs, ASMWN = 200 and ASMNK = 100, into (5.27). If we do this, we obtain the following simplified numerical expression for the change in WN revenue: dRWN = [0.19–0.00015×100–0.0003 × 200]dASMWN - [0.00015×0.5 + 0.0003]×200×dASMWN This simplifies to: dRWN = (0.115 – 0.075) × dASMWN = 0.04 × dASMWN

(5.28)

Equation (5.28) indicates that Southwest revenue will increase by 4% of the small increase in Southwest capacity. 5.5.5.2 Short-Cut Formula Let us now see how we can estimate the impact on WN revenue of a change in WN ASMs using the short-cut formula at (5.18). To use (5.18) we must first calculate the own yield elasticity. The own yield elasticity is the elasticity of WN PRASM to a change in WN ASMs. This is equal to: WN Own Yield Elasticity =

ୢ୔ୖ୅ୗ୑౓ొ ୢ୅ୗ୑౓ొ

×

୅ୗ୑౓ొ ୔ୖ୅ୗ୑౓ొ

= െ0.0003 ×

୅ୗ୑౓ొ

(5.29)

୔ୖ୅ୗ୑౓ొ

If we determine the WN own yield elasticity at the pre-change values for WN and NK ASMs and WN PRASM we obtain: WN Own Yield Elasticity = െ0.0003 ×

ଶ଴଴ ଴.ଵଽ – ଴.଴଴଴ଵହ × ଵ଴଴ – ଴.଴଴଴ଷ × ଶ଴଴

= െ0.52174

This WN own yield elasticity estimate means that a 10% increase in WN ASMs leads to a 5.22% reduction in WN PRASM, other things being equal. We next calculate the cross-yield elasticity. This measures the impact on WN PRASM of a change in NK ASMs. This is equal to: WN Cross Yield Elasticity =

ୢ୔ୖ୅ୗ୑౓ొ ୢ୅ୗ୑ొే

×

୅ୗ୑ొే ୔ୖ୅ୗ୑౓ొ

= െ0.00015 ×

୅ୗ୑ొే

(5.30)

୔ୖ୅ୗ୑౓ొ

If we determine (5.30) at the pre-change levels of WN and NK ASMs we obtain: WN Cross Yield Elasticity = െ0.00015 ×

ଵ଴଴ ଴.ଵଽ – ଴.଴଴଴ଵହ × ଵ଴଴ – ଴.଴଴଴ଷ × ଶ଴଴

= െ0.13043

The cross-yield elasticity estimate says that a 10% increase in NK ASMs leads to a 1.3% reduction in WN PRASM other things being equal. We next calculate the competitor reaction elasticity. From (5.26), this is found to be: WN Competitor reaction elasticity =

ୢ୅ୗ୑ొే ୢ୅ୗ୑౓ొ

×

୅ୗ୑౓ొ ୅ୗ୑ొే

= 0.5 ×

୅ୗ୑౓ొ ୅ୗ୑ొే

= 0.5 ×

ଶ଴଴ ଵ଴଴

=1

This means that a 10% increase in Southwest ASMs causes Spirit to increase ASMs by 10% in response. If we substitute the WN own yield elasticity, the WN cross-yield elasticity, and the WN competitor reaction elasticity into (5.18) together with the WN PRASM function we obtain: οRevenue୛୒ = [1 െ 0.52174 െ 1 × 0.13043] × (0.19 െ 0.0003 × 200 െ 0.00015 × 100) × οASM୛୒ This simplifies to: οRevenue୛୒ = 0.04 × οASM୛୒

(5.31)

Equation (5.31) is precisely our estimate at (5.28). This demonstrates that equation (5.18) in our theory section above actually works. You should now estimate the impact on Southwest revenue under different assumptions about the way that Spirit reacts to Southwest. For example, you can assume that Spirit doesn’t change its capacity, you can assume that Spirit increases capacity but only by 5% (so that the competitor reaction elasticity is 5%/10% = 0.5), you can assume that Spirit overreacts by increasing capacity by more than 10% (in which case the competitor reaction elasticity exceeds 1) and you can assume that Spirit reacts by reducing capacity (the competitor reaction elasticity is less than zero). These are all exercises that are left to you as homework.

Medium Run Airline Revenue

115

Quiz 5-3. PRASK, Own ASKs and Competitor Reaction easyJet and Norwegian Air International compete on the London Gatwick (LGW) to Copenhagen (CPH) route. Norwegian Air International has decided to increase its capacity by 10%. Norwegian estimates that its own yield elasticity is -0.25 and the cross-yield elasticity is -0.25 as well. It expects that easyJet will increase ASKs by 5% in response to its 10% increase. Use this information to answer the following 5 questions. 1. What is the percentage change in Norwegian Air International PRASK due to its own ASKs? (a) -2.5% (b) +2.5% (c) -10% (d) +10% 2. What is the percentage change in Norwegian Air International PRASK due to easyJet’s ASKs? (a) -2.5% (b) +2.5% (c) -5% (d) -1.25% 3. What is size of the competitor reaction elasticity? (a) 0.5 (b) 1 (c) 0.25 (d) 0 4. What condition leads you to the conclusion that Norwegian Air International revenue will rise in response to the increase in ASKs? (a) [1 + own yield elasticity + competitor reaction u cross-yield elasticity] = [1 – 0.25 - 0.5 u 0.25] < 0 (b) [1 + own yield elasticity + competitor reaction + cross-yield elasticity] = [1 – 0.25 + 0.5 - 0.25] > 0 (c) [1 + own yield elasticity u competitor reaction u cross-yield elasticity] = [1 – 0.25 u 0.5 u 0.25] > 0 (d) [1 + own yield elasticity + competitor reaction u cross-yield elasticity] = [1 – 0.25 - 0.5 u 0.25] > 0 5. What is the percentage change in Norwegian Air International revenue? (a) [1 – 0.25 + 0.5 u 0.25] u 0.1 (b) [1 – 0.25 + 0.5 - 0.25] u 0.1 (c) [1 – 0.25 u 0.5 u 0.25] u 0.1 (d) [1 – 0.25 - 0.5 u 0.25] u 0.1

5.6 Airline Group Revenue and Cannibalisation Effects 5.6.1 Examples of Airline Groups There are many airline groups around the world that operate full-service and low-cost carriers on the same routes at adjacent times of the day. When this occurs, it is possible that an expansion in the capacity of the low-cost carrier within the airline group will adversely affect the yields and thus revenue of the full-service airline within the airline group. This effect is referred to as a cannibalisation effect because an increase in the production of one product line within a company hurts the revenue associated with another product line within the same company. Table 5-1 below presents examples of airline groups that operate full-service and low-cost carriers on the same routes at adjacent times of the day. Where a low-cost carrier operates on the same route as a full-service carrier, and both airlines are a part of the same airline group, then we refer to the low-cost carrier as an internal competitor to the fullservice airline and the full-service airline as an internal competitor to the low-cost airline. This is to be distinguished from an external competitor, which is a competitor that is not a part of the airline group of the airline in question. For example, the Qantas Group owns two airline brands – Qantas Mainline and Jetstar. Qantas is a full-service carrier and Jetstar is a low-cost carrier. Both airlines operate on the Melbourne to Sydney route on the same day, at adjacent times of the day. If we examine Qantas yields on this route, these will be affected by Jetstar capacity. Jetstar capacity is referred to as internal, low-cost competitor capacity in this case because Jetstar is internal to the Qantas Group. Virgin Australia also operates on Sydney-Melbourne. Virgin Australia is a full-service airline which could influence the yields of both Qantas and Jetstar. Virgin Australia therefore represents an external, full-service competitor to the full-service airline Qantas and the low-cost carrier Jetstar on this route. Consider another example. British Airways and Iberia, both part of the International Airlines Group, operate services on the route Barcelona (BCN) to London Heathrow (LHR) at the time of writing. Both carriers are full-service airlines.

Chapter 5

116

Airline Group Qantas Group International Airlines Group

Singapore Airlines Group Lufthansa Group

Air France Group

Flying Brands Qantas Mainline Jetstar British Airways Iberia Aer Lingus Vueling Level Singapore Airlines Silk Air Scoot Lufthansa Austrian Eurowings Swiss Air Brussels Airlines Air France KML Transavia Air France Hop

Carrier Type Full-Service Low-Cost Full-Service Full-Service Full-Service Low-Cost Low-Cost Full-Service Full-Service Low-Cost Full-Service Full-Service Low-Cost Full-Service Full-Service Full-Service Full-Service Low-Cost Low-Cost

Table 5.1: Global Airline Groups An increase in Iberia capacity on this route may influence the yields of British Airways. In this case, Iberia is an internal full-service competitor to British Airways on this route. Singapore Airlines and Silk Air, both a part of the Singapore Airlines Group, operate services on the Singapore (SIN) to Bali (DPS) route. An expansion in Silk Air capacity on the route may negatively impact Singapore Airlines yields. Silk Air is a full-service airline and represents an internal, full-service competitor to the full-service airline Singapore Airlines. Jetstar Asia also operates on the route and is a low-cost carrier. If Jetstar Asia expands capacity on SIN-DPS then this may influence Singapore Airlines and Silk Air yields on the route. Jetstar Asia is not a part of the Singapore Airlines Group thus it represents a low-cost, external competitor to Singapore Airlines and Silk Air.

5.6.2 Cannibalisation Effect Flows Let us now analyse the impact on the revenue of a full-service airline in the case of an increase in the ASKs of a lowcost internal competitor. We summarise the complexity of this impact in Figure 5-7 below. To understand Figure 5-7, we start with the very top box labelled n Own LCC ASKs. In this box we assume that there is an increase in the ASKs of the low-cost carrier airline operating services within an airline group. For example, it might be Scoot increasing ASKs on services between Sydney and Singapore when the airline group knows that its mainline, full-service carrier Singapore Airlines also operates services on the same city pair. We describe the impact of an increase in own LCC ASKs on group revenue in a series of flows that are denoted by capital letters of the alphabet in Figure 5-7. Flow A This flow represents the volume effect of an increase in LCC ASKs. When an airline increases ASKs it usually flies more passengers which in turn contributes to the revenue of the LCC. Flow B What is relevant to this analysis is the impact on airline group passenger revenue of the change in LCC ASKs, not the impact on LCC passenger revenue of a change in LCC ASKs. Flow B describes how the LCC passenger revenue contributes to group passenger revenue and reinforces the fact that the impact we wish to measure is the impact of LCC ASKs on group passenger revenue not LCC or FSA passenger revenue individually. Flow C The expansion in LCC ASKs leads to a reduction in LCC PRASK. This occurs via the own LCC yield elasticity. This elasticity measures the percentage change in LCC PRASK in response to a 1% increase in LCC own ASKs. It represents the LCC’s Own Yield Elasticity and is a component of the price effect. Flow D This flow measures the impact of a reduction in LCC PRASK on lower LCC revenue. LCC PRASK and revenue will be influenced not only by its own ASKs, but also external competitor ASKs, which may react to the ASKs of the own LCC. Flow E The increase in LCC ASKs reduces the PRASK of the group’s FSA through the cannibalisation effect. This is measured by a cannibalisation elasticity. This elasticity measures the percentage reduction in FSA PRASK in response to a 1%

Medium Run Airline Revenue

117

increase in own LCC ASKs. The size of this cannibalisation elasticity at the total network level will depend on five forces - the extent to which the group airlines fly on the same routes, the extent to which they fly on the same day, the n Own LCC ASKs C

A

E

H F

LCC Passenger Revenue

D

p LCC PRASK

p FSA PRASK

'Competitor ASKs

I

B

F

J Group Passenger Revenue

G

pFSA Revenue

Fig. 5-7: Impact of Own ASKs on Revenue with Cannibalisation and Competitor Reaction extent to which they fly at adjacent times of the day, the extent of the product differences between the airlines, and the extent to which there is “cross-selling” between the two airlines. What do we mean by cross-selling? Examples of cross-selling include the low-cost carrier services being advertised on the full-service carrier’s website. For example, if you wanted to book a flight from Sydney to Melbourne on Qantas, then on the Qantas website you would be offered flight options not just for Qantas on Sydney to Melbourne but also on Jetstar from Sydney to Melbourne. In the case of Singapore Airlines, if you wanted to book a flight on Singapore Airlines from Singapore to a destination in Asia and Singapore does not fly to that destination in Asia but Scoot does, then Singapore will ask if you want to be redirected to the Scoot website to make the booking. Another type of cross-selling illustration involves frequent flyer programs. Full-service airline frequent flyer members can earn frequent flyer points by flying on the low-cost carrier. And full-service carrier frequent flyers can redeem their points earned on the fullservice carrier by burning them on the low-cost carrier services. Low-cost carrier passengers can also use the airport lounges of the full-service airline. This is another form of cross-selling as it enriches the low-cost carrier product at the potential expense of the full-service carrier product. Flow F This flow indicates how a reduction in the FSA PRASK leads to a reduction in FSA passenger revenue. This effect is a part of the cannibalisation effect because it involves a fall in FSA revenue because of its own yield falling, and its own yield falls because of an expansion in the capacity of the group’s LCC. Flow G This flow indicates how a reduction in FSA revenue leads to a reduction in Group passenger revenue, following the same logic as Flow B. Flow H The increase in LCC ASKs draws a reaction from external competitors, which we call 'Competitor ASKs. This may involve an increase or a decrease in external competitor ASKs depending on the relationship between the external competitor and the low-cost carrier. If the external competitor is a full-service airline, then it is conceivable that the external competitor will lower ASKs compared to a world in which the LCC does not increase ASKs. This is a defensive strategy that group airlines often use – they use strong growth in the LCC to stop the growth in a competitor to minimise the impact of the competitor on the FSA. If they see an external FSA competitor that is hurting their own FSA, then they will often grow a LCC airline to take the ‘wind out of the sails’ of the external FSA competitor. Alternatively, if the external competitor is an LCC that is in direct competition with the LCC of the group airline, then it is more likely to expand capacity so that it does not concede its share of the LCC market. Flow I The change in external competitor ASKs causes a change in the PRASK of the FSA. The extent to which this flowsthrough into a change in FSA revenue depends on whether the external competitor is a FSA or a LCC. If the external competitor is a FSA, then we would expect the change in external competitor ASKs would have a more powerful impact on FSA PRASK than if the external competitor were a LCC. The change in FSA PRASK in turn leads to a change in FSA passenger revenue through flow F which causes a change in group passenger revenue through flow G.

Chapter 5

118

Flow J The change in external competitor ASKs causes a change in the PRASK of the LCC. If the external competitor is a LCC we would expect the impact to be deeper than if the external competitor were a FSA. The change in PRASK flows through into a change in LCC revenue through flow D and then onto group revenue through flow B. The net impact of flows A to J on group revenue is exceptionally complex as will be evident when we look at the analytics of the impact. A topic to which we now turn with immense joy.

5.6.3 Cannibalisation Analytics Like other analytical sections in this chapter, this sub-section has complicated calculus and algebra that you may wish to skip if you are not mathematically inclined. If that is the case, then make sure that you understand the logic and intuition presented in sections 5.6.1 and 5.6.2 above and the summary formulae presented in this sub-section. To demonstrate cannibalisation effects analytically, we start with an expression for the definition of total passenger revenue of the airline Group, which is the sum of LCC revenue and FSA revenue as follows: R ୋ = LCC Passenger Revenue + FSA Passenger Revenue

(5.32)

The right-hand side of equation (5.32) can be decomposed into ASK and PRASK components for the LCCs and FSAs as follows: R ୋ = P୐ × A୐ + P୊ × A୊

(5.33)

The right-hand side of (5.33) says that LCC passenger revenue is LCC PRASK (PL) times LCC ASKs (AL), and FSA passenger revenue is FSA PRASK (PF) times FSA ASKs (AF). We have had to use the shorter-hand notation PL, AL, PF and AF for respectively PRASKLCC, ASKLCC, PRASKFSA and ASKFSA so that we can simplify the algebra and the calculus to follow. We wish to examine the impact on the left-hand side of (5.33), which is group passenger revenue, in the case of an increase in the right-hand side variable AL. Before we show you the calculus, it might be useful to set out the variables that a change in AL will impact. Firstly, it will have a volume impact, which means it will increase LCC revenue for a given PL. Secondly, it will have an own-yield effect, so PL will change in response to a change in AL. Thirdly, it will have a cannibalisation effect, so AL will influence PF. Fourthly, it will influence the capacity decisions of competitors, which in turn affects both PL and PF. To estimate the own-yield effects, we need to construct the PL and PF functions. We assume the following general functions for FSA and LCC PRASK: PL = fL(AL, AF, AC)

(5.34)

PF = fF(AL, AF, AC)

(5.35)

where AC in (5.34) and (5.35) represents competitor ASKs. The two PRASK functions (5.34) and (5.35) tell us that AL and AF can potentially influence the PRASK of both airlines. It is also the case that PRASK can be influenced by the ASK decisions of external competitors, AC. We also assume that we can differentiate these functions, which effectively means that the functions are continuous and smooth. Using (5.34) and (5.35) together with (5.33), the change in group passenger revenue in response to an increase in AL is found by differentiating (5.33) with respect to AL: οୖృ ο୅ై

= P୐ + A୐ × ቀ

ο୔ై ο୅ై

+

ο୔ై ο୅ి

×

ο୅ి ο୅ై

ο୔

ο୔

ο୅

ቁ + A୊ × ቀο୅ూ + ο୅ూ × ο୅ిቁ ై

(5.36)

ై

ి

Equation (5.36) says that when AL increases, there are five effects. The first is that there is a positive volume impact on group passenger revenue because the low-cost carrier will fly more passengers = PL. The second is that there is a negative ο୔ price effect on group passenger revenue which occurs because the increase in AL causes a reduction in PL = AL u ై . ο୅ై

The third is an external competitor reaction and LCC cross-yield effect which says the increase in AL provokes a change ο୔ ο஺ in AC, which in turn affects PL and thus group passenger revenue = AL u ಽ × ಴. The fourth is a cannibalisation effect which says the increase in AL leads to a reduction in PF = AF u

ο୔ಷ ο୅ಽ

ο஺಴

ο஺ಽ

. The fifth is an external competitor reaction and FSA

cross-yield effect which says the increase in AL provokes a reaction by AC, which in turn affects PF and thus group ο୔ ο஺ passenger revenue = AF u ಷ × ಴. ο஺಴

ο୅ಽ

We can re-arrange (5.36) so that it is written in terms of elasticities as follows: οୖృ ο୅ై

= ቂ1 +

ο୔ై ୅ై ο୅ై ୔ై

+ቀ

ο୔ై ୅ి ο୅ి ୔ై

ο୅ ୅

ୖ

ο୔ ୅

ο୔ ୅

ై

ి

ο୅ ୅

ቁ × ቀο୅ి ୅ై ቁቃ × P୐ + ୅ూ × ቂቀο୅ూ ୔ై ቁ + ቀο୅ూ ୔ిቁ × ቀο୅ి ୅ై ቁቃ ై

ి

ై

ూ

ూ

ై

ి

(5.37)

Medium Run Airline Revenue

119

Written in words we can write the change in group passenger revenue in response to a change in AL at (5.37) as: οୖృ ο୅ై

= [1 + Own Elas. L + Cross Elas. L × CR Elas. L] × P୐ +

ୖూ ୅ై

×ቂ

F Cann. Elas. +Cross Elas. F × ቃ (5.38) CR Elas. L

Equation (5.38) says that the change in group revenue in response to an increase in the ASKs of the low-cost carrier airline within the group depends on a variety of parameters. The first is the own yield elasticity of the LCC (Own Elas. L.), which determines the extent to which PRASKLCC declines in response to an increase in ASKLCC. The second is the elasticity describing the reaction of the capacity of the external competitor to the increase in ASKLCC (CR Elas. L). The third is the LCC Cross Yield Elasticity with respect to external competitor capacity (Cross Elas. L.), which measures the extent to which PRASKLCC declines when there is an increase in the ASKs of external competitors. The fourth is the cannibalisation elasticity (F. Cann. Elas.), which measures the extent to which PRASKFSA falls in response to an increase in ASKLCC. The fifth is the FSA Cross Yield Elasticity with respect to external competitor capacity (Cross Elas. F), which measures the impact of external competitor capacity on PRASKFSA. We can also write the change in group passenger revenue relative to the change in FSA capacity as follows: οୖృ ο୅ూ

= [1 + Own Elas. F + Cross Elas. F × CR Elas. F] × P୊ +

ୖై ୅ూ

×ቂ

L Cann. Elas. +Cross Elas. L × ቃ (5.39) CR Elas. F

The content of (5.39) has a similar interpretation as the content of (5.38). In this case the cannibalisation effect is in reverse, measuring the impact of FSA capacity on the PRASK of the LCC. I realise that these are super complicated expressions. If you are not great with mathematics, focus on the intuition and the logic and try to follow the numerical example to follow.

5.6.4 A Numerical Example – Qantas and Jetstar Qantas and Jetstar both operate services on the city pair Melbourne (MEL)-Brisbane (BNE). The airline group wishes to determine the impact of an increase in Jetstar’s capacity on Qantas Group revenue on MEL-BNE. An analyst for the airline has estimated the following PRASK functions for Qantas (Q) and Jetstar (J) as a function of the ASKs of Q and J as well as an external competitor Virgin Australia (V): ି଴.ଵ ି଴.ଶ A୚ PQ = 0.25 × Aି଴.ହ ୕ A୎

(5.40)

PJ = 0.1 × Aି଴.଻ Aି଴.ଷ ୎ ୚

(5.41)

We can see in these functions that Q’s PRASK is affected by all three sets of ASKs, while the PRASK of J is only affected by its own ASKs and the ASKs of the external competitor. This implies that the cannibalisation effect only runs in one direction – from J capacity to Q yields, which is our dominant expectation. The various elasticities implied by (5.40) and (5.41) are as follows: Own Yield Elasticity Q =

ப୔్ ப୅్

×

Cross Yield Elasticity Q െ V = Cross Yield Elasticity J െ V =

୅్

୔్ ப୔్

ப୅౒

= െ0.5 ×

୅౒ ୔్

= െ0.2

Cannibalisation Elasticity Q = Own Yield Elasticity J =

ப୔ె ப୅ె

×

ப୔్

ப୅ె ୅ె ୔ె

×

୅ె ୔్

= െ0.1

= െ0.7

μP୎ A୚ × = െ0.3 μA୚ P୎

The various elasticities are interpreted as follows. Own Yield Elasticity Q means that a 10% increase in AQ leads to a 5% reduction in PQ other things being equal. Cannibalisation Elasticity Q means that a 10% increase in AJ leads to a 1% reduction in PQ other things being equal. Cross Yield Elasticity Q-V means that a 10% increase in AV leads to 2% reduction in PQ other things being equal. Own Yield Elasticity J means that a 10% increase in AJ leads to a 7% reduction in PJ other things being equal. Cross Yield Elasticity J-V means that a 10% increase in AV leads to a 3% reduction in PJ other things being equal. The revenue that the airline group earns from full-service airline Q is (5.40) multiplied by AQ as follows: ି଴.ଵ ି଴.ଶ A୚ RQ = 0.25 × A଴.ହ ୕ A୎

(5.42)

The revenue that the airline group earns from low-cost carrier J is (5.41) multiplied by ASKJ as follows: ି଴.ଷ RJ = 0.1 × A଴.ଷ ୎ A୚ The group revenue is the sum of Q revenue (5.42) and J revenue (5.43), which is:

(5.43)

Chapter 5

120 ି଴.ଵ ି଴.ଶ ି଴.ଷ RG = RQ + RJ = 0.25×A଴.ହ A୚ + 0.1×A଴.ଷ ୎ A୚ ୕ A୎

(5.44)

We wish to analyse the impact of a change in ASKJ on RG in (5.44). We assume that the capacity of the external competitor is related to the capacity of the low-cost carrier in the following way: AV = 0.75×AJ

(5.45)

Substituting (5.45) into (5.44) yields: ି଴.ଵ ൫0.75A୎ ൯ RG = 0.25A଴.ହ ୕ A୎

ି଴.ଶ

+ 0.1×A଴.ଷ ୎ ൫0.75A୎ ൯

ି଴.ଷ

ି଴.ଷ = 0.265A଴.ହ + 0.109 ୕ A୎

(5.46)

The change in RG in response to a change in AJ is then: ିଵ.ଷ ିଵ.ଷ οA୎ = െ0.0795 × A଴.ହ οA୎ οR ୋ = െ0.3 × 0.265 × A଴.ହ ୕ A୎ ୕ A୎

(5.47)

We can see in this case that an increase in AJ will reduce group revenue because (5.47) is less than zero. This is because the positive volume effect is cancelled out by the negative external competitor effect on J, leaving only the negative cannibalisation effect and external competitor effect on Q, which are both negative. If the airline group were to expand Jetstar capacity in this case, it would unambiguously reduce Qantas Group profit.

5.6.5 A Detailed Illustration – Singapore Airlines Group 5.6.5.1 No External Competitor Reaction Singapore Airlines the mainline full-service carrier, and Scoot Airlines the low-cost carrier, both owned and operated by the Singapore Airlines Group, operate services to the same city pairs at the time of writing. These include SIN-MEL, SIN-SYD, SIN-CAN, SIN-KIX, SIN-TPE, SIN-BKK, SIN-PER, SIN-KUL, SIN-DPS, SIN-PEN, SIN-MAA, and SIN-SZX (Singapore Airlines Destinations 2021, Scoot Airlines Destinations 2021). If you do not know some of these three-letter airport codes, then please familiarise yourself with them by googling “airport code XXX”. For many of these city pairs the airlines operate reasonably adjacent frequencies. For example, on SIN-DPS (Singapore-Denpasar) the airlines operate to the schedules presented in Table 5-2 at the time of writing. Singapore Airlines Daily Flights 6:55am o 9:35am Daily 8:20am o 11:05am Daily 9:05am o 11:50am Daily 4:20pm o 7:10pm Daily 6:05pm o 8:50pm Daily

Scoot Daily Flights 7:15am o 10:05am Daily 10:30am o 1:20pm Daily 3:20pm o 6:05pm Daily

Source: Google search “flights Singapore to Denpasar” September 20, 2019

Table 5.2: Singapore to Denpasar Flights for Singapore Airlines and Scoot In Table 5-2 we can see that Singapore Airlines flies 20 minutes prior to Scoot in the morning and an hour after Scoot in the afternoon in the case of Singapore to Denpasar flights. Table 5-3 below presents adjacent services operated by both flying brands in the case of Bangkok services. We can see in Table 5.3 that the airlines fly 20 minutes apart in the morning and at the same time in the afternoon on Singapore to Bangkok services. Table 5.4 below presents adjacent services in the case of Taipei flights. In this case there are morning flights that are within 15 and 40 minutes of each other. There would therefore seem to be significant opportunities for Scoot capacity expansion to dilute the yields of Singapore Airlines services. To test this proposition, an airline analyst has used statistical analysis to estimate a relationship between the logarithm of Singapore Airlines mainline passenger revenue per ASK and Scoot Airlines ASKs, amongst other variables. A summary of the estimated PRASK equation which is estimated at a quarterly frequency using data over time is given as follows: Log (SQ PRASKt) = 0.8051 – 0.332 u log (SQ ASKt) – 0.0000121 u Scoot ASKt where SQ ASK is Singapore Airlines mainline ASKs in millions and Scoot ASK is Scoot Airlines ASKs in millions. We can use information in this estimated equation to substitute into our formula at (5.38) to determine the extent to which an expansion in Scoot capacity affects Singapore Group revenue. The first piece of information that we can extract from the estimated regression is the cannibalisation elasticity. The cannibalisation elasticity is computed by using the

Medium Run Airline Revenue

Singapore Airlines Daily Flights 7:15am o 8:40am Daily 9:45am o 11:10am Daily 1:10pm o 2:35pm Daily 4:00pm o 5:25pm Daily 5:35pm o 7:00pm Daily 6:45pm o 8:10pm Daily

121

Scoot Daily Flights 6:55am o 8:25am Daily 7:00am o 8:30am Daily 9:30am o 10:55am Daily 12:45pm o 2:10pm Daily 4:00pm o 5:35pm Daily 5:35pm o 6:55pm Daily 10.05pm o 11:30pm Daily

Source: Google search “flights Singapore to Bangkok” September 20, 2019

Table 5.3: Singapore to Bangkok Flights for Singapore Airlines and Scoot Singapore Airlines Daily Flights 8:15am o 1:00pm Daily 11:40am o 4:30pm Daily

Scoot Daily Flights 12:55am o 5:30am Daily 5:55am o 10:30am Daily 8:30am o 1:15pm Daily 11:00am o 3:45pm Daily 12:30pm o 5:05pm Daily 1:20pm o 6:05pm Daily 1.5opm o 6:35pm Daily

Source: Google search “flights Singapore to Taipei” September 20, 2019

Table 5.4: Singapore to Taipei Flights for Singapore Airlines and Scoot coefficient attached to Scoot ASKs in the following way:32 Cannibalisation Elasticity = -0.0000121 u Scoot ASKt You will notice that we only multiply the coefficient attached to Scoot ASKs by Scoot ASKs to obtain the cannibalisation elasticity, and there is no division by Singapore PRASK, because the left-hand side of the PRASK equation is in logarithms. The level of quarterly Scoot ASKs at the time of writing is 7.724m. The cannibalisation elasticity at current Scoot ASKs is thus: Cannibalisation Elasticity (ASKScoot = 7,724m) = -0.0000121×7724 = -0.093 This means that a 10% increase in Scoot ASKs will lead to a 0.93% decrease in Singapore Airlines passenger revenue per ASK. There is not enough data for Scoot airlines to determines its Own Yield Elasticity, but we will assume that it is the same as that for Singapore Airlines, which is the coefficient attached to the logarithm of SQ ASKs in the above regression equation, equal to -0.33. The passenger revenue of Singapore Airlines over the 12 months to June 2018 is SG$9,880m. Scoot ASKs over the same period is 30,896m and Scoot PRASK is 5.78 Singapore cents per ASK. If we substitute this information into equation (5.38), and assume there is no competitor reaction, then we obtain the following estimated impact on Singapore Group revenue for each 1 million increase in Scoot ASKs: οR ୋ 0.093 × 9,880 = (1 െ 0.33) × 0.0578 െ = 0.008986 30,896 οA୐ This result says that if the Singapore Airlines Group increases Scoot Airlines ASKs by 1b then the increase in Singapore Airlines Group revenue in millions of Singapore dollars is: 'Singapore Group Revenue =

οୖృ ο୅ై

u οA୐ = 0.008986 u 1,000 = SG$8.986m

Let us break down this estimate into its component parts:

32

This uses the logarithm rule of calculus associated with the function loge y = ax (that is, the dependent variable is in logarithms, ௗ௬ ଵ but the independent variable is a linear function and is not in logarithms). In this case the first derivative is: ௗ௫ × ௬ = ܽ. The elasticity of y to x is

ୢ୷ ୢ୶

୶

× ୷ = a × x.

Chapter 5

122

1.

1 × 0.0578 is the volume effect of an increase in Scoot ASKs, which says that if the airline Group increases Scoot ASKs by 1b then this will increase Singapore Airline’s Group passenger revenue by 0.0578 u 1,000 = SG$57.8m because the airline will fly more low-cost passengers;

2.

െ0.33 × 0.0578 is the effect of a reduction in Scoot PRASK on Group revenue, which says that if the airline increases Scoot ASKs by 1b this will require the airline to reduce Scoot PRASK, which in turn reduces Group passenger revenue by SG$19.074m; and

3.

െ

଴.଴ଽଷ×ଽ.଼଼଴ ଷ଴,଼ଽ଺

= -0.02974 is the impact of the 1b increase in Scoot ASKs on the PRASK of Singapore

Airlines mainline and thus the Group’s revenue, which is -SG$29.740m. The net effect on the Group’s passenger revenue of a 1b increase in Scoot ASKs is thus: +$57.8m – $19.074m – $29.74m = SG$8.986m In this case Singapore Airlines Group will earn more revenue if it decides to expand Scoot ASKs by 1b. 5.6.5.2 With External Competitor Reaction In the analysis above, we do not consider the impact of a change in the capacity of an external competitor. Let us suppose now that in response to an increase in Scoot’s ASKs by 1b, which represents an ASK increase of 3.2%, the low-cost carrier Air Asia increases its ASKs by 3.2%. This implies the competitor reaction elasticity is 1. We assume that this has the same impact on Scoot’s PRASK as Scoot ASKs, so that the Scoot cross-yield elasticity with respect to Air Asia is equal to -0.33. We also assume that Air Asia’s ASKs do not impact the PRASK of Singapore Airlines so that the Singapore cross-yield elasticity with respect to Air Asia ASKs is 0. Substituting this additional information into equation (5.38) and we obtain: οR ୋ 9,880 = [1 െ 0.33 െ 0.33 × 1] × 0.0578 + × [െ0.093 + 1 × 0] = െ0.01009 30,896 οA୐ We can see in this case that the sign of the impact on passenger revenue has changed so that it is negative. A 1b increase in Scoot ASKs will reduce Singapore Group passenger revenue by: 'Singapore Group Revenue =

οୖృ ο୅ై

u 1,000 = -0.01009 u 1,000 = -SG$10.09m

Let us now assume that the expansion in Scoot Airways capacity by 1b ASKs now causes Air Asia to reduce its ASKs by 3.2% in response to lower yields. In this case the competitor reaction elasticity is -1, since Air Asia reduces ASKs by 3.2% in response to a 3.2% increase in Scoot ASKs. We further assume that each 1% decrease in Air Asia ASKs will increase Singapore Airline’s PRASK by 0.2%, which means the Singapore Airlines cross-yield elasticity with respect to Air Asia ASKs is -0.2. The impact of an increase in Scoot ASKs by 1 unit on Singapore Group passenger revenue in this case using (5.38) is then: οୖృ ο୅ై

= [1 െ 0.33 + 0.33 × 1] × 0.0578 +

ଽ,଼଼଴ ଷ଴,଼ଽ଺

× [െ0.093 + 1 × 0] = 0.09202

In this case a 1b increase in Scoot ASKs causes an increase in Singapore Airlines Group passenger revenue by: 'Singapore Group Revenue =

οୖృ ο୅ై

u 1,000 = 0.09202 u 1,000 = +SG$92.02m

The reduction in Air Asia ASKs completely offsets the negative impact on Scoot’s PRASK of Scoot increasing ASKs so that the impact on Scoot revenue is a pure volume effect. In addition to this, the cannibalisation effect is more than offset by the favourable impact on SQ’s PRASK of a reduction in Air Asia’s ASKs. What this tells us is that the business case for introducing a low-cost carrier on a particular route that is also serviced by the group’s full-service carrier depends on whether the low-cost capacity can inhibit the capacity growth of external competitors on the route. In other words, the benefits to the group will depend on whether the use of the low-cost carrier as a defensive strategy is effective.

Quiz 5-4. Cannibalisation Effects 1. On the Singapore (SIN) to Bali (DPS) route there are five carriers that provide direct services – Jetstar, Singapore Airlines, Scoot, Indonesia Air Asia, and Garuda. In relation to Singapore Airlines services, use the capacity descriptions in the following list to best describe the capacity of Jetstar, Scoot, Indonesia Air Asia, and Garuda.

Medium Run Airline Revenue

(a) (b) (c) (d)

123

Internal low-cost competitor capacity. Internal full-service competitor capacity. External low-cost carrier capacity. External full-service carrier capacity.

2. On the London Gatwick (LGW) to Madrid (MAD) route there are six carriers that provide direct services – Norwegian Air International, easyJet, Iberia, Iberia Express, British Airways and Air Europe. In relation to British Airways services, describe the capacity of the remaining five carriers. Use the following list of descriptions: (a) Internal low-cost competitor capacity. (b) Internal full-service competitor capacity. (c) External low-cost carrier capacity. (d) External full-service carrier capacity. On the London Heathrow (LHR) to Barcelona (BCN) route there are just two carriers that service the route – Iberia and British Airways. The International Airlines Group wishes to expand Iberian capacity on the route by 10%. The Group has an estimate of the own yield elasticity of -0.5. It also believes that British Airways PRASK on the route will fall by 2%. Use this information to answer the following 3 questions. 3. By how much will Iberian PRASK change on the London Heathrow to Barcelona route? (a) -5% (b) +5% (c) -10% (d) -2% 4. What does International Airlines Group believe is the elasticity of British Airways PRASK to a change in Iberian capacity? (a) -0.5 (b) -0.2 (c) -0.1 (d) +0.5 5. How would you define the capacity of British Airways relative to Iberian? (a) Internal low-cost competitor capacity. (b) Internal full-service competitor capacity. (c) External low-cost carrier capacity. (d) External full-service carrier capacity. On the Sydney (SYD) to Melbourne (MEL) route there are just four carriers that service the route – Qantas Mainline, Jetstar, Virgin Australia, and Tiger Airways. The Virgin Group wishes to expand the capacity of Tiger Airways by 10%. Qantas Group reacts by increasing the capacity of Jetstar by 5%. Qantas Mainline and Virgin Australia capacity remains unchanged. Answer the following 8 questions using this information. 6. How would you define the capacity of Qantas Mainline relative to Tiger Airways? (a) Internal low-cost competitor capacity. (b) Internal full-service competitor capacity. (c) External low-cost carrier capacity. (d) External full-service carrier capacity. 7. How would you define the capacity of Jetstar relative to Tiger Airways? (a) Internal low-cost competitor capacity. (b) Internal full-service competitor capacity. (c) External low-cost carrier capacity. (d) External full-service carrier capacity. 8. How would you define the capacity of Virgin Australia relative to Tiger Airways? (a) Internal low-cost competitor capacity. (b) Internal full-service competitor capacity. (c) External low-cost carrier capacity. (d) External full-service carrier capacity.

Chapter 5

124

9. What is the competitor reaction elasticity of Jetstar? (a) 0.1 (b) 1 (c) -1 (d) 0.5 10. How would you best describe the impact of Tiger’s capacity expansion on Qantas PRASK? (a) Cannibalisation elasticity. (b) Own yield cannibalisation elasticity. (c) Own yield elasticity. (d) Cross-yield elasticity. 11. How would you best describe the impact of Tiger’s capacity expansion on Virgin’s PRASK? (a) Cannibalisation elasticity. (b) Own yield elasticity. (c) Cross-yield elasticity. (d) Competitor reaction elasticity. 12. How would you best describe the impact of Tiger’s capacity expansion on Jetstar’s PRASK? (a) Cannibalisation elasticity. (b) Own yield elasticity. (c) Cross-yield elasticity. (d) Competitor reaction elasticity. 13. Which airline’s PRASK do you expect will be affected the most? (a) Tiger (b) Jetstar (c) Virgin (d) Qantas The full-service airline Iberia and the low-cost carrier Vueling Airlines operate services on the city pair Charles De Gaulle (CDG) to Barcelona (BCN). Both airlines are a part of the International Airlines Group (IAG). IAG is considering increasing the capacity of Vueling by 10% on CDG-BCN. This is estimated to reduce Vueling PRASK by 4% and it is estimated to reduce Iberia PRASK by 1%. There are assumed to be no competitors on the route. The Vueling PRASK is 11.52 Euro cents, the Iberia PRASK = 13.75 Euro cents, Vueling ASKs on the city pair over a year are 450m and Iberia ASKs over a year are 100m. Use this information to answer the following three questions. 14. What is the Vueling Own Yield Elasticity? (a) -0.3 (b) -0.4 (c) -0.5 (d) -0.6 15. What is the Cannibalisation Elasticity? (a) -0.1 (b) -0.2 (c) -0.3 (d) -0.4 16. What is the change in International Airlines Group Revenue in millions of Euro per year per Vueling ASK? ଴.ଵଷ଻ହ×ଵ଴଴ (a) [1 െ 0.4] × 0.1152 െ × 0.1 (b) െ0.4 × 0.1152 െ

ସହ଴ ଴.ଵଷ଻ହ×ଵ଴଴

× 0.1

ସହ଴ ଴.ଵଷ଻ହ×ଵ଴଴

(c) [1 െ 0.4] × 0.1152 +

(d) [1 െ 0.4] × 0.1152 +

ସହ଴ ଴.ଵଷ଻ହ×ଵ଴଴ ସହ଴

× 0.1 × [1 െ 0.1]

17. Now assume in questions 14. to 16. that on the city pair CDG-BCN Air France is an external competitor to both airlines. The elasticity of Vueling yields to Air France capacity is 0 and the elasticity of Iberia yields to Air France capacity is -0.2. In response to the increase in Vueling capacity, Air France will increase capacity by 5%. Use this information to determine the impact on IAG Group passenger revenue of a 10% increase in Vueling capacity.

Medium Run Airline Revenue

125

5.7 Is it Realistic to Believe that Airline Revenue Can Fall, when ASKs Increase? One of the important implications of the use of a linear and negative relationship between PRASK and own capacity is that an increase in capacity could lead to a reduction in passenger revenue. The intuition is that there is a level of capacity above which an airline’s revenue will begin to fall as capacity is increased. This is because the own yield elasticity turns from being inelastic to capacity to elastic. In the case in which the yield elasticity is elastic, an increase in capacity leads to a reduction in PRASK that is proportionately greater than the increase in ASK that caused it. For example, if the own yield elasticity is -1.2 and own ASKs increase by 10% then this means that PRASK falls by 2% more than ASKs increase, resulting in a reduction in revenue of 2%. In some ways it is difficult to conceive that an increase in production can result in lower revenue. In the case of airlines, however, there are many instances in recent times in which an airline has lifted capacity, and this has coincided with a reduction in revenue. These are presented in Figures 5-8 through to 5-13 below, in the case of a variety of airlines across the Asia Pacific, Europe and the Americas. Air NZ Revenue (NZ$ m)

$2,533

$2,535 $2,530

$2,525

$2,525 $2,520 6 Months to Jun-16 Air NZ ASKs (m)

21,000 20,500 20,000 19,500 19,000

6 Months to Jun-17

20,762 19,699 6 Months to Jun-16

6 Months to Jun-17

Source: Airline Intelligence and Research Database 2021

Fig. 5-8: Air New Zealand Revenue and Capacity China Eastern Revenue (RMB m)

51,000

50,776

50,500 50,000

49,598

49,500 49,000 6 Months to December 2014

6 Months to December 2015

China Eastern ASK (m)

13,500 13,000 12,500 12,000 11,500 11,000 10,500

13,148 11,647

6 Months to December 2014 Source: Airline Intelligence and Research Database 2021

Fig. 5-9: China Eastern Revenue and Capacity

6 Months to December 2015

Chapter 5

126 China Southern Revenue (RMB m)

58,400

58,359

58,350

58,293

58,300 58,250 6 Months to Dec-14

6 Months to Dec-15

China Southern ASK (m)

130,000 120,000 110,000 100,000

122,681 111,423

6 Months to Dec-14

6 Months to Dec-15

Source: Airline Intelligence and Research Database 2021

Fig. 5-10: China Southern Revenue and Capacity ANA Revenue (JPY b)

400 395 390 385 380

396 389

3 months to Dec-16 ANA ASKs (JPY m)

30,641

31,000 30,000

3 months to Dec-17

29,675

29,000 3 months to Dec-16

3 months to Dec-17

Source: Airline Intelligence and Research Database 2021

Fig. 5-11: All Nippon Airways Revenue and Capacity United Revenue (USD m)

8,650 8,600 8,550 8,500 8,450

8,603 8,528

3 months to Sep-16

3 months to Sep-17

United ASM (m)

72,000 70,000 68,000 66,000

70,083 68,074

3 months to Sep-16 Source: Airline Intelligence and Research Database 2021

Fig. 5-12: United Airlines Revenue and Capacity

3 months to Sep-17

Medium Run Airline Revenue IAG Revenue (EUR m)

5,100 5,050 5,000 4,950 4,900 4,850

127

5,078 4,934

3 months to Mar-16

3 months to Mar-17

IAG ASK (m)

69,000 68,000 67,000 66,000 65,000

68,304 66,151

3 months to Mar-16

3 months to Mar-17

Source: Airline Intelligence and Research Database 2021

Fig. 5-13: International Airlines Group Revenue and Capacity For example, in the case of Air New Zealand in Figure 5-8, the airline’s revenue fell from NZ$2.533b down to NZ$2.525 between the 6 months to June 16 and the 6 months to June 2017. Over the same period, Air New Zealand lifted total network level ASKs from 19,699 to 20,762. This type of negative relationship can occur if the New Zealand economy is in recession, but between 2014 and 2017 the New Zealand economy grew at the following rates, 2014 = 3.2%, 2015 = 4.2%, 2016 = 4.1% and 2017 = 3.0%. The average growth rate for the New Zealand economy between 1980 and 2017 is 2.7%, which is the long run average growth rate. We can clearly see that during the period over which Air New Zealand revenue fell at the same time as its ASKs increased, aviation demand was strong. We can therefore say that it was not weak market demand that is likely to have caused the weakness in revenue, rather it was Air New Zealand growing capacity on the right-hand side of the turning point of its revenue function. Figures 5-9 to 5-13 provide similar examples in the case of China Eastern, China Southern, All Nippon Airways, United Airlines and IAG. The dominant reason for the negative relationship between revenue and airline ASKs found across a wide range of airlines is that passenger demand is naturally relatively inelastic to changes in airfares. This means that the elasticity of PRASK to a change in ASKs is naturally a high negative number or high in absolute terms. The reason for the inelasticity of airline demand to fares largely follows from the fact that air travel is just one product or service that is purchased by the traveller when travelling by air. The other goods or services include accommodation, land transport (car hire, buses, taxis, rail, ferries), food and beverages, entertainment, gambling, and retail goods (see also chapter 3). Spending on air travel varies between 15% and 30% of total travel spend. This means that when an airline expands capacity and it is required to reduce fares to fill the plane, this fare reduction only affects 15% to 30% of the total price of travel of the average consumer. This means that travel can be highly responsive to the total travel price but inelastic to the components of that price, such as airfares, as discussed in detail in Chapter 3.

PART D: AVIATION COST

CHAPTER 6 AIRLINE COST

Airlines face a wide variety of costs. The main cost items are fuel, manpower or staff costs, the wear and tear of aircraft that an airline owns, the rental paid to aircraft leasing companies in the case of leased aircraft, maintenance costs, ground handling costs and airport charges. Each cost item is driven by a price component and a volume component. For example, the cost of employing pilots depends on the average wage paid to pilots, which is the price component, and the number of pilots or the hours worked by pilots, which is the volume component. The cost of airline fuel is equal to the price of jet fuel, which is the price component, and the quantity of jet fuel consumed, which is the volume component. The fees that an airline pays to airports is equal to the airport charge, which is the price component, multiplied by the number of landed passengers, which is the volume component. Different cost types are driven by different volume drivers. Fuel costs are driven by airline capacity. Airport charges are driven by the number of passengers that the airline carries. Maintenance costs will depend on the number of hours that are flown by the aircraft. The fact that different cost categories are driven by different volume drivers means that it is difficult for airlines to construct a single, infallible method of comparing their unit cost to the unit cost of other airlines. The rapid expansion of low-cost carriers such as Ryanair, Southwest, Spirit, easyJet, JetBlue, Jetstar, Air Asia, Indigo, Volaris, Peach, Vanilla, Allegiant and Wizz Air has brought more attention to the best way of measuring unit cost, and the best way to compare unit cost across airlines, especially low-cost carriers. This is because low-cost airlines, as the low-cost title suggests, strive to be the lowest unit cost carrier in the markets in which they operate. Their aim is to be the lowest cost carrier because the lowest unit-cost airline in a market or on a route can offer the lowest fare, and the low-cost airline with the lowest fare usually captures the greatest share of the leisure passenger market. How do low-cost carriers normally compare unit cost? They do this by computing what is called CASK or the cost per available seat kilometre. While this measure of unit cost is not perfect, it is just about the best measure that can be used to compute and compare unit costs across airlines, especially low-cost carrier airlines. An important aim of this chapter is to understand how CASK is calculated, understand the forces that may make it differ across airlines, and understand any adjustments that may need to be made for a meaningful comparison of unit cost across airlines. We did introduce CASK in chapter 2, but in this chapter, we will examine it in much more detail. I hope you enjoy this chapter on airline costs.

6.1 What are the Different Costs that Airlines Face? In this section we provide a detailed list of the cost items that are typically presented as part of an airline’s profit and loss statement or the notes to that profit and loss statement. Table 6-1 below presents the reported airline cost item in the first column and a brief description of those cost items in the second column. The table splits the cost items into major and minor items. The major cost items account for more than 80% of airline costs. To illustrate the importance of the different cost categories presented in Table 6-1, we present a breakdown of the operating expenses that are reported by three different airlines - Qantas Group, Singapore Airlines and Turkish Airlines. All three airlines have some of the best cost reporting in global aviation. Table 6-2 below presents the operating cost breakdown found in the Qantas Group 2020 Databook report, along with a bar chart of those costs in Figure 6-1 below. We can see in both the table and the figure that the major Qantas Group cost items in order of importance are manpower, fuel, depreciation of aircraft and non-aircraft assets, impairment of assets, and route navigation and landing charges (shortened to airport charges in the case of Figure 6-1). These five cost items alone represent 73% of the Qantas Group’s costs in 2020, with the only major cost category not reported in Table 6-1 being the asset impairment cost category. Impairment of assets is not a cost category that is presented in Table 6-1 because it doesn’t regularly feature in reported costs. It will feature in reported costs from time-to-time when there is a major shock to the global aviation market, caused by, for example a major economic downturn or a pandemic. In this case the aircraft assets that sit on the balance sheet of the airline experience a material reduction in market value, which must be recorded as an asset impairment cost in the non-operating component of the profit and loss statement of the airline. Let me present an example of the type of costs that will be included in asset impairment. Let us suppose that the Qantas Group, hypothetically, purchased a B787-800 aircraft for A$200m in 2020. The airline expected that the aircraft’s value would depreciate by 5% or A$10m in the first year, and so the written down value of the aircraft that the airline would expect to appear in its balance sheet at the start of 2021 is A$190m. The Coronavirus, however, significantly reduces the demand for air travel, which in turn means that aircraft all around the world are stood down from operations. To sure-up cash balances, many airlines offer their aircraft for sale in the global second-hand aircraft market. This increase in the supply of aircraft leads to a reduction in the second-hand price of aircraft, including the second-hand price of B787-800 aircraft. The price of a one-year-old B787-800 aircraft in the global second-hand

Airline Cost

Airline Cost Category Major Cost Categories

Brief Description

Fuel

The cost of jet fuel used by aircraft, and diesel and gasoline used by land transport. The wages that are paid to cabin crew. The wages that are paid to pilot and flight engineers. Wear and tear of owned aircraft. Payments made to aircraft lessors for the rental of aircraft.

Cabin Crew Salaries Tech Crew Salaries Aircraft Depreciation Operating leases and capacity hire Landing Fees Route Navigation Ground Handling Maintenance, Repair and Overhaul Non-Operational Manpower costs General Overhead Commissions Finance Charges Advertising and marketing Crew Expenses and Meals Station Engineering Crew Training Passenger Air Meals Customer Reservation System Fees Terminal Navigation Minor Cost Categories

Fees paid by the airline to airports for use of the airport’s aeronautical assets and services. Fees paid to air traffic control for navigation whilst the aircraft is flying, including overfly charges for international services. Include costs such as cleaning, refuelling, waste removal, baggage and freight handling, ground marshalling, aerobridge services, check-in, and gate services. Engineering labour, hanger and spare part and material costs associated with major checks on aircraft. Includes wages paid to back-office staff such as the CEO and CEO, and those employed in various departments such as human resources, finance, strategy, audit, government relations, communications, and frequent flyers. Includes building rental and utilities. Payments that airlines make to agents to incentivise them to sell tickets to passengers and space to freight distributors. Interest and other payments made on borrowed finance and corporate bonds. Costs associated with the production of advertising and marketing material and purchase of media time. The cost of accommodation, land transport (airport to hotel and return, base to airport and return), uniforms and meals for crew. Costs associated with paying staff and engineers to inspect the aircraft when it is waiting at the gate. This involves inspecting things like wheels, brakes and fluid levels and any running repairs that are needed. Simulator and other training costs. The cost of the meals that are prepared for passengers. Fees paid by the airline to relevant entities for the use of systems that present seat availability and fares on offer. Fees paid to air traffic control for navigation around the airport terminal.

Payment Card Charges

Merchant service fees paid by the airline to banks.

Frequent Flyer Costs

Costs associated with administering the frequent flyer program of the airline.

Depreciation, Amortisation and Lease of Non-Aircraft

Depreciation, amortisation and lease of buildings and other non-aircraft plant and equipment, including the airline’s fleet of land vehicles.

Inflight Passenger Expenses

Other passenger costs, such as costs associated with the cutlery that passengers use, the lavatory consumables and the cost of duty-free products that are sold.

Insurance

The costs associated with insurance premiums paid against possible injury, sickness or death of passengers and damage to aircraft. Customer-facing staff employed at airports, such as check-in staff, security staff, ticket sales staff and other staff. Airline staff employed in regional offices and the lease or rent payments for buildings that house these resources. The costs associated with running an airport club lounge, including staff, consumables, maintenance and repair, space rental and depreciation costs.

Domestic Airport Support Port Support Overhead Airport Lounge Expenses

Table 6-1: List of Major and Minor Airline Cost Categories

131

Chapter 6

132

Qantas Group Databook Expenditure Items FY20 $M 3,646 2,895 2,045 1,456 1,373 845

Manpower and staff related Fuel Depreciation and amortisation Impairment of assets and related costs Route navigation and landing fees Maintenance (including materials and labour) Redundancy and related costs Commissions and other selling costs Passenger expenses Airmeals Ground handling Capacity hire Crew expenses Property Marketing and advertising Airport security charges Short-term non-aircraft lease payments Other Total

565 506 317 352 317 268 176 176 160 106 49 382 16,694

Source: Qantas Group Databook 2020

Table 6-2: Qantas Group Operating Cost Breakdown 12 Months to 30 June 30, 2020

Cost Items % Total Cost 25% 23.3% 18.5%

20% 15%

13.1% 9.3% 8.8%

10%

5.4%

5%

3.6% 3.2% 2.4% 2.3% 2.0% 2.0% 1.7% 1.1% 1.1% 1.0% 0.7% 0.3%

0% Lease

Security

Marketing

Property

Crew

Capacity Hire

Ground Handling

PAX Expenses

Airmeals

Other

Commissions

Redundancy

Maintenance

Airport Charges

Asset Impair.

Depreciation

Fuel

Staff

Source: Qantas Group Databook 2020

Fig. 6-1: Major Cost Breakdown – Qantas Group FY20 market falls to A$150m by the start of 2021. The difference between the written down value of the Qantas Group B787800 and the price in the second-hand market is therefore A$40m. This amount represents an impairment of the Qantas Group Boeing 787-800 asset, which is a cost item that will sit in the non-operating component of the profit and loss statement of the Qantas Group. Table 6-3 presents the expenses that appear in the annual profit and loss statement of the Singapore Airlines Group for the 12 months to March 31, 2020, with the corresponding cost bar chart presented in Figure 6-2 below. The table and figure indicate that the top five cost categories of Singapore Airlines are fuel, manpower, depreciation, handling, and landing, parking and overfly charges (shortened to ‘airport’ in Figure 6-2). These five cost categories contributed 72% to the operating cost of the airline over the 12 months to March 31, 2020, with the three most important Singapore

Airline Cost

133

Singapore Airlines CONSOLIDATED PROFIT AND LOSS ACCOUNT FOR THE FINANCIAL YEAR ENDED 31 MARCH 2021 (in $ million) EXPENDITURE FY 2019/20 4,636.5 2,563.6 2,134.2 1,276.7 886.4 835.4 709.8 539.0 489.8 334.2 187.7 172.4 79.2 61.6 57.5 50.2 14.2 888.4 15,916.8

Fuel costs Staff costs Depreciation Handling charges Landing, parking, and overflying charges Aircraft maintenance and overhaul costs Fuel hedging ineffectiveness Inflight meals Commission and incentives Advertising and sales costs Other passenger costs Crew expenses Rentals on leased aircraft Impairment of amount owing by a joint venture company Amortisation of intangible assets Company accommodation and utilities Impairment of property, plant, and equipment Other operating expenses

Source: Singapore Airlines SGX Announcement 2020.

Table 6-3: Singapore Airlines Operating Cost Breakdown Full Year Ended March 31, 2020

Cost Items % Total Cost 35% 30%

29.1%

25% 20%

16.1% 13.4%

15%

8.0%

10% 5%

5.6% 5.6% 5.2% 4.5% 4.5%

3.1% 2.1%

0%

1.2% 1.1% 0.5% 0.5% 0.4% 0.3% Utilities

Amortisation

Asset Impair.

Rentals

Crew

PAX Costs

Advertising

Commissions

PAX Meals

Fuel Hedging

Maintenance

Other

Airport

Handling

Depreciation

Staff

Fuel

Source: Singapore Airlines SGX Announcement 2020

Fig. 6-2: Major Cost Breakdown Singapore Airlines 12 Months to March 31, 2020 Airlines cost items the same as the three most important Qantas Group cost items, specifically fuel, manpower and depreciation. A major cost category that is included in Singapore Airlines expenses that does not appear to be important in the case of Qantas Group expenses over the period examined in Figures 6-1 and 6-2, is fuel hedge ineffectiveness. Like asset impairment costs, the cost of fuel hedge ineffectiveness does not regularly appear as a reported cost in the case of Singapore Airlines, and most airlines around the world, which is why it is not listed in Table 6-1. Let me provide a simple and brief explanation of the cost of fuel hedge ineffectiveness.

Chapter 6

134

Most airlines around the world hedge their exposure to jet fuel prices by using financial instruments such as swaps, options, and collars. Sometimes those hedges are effective in protecting the airline against higher jet fuel prices, and sometimes they are ineffective. When they are ineffective according to accounting rules the airline experiences fuel hedge losses, and these losses must be reported in the non-operating component of the profit and loss statement of the airline. Fuel hedging is typically ineffective, in simple terms, when the airline locks in a certain price for jet fuel, but the market jet fuel price falls, resulting in the airline experiencing a hedge loss. The extent of the ineffectiveness of Singapore Airline’s fuel hedging is also influenced by the airline hedging more barrels of fuel than necessary. This is also called over-hedging. This can occur when the airline consumes less fuel than expected. Singapore Airlines, like many airlines across the world, is likely to have consumed less jet fuel than expected over the 12 months to March 31, 2020, because the Coronavirus pandemic forced the airline to shutdown much of its flight operations. Table 6-4 below presents a breakdown of the operating costs of Turkish Airlines which is obtained from its 12 months to December 31, 2019, financial results report, in the notes section. TÜRK HAVA YOLLARI ANONÏM ORTAKLIöI AND ITS SUBSIDIARIES Notes to the Condensed Interim Financial Statements As at And for the Three-Month Period Ended 31 March 2021 (All amounts are expressed in Million US Dollars (USD) unless otherwise stated.) 18. EXPENSES BY NATURE

Fuel Expenses Personnel expenses Depreciation and amortisation charges Ground services expenses Aircraft maintenance expenses Passenger services and catering expenses Airport expenses Air traffic control expenses Commissions and incentives Wet lease expenses Reservation systems expenses Advertisement and promotion expenses Rents Service expenses Insurance expenses Taxes and duties Transportation expenses IT and communication expenses Aircraft rent expenses Consultancy expenses Systems use and associateship expenses Other expenses

1 January – 31 December 2019 3,873 2,067 1,521 815 791 622 623 553 504 284 267 175 85 83 54 52 44 41 27 21 9 133 12,644

Source: Turkish Airlines Financial Results Report 2019

Table 6-4: Turkish Airlines Operating Cost Breakdown 12 Months to December 31, 2019 Table 6-4 indicates that the top five cost categories of Turkish Airlines are fuel, manpower, depreciation, ground handling and maintenance, which is a more conventional set of top 5 airline cost categories, likely reflecting the fact that it is derived from a period that is just prior to the start of the Coronavirus pandemic. These top five cost categories generated 72% of total cost for Turkish Airlines over the calendar year 2019, as also indicated in the bar chart of Figure 6-3 below. The top three Turkish Airlines cost categories, fuel, manpower and depreciation, are the same as those for Singapore Airlines and the Qantas Group.

6.2 Changing Importance of Fuel and Non-Fuel Costs The common thread that we see in each of the Qantas, Singapore Airlines and Turkish Airlines cost breakdowns presented in section 6.1 is the importance of fuel, staff, and depreciation costs as the top three cost categories, while the top 5 cost categories generate more than 70% of operating costs for all three airlines. The relative importance of fuel and non-fuel costs vary considerably over time, however, because of changes in the price of jet fuel. To demonstrate

Airline Cost

135

35%

30.6%

Cost Items % Total Cost 30%

20%

0.3%

0.3%

Taxes

Transportation

IT

Systems use

0.4%

Insurance

0.1%

0.4%

Services

Consulting

0.7%

Property Rent

0.2%

0.7%

Other

Aircraft rent

1.1%

Advertising

0.2%

1.4%

Reservation Systems

Commissions

2.1%

4.0%

Air Traffic Control

Lease

4.4%

Airport

2.2%

4.9%

PAX Services

Maintenance

5%

4.9%

6.3%

10%

Ground Handling

15%

6.4%

12.0%

16.3%

25%

0% Depreciation

Staff

Fuel

Source: Turkish Airlines Financial Results Report 2019

Fig. 6-3: Major Cost Breakdown – Turkish Airlines 12 Months to December 31, 2019 this, consider Figure 6-4 below, which presents a time series of Qantas Group non-fuel costs as a percentage of total costs between 2000 and 2020, with the cost of major asset impairment excluded. Qantas NonFuel Cost % Total Cost 95% 90%

90% 85%

81%

80% 75% 72%

70%

Jun-20

Jun-19

Jun-18

Jun-17

Jun-16

Jun-15

Jun-14

Jun-13

Jun-12

Jun-11

Jun-10

Jun-09

Jun-08

Jun-07

Jun-06

Jun-05

Jun-04

Jun-03

Jun-02

Jun-01

Jun-00

Source: Airline Intelligence and Research Database 2021

Fig. 6-4: Non-Fuel Costs as a Percentage of Qantas Group Total Cost The importance of non-fuel costs in total costs has declined for the Qantas Group from 90% over the 12 months to June 30, 2000, to 72% over the 12 months to June 30, 2014. Much of this movement is related to the trend increase in jet fuel prices, with the spike in jet fuel prices just prior to the GFC in 2008 sending non-fuel costs as a percentage of total costs to a low of 72%. The importance of non-fuel costs has since recovered to 81% by the 12 months to June 30, 2020, in response to a trend decline in jet fuel prices. Figure 6-5 below presents non-fuel costs as a percentage of total cost in the case of Southwest Airlines over the longer annual timeframe 1972 to 2019.

Chapter 6

136

Southwest Airlines Non-Fuel Cost % Total Cost 95% 90%

88.8%

87.5%

85% 80%

77.7%

75% 70% 65% 60%

62.3%

59.5%

55%

2019 2018 2017 2016 2015 2014 2013 2012 2011 2010 2009 2008 2007 2006 2005 2004 2003 2002 2001 2000 1999 1998 1997 1996 1995 1994 1993 1992 1991 1990 1989 1988 1987 1986 1985 1984 1983 1982 1981 1980 1979 1978 1977 1976 1975 1974 1973 1972 Source: Airline Intelligence and Research Database 2021

Fig. 6-5: Non-Fuel Costs as a Percentage of Southwest Airlines Total Cost We can see in Figure 6-5 that the importance of non-fuel costs in the case of Southwest Airlines has cycled through highs and lows, corresponding with low and high cycles in jet fuel prices. Between 1973 and 1981 the importance of non-fuel costs at Southwest dropped from 87.5% down to an absolute low of 59.5%, due to a steep climb in jet fuel and oil prices. Between 1981 and 1998 the non-fuel cost percentage of total cost peaked at 88.8% by 1998. The importance of non-fuel costs then fell once again as jet fuel prices cycled upward, falling to just over 62% by 2011, at which point the importance of non-fuel costs in total costs began to climb once again to stand at 78% in 2019. Figure 6-6 provides time series movements in non-fuel costs as a percentage of total costs for the Lufthansa Group. Lufhansa Group Non-Fuel Cost as % Total Cost 94% 92%

92.5%

90% 88% 86% 84%

82.0%

82% 80% 78% 76%

74.8%

74%

2019 2018 2017 2016 2015 2014 2013 2012 2011 2010 2009 2008 2007 2006 2005 2004 2003 2002 2001 2000 1999 1998 1997 1996 1995 1994 1993 1992 1991 1990 1989 1988 Source: Airline Intelligence and Research Database 2021

Fig. 6-6: Non-Fuel Costs as a Percentage of Total Cost Lufthansa Group The Lufthansa Group has much larger operations in the non-flying parts of the airline business than most other airlines, including businesses such as catering, IT and maintenance, repair, and overhaul, which raises the importance of its nonfuel costs. Lufthansa non-fuel costs peaked at a high of 92.5% over the 12 months to December 31, 1999 but fell to as low as 74.8% by 2012 in response to surging jet fuel prices. The non-fuel cost percentage has since returned to 82% by 2019 as jet fuel prices have trended lower.

Airline Cost

137

Quiz 6-1: Cost Categories 1. In order of importance, which costs are approximately the most significant for airlines? (a) Maintenance, Depreciation and Amortisation, Manpower, Fuel, Route and Navigation charges, Operating Lease. (b) Route and Navigation Charges, Manpower, Operating Lease, Fuel, Depreciation and Amortisation, Maintenance. (c) Fuel, Manpower, Depreciation and Amortisation, Route and Navigation Charges, Operating Lease, Maintenance. (d) Depreciation and Amortisation, Maintenance, Manpower, Route and Navigation charges, Operating Lease, Fuel. 2. What are route and navigation costs for airlines? (a) Payments made to the Tax Office by the airline. (b) Payments made by the airport to the airline. (c) Payments made by the airline to the government agency that looks after air and airport terminal services. (d) Payments made by the airline to the airport for the use of the airport terminals, aprons, taxiways, and runway. 3. What is an example of a passenger ground handling cost? (a) The cost associated with cleaning the rest rooms on the aircraft after a long-haul flight. (b) The cost associated with loading the belly of the aircraft with passenger luggage. (c) The cost of replacing the passenger consumables on the aircraft, such as soap, toilet paper, pillows, blankets, and headphones. (d) All the above. 4. What is an aircraft operating lease cost? (a) The payments that an airline makes to an aircraft manufacturer when it is paying-off the aircraft that it has bought. (b) The payments that an airline makes to a bank to pay for the loan that it has used to pay for its aircraft. (c) The money that an airline pays to a lessor for the use of aircraft owned by the lessor. (d) The money that an airline pays to another airline for the use of aircraft owned by a leasing company. 5. Commission costs are incurred by airlines because: (a) The airline tries to provide incentives to passengers to sell its seats and freight space. (b) The airline tries to provide incentives to other airlines to sell its seats and freight space. (c) The airline tries to provide incentives to agents to sell its seats and freight space. (d) The airline tries to provide incentives to airports to sell its seats and freight space. 6. The staff costs of an airline employee who helps passengers at the check-in kiosks at local airports would be included in which cost category? (a) Airport lounge expenses. (b) Passenger ground handling. (c) Domestic airport support. (d) Airport staff expenses.

6.3 The Link Between Productivity, Resource Prices, and Airline Unit Cost 6.3.1 Schematic Representation of the Linkage An airline employs resources such as pilots, cabin crew, engineers, jet fuel, a fleet of aircraft and airport services to produce output in the form of seats and freight air transport. The relationship between the input of resources and the output of the airline is determined by the productivity of those resources. The airline pays for the use of the resources based on a set of resource prices, which generates operating cost. Dividing the operating cost by the seat and freight output of the airline determines an airline’s unit cost. As a result, unit cost depends heavily on both the productivity of the resources it employs and the price it pays for labour, capital, energy, and other resources. This relationship between production, resource usage, resource prices, operating cost and unit cost is described in Figure 6-7 below. Figure 6-7 is a simple schematic representation of the way that airline unit cost is determined by the interaction of resource usage and airline output (ASK and AFTK production), simultaneously with the linkage between resource usage and cost. The middle row of Figure 6-7 describes the volume of resources that are employed by the airline, including hours of cabin crew and pilot labour, barrels of jet fuel, the number of fleet units, kilowatts of energy and litres of water usage, the number of airport movements and the labour time required by engineers. The use of these resources by the airline represents the quantity of resources used. This resource use translates into resource costs (or are translated into a dollar value) by multiplying the volume of resource by the price of those resources in the top row of Figure 6-7. For example, if the airline uses 30m barrels of jet fuel, which is the volume of resource, and pays a price of U$100 per barrel for fuel, which is the price of that resource, this results in a resource cost of US$3b. Resource prices also include the wages paid per hour, the price of aircraft, the price of energy and utilities (electricity, diesel, and water), and airport charges.

138

Aircraft Cost

Airline Output

Resource Inputs

Via Productivity of Fleet

Fleet

Via Aircraft Price

Via Productivity of Energy

Energy

Via Energy Prices

Energy Cost

Fig. 6-7: The Linkage between Airline Cost, Production and Unit Cost

Unit Cost

Via Productivity of Crew Labour

Cabin & Tech crew

Via Wage Rate

Cabin & Tech Crew Cost

Total Cost

Chapter 6

AFTKs

ASKs

Via Productivity of Fuel

Fuel

Via Fuel Prices

Fuel Cost

Engineers

Via Wage Rate

Engineers Cost

ATKs

Via Productivity of Airports Via Productivity of Engineers

Airports

Via Airport Charges

Airport Cost

Airline Cost

139

At the same time as the resource usage incurs costs it generates a certain level of production, including passenger production as described by ASKs and freight production as described by AFTKs, which combine to generate total production, ATKs. The amount of output produced is determined by the combined productivity of the resources that are employed by the airline (we will develop this point in more detail below when we consider resource productivity in section 6.3.2). Combining the cost of the resources with the production that the resource use generates, enables us to construct a measure of unit cost: Unit Cost =

େ୰ୣ୵ େ୭ୱ୲ା୉୬ୣ୰୥୷ େ୭ୱ୲ା୅୧୰୮୭୰୲ େ୭ୱ୲ା୊୳ୣ୪ େ୭ୱ୲ା୉୬୥୧୬ୣୣ୰ େ୭ୱ୲ ୅୧୰୪୧୬ୣ ୓୳୲୮୳୲

(6.1)

The most common denominator that is used in (6.1) for airline output is ASKs. If ASKs are used in the denominator this gives rise to the unit cost metric CASK or cost per available seat kilometre, which we described in some detail in chapter 2. An alternative to the use of ASKs in the denominator of (6.1) is the use of ATKs. This is the preferred denominator of (6.1) in the case of airline businesses that transport a considerable volume of air freight. If ATKs are used in the denominator of (6.1) then the unit cost measure of the airline is called CATK. In the case of jurisdictions that use miles rather than kilometres the relevant unit cost metrics are CASM and CATM.

6.3.2 Airline Production and Productivity 6.3.2.1 Airline Production Function The job of a passenger airline is to produce available seat kilometres and available freight tonne kilometres. The production of ASKs and AFTKs requires many important resource inputs, including but not limited to jet fuel, technical crew, cabin crew, engineers, ground handlers, airports, aircraft, management and analysts, energy, advertising hours, communication, and IT equipment. Management and analysts are a part of the non-operational labour of the airline, which includes staff such as the CEO, CFO, board members, and people who work in back-office functions such as yield managers, accountants, treasurers, auditors, lawyers, and salespersons amongst many others. The way that these resources combine to determine the level of airline output is referred to as the production relationship. In economics this relationship is described by a production function. The production function is a function that takes the quantity of resources used by the airline and converts it into a single measure of airline output, such as ASKs and AFTKs. In the case of airlines, it is not an easy task developing a production function that adequately captures the relationship between the resource inputs and airline output, particularly at a route level. This is because at the route level, the relationship between many of the resource inputs and the airline’s output is a fixed proportion or Leontief relationship.33 To illustrate this point, consider a Qantas flight from Sydney to Melbourne. This flight will operate with the following resource inputs (1) a 180 seat Boeing 737-800 aircraft, (2) two pilots, a captain and a second officer, who will fly for 1 and a half hours, (3) 5 cabin crew, who will each fly for one and a half hours, (4) 25 barrels of jet fuel, and (5) use of the runways, taxiways, apron areas and gates of Sydney and Melbourne Airport domestic terminals. This resource usage will generate available seat kilometres equal to ASKSYD-MEL = 180 u 706 = 127,080, where 706 km is the great circle distance between Sydney and Melbourne. In this case the resource inputs are used in the following fixed proportions: 3 pilot hours : 7.5 cabin crew hours : 1 aircraft : 25 barrels of jet fuel : 2 airports Ÿ 127,080 ASKs When production and resource usage occur in fixed proportions like this, an increase in any one of the inputs will not result in an increase in output. For example, if Qantas used one more cabin crew member on the Sydney to Melbourne flight this would not lead to an increase in ASKs on the flight. If the pilot added more jet fuel to the flight this also would not lead to an increase in ASKs. For ASKs to increase, the airline would need to increase most of its resource inputs in fixed proportions. For example, to double ASKs the airline would need 6 pilot hours, 15 cabin crew hours, 50 barrels of jet fuel and 4 movements at Sydney and Melbourne Airports. One of the complications associated with this fixed proportions approach to analysing airline production relationships is that to increase output it isn’t necessary to increase all inputs in fixed proportions. Some resources can remain fixed up until a certain limit. For example, continuing our Sydney to Melbourne illustration, we can double ASKs by assuming that there is a return flight from Melbourne to Sydney. For this return flight we require an additional 3 pilot hours, 7.5 cabin crew hours, 25 barrels of jet fuel and use of the Melbourne runway for take-off and the Sydney runway for landing. We can use, however, the same aircraft for the Melbourne-Sydney flight as was used for the SydneyMelbourne flight, which means we need to double most other inputs except for the number of aircraft that is used.

33

If you wish to learn more about the Leontief production relationship, which was discovered by the Nobel Prize winning economist Wassily Leontief, I recommend reading the excellent book by R.G.D. Allen. 1968. Macroeconomic Theory: A Mathematical Treatment, London: Macmillan on page 35. If you wish to read a more recent treatment of the Leontief Production Function, see R.E. Miller and P.D. Blair. 2009. Input-Output Analysis: Foundations and Extensions, second edition, New York, Cambridge University Press.

Chapter 6

140

In the Sydney-Melbourne-Sydney illustration above, it was not necessary to increase the number of fleet units to double ASKs because the return service from Melbourne to Sydney took place not long after the arrival of the aircraft on the flight from Sydney to Melbourne. If the airline were to use just one aircraft on Sydney-Melbourne, this would place significant restrictions on the frequency of services that can be flown, which in turn places restrictions on the number of ASKs that can be flown on the route. By utilising only one aircraft on the route this would restrict the airline to running return services every four hours, which for this busy route is not likely to be profit maximising. The extent to which ASKs can be increased will therefore depend on the airline’s desired frequency of services. This also further demonstrates the complexity of analysing the production relationship at the route level. While there are complexities associated with trying to understand relationships between resource inputs and output at the route level for an airline, some of this complexity disappears when examining the relationships at the network or total airline level. Let us illustrate this point by analysing the co-movement over time in the volume of key resource inputs employed by the U.S. low-cost carrier JetBlue. Figure 6-8 below presents the relationship between the number of block hours flown by JetBlue pilots and cabin crew and the number of barrels of jet fuel consumed by the airline over the calendar annual period 2001 to 2020. JetBlue Fuel Consumption (m gallons) 1,000

JetBlue Block Hours Block Hours

1,200,000 1,000,000 800,000 600,000 400,000 200,000 0

Fuel Consumption

800 600 400 200 0 2020

2019

2018

2017

2016

2015

2014

2013

2012

2011

2010

2009

2008

2007

2006

2005

2004

2003

2002

2001

Source: Airline Intelligence and Research Database 2021

Fig. 6-8: JetBlue Pilot and Cabin Crew Block Hours and Fuel Consumption – 2001 to 2020 We can see from Figure 6-8 that there is a strong relationship between block hours and jet fuel consumption across the period examined. In fact, the correlation between the two variables is 99.7% over the 19-year period. The two series clearly separate between 2007 and 2019 but it appears to be a step-separation in 2007, with the gap opening-up at that point and only narrowing slightly from around 2014 onwards. In 2019, there were 1,102,655 block hours for 885 million gallons of jet fuel consumed with a ratio of 803 gallons of jet fuel consumed per block hour. Figure 6-9 below presents a time series graph of the number of JetBlue airport departures against the average number of fleet units for calendar years between 2001 and 2019. Fleet Units

JetBlue Departures

Departures

421,000 371,000 321,000 271,000 221,000 171,000 121,000 71,000 21,000

Fleet

300 250 200 150 100 50 0 2019

2018

2017

2016

2015

2014

2013

2012

2011

2010

2009

2008

2007

2006

2005

2004

2003

2002

2001

Source: Airline Intelligence and Research Database 2021

Fig. 6-9: JetBlue Flight Departures and Fleet Units – 2001 to 2019 It shows a high level of co-movement between the two resource inputs, with a correlation of 99.9%. Gaps open-up between the two series from time to time, but these tend to narrow relatively quickly, reflecting the extent to which the airline flexes the utilisation of its aircraft up and down over time. In 2019 there were approximately 1,452 airport departures per fleet unit per year. The relationships that we observe in Figures 6-8 and 6-9 are relationships that can be

Airline Cost

141

adequately captured and modelled by analysing measures of resource productivity such as the average product. This is a topic to which we turn in the following sub-section. 6.3.2.2 Airline Productivity Airlines typically measure resource productivity by dividing the quantity of output by the quantity of resource input – this is called the average product of the input. The average product can be estimated for each input into production that is used. In the case of airlines, the key measures of average product that are computed are for fuel, fleet, and labour inputs. The average products of these airline resource inputs are computed in the following way: Average product of labour = Average product of fuel = Average product of fleet =

୅୴ୟ୧୪ୟୠ୪ୣ ୗୣୟ୲ ୏୧୪୭୫ୣ୲୰ୣୱ

୊୳୪୪ି୘୧୫ୣ ୉୯୳୧୴ୟ୪ୣ୬୲ ୗ୲ୟ୤୤ ୅୴ୟ୧୪ୟୠ୪ୣ ୗୣୟ୲ ୏୧୪୭୫ୣ୲୰ୣୱ ୆ୟ୰୰ୣ୪ୱ ୭୤ ୊୳ୣ୪ େ୭୬ୱ୳୫ୣୢ ୆୪୭ୡ୩ ୌ୭୳୰ୱ ୒୳୫ୠୣ୰ ୭୤ ୊୪ୣୣ୲ ୙୬୧୲ୱ×୒୳୫ୠୣ୰ ୭୤ ୈୟ୷ୱ ୧୬ ୲୦ୣ ଢ଼ୣୟ୰

(6.2) (6.3) (6.4)

The labour productivity metric at (6.2) tells us the average number of ASKs that the airline is likely to generate for each full-time equivalent staff member. It is typically interpreted in terms of millions of ASKs per FTE. The fuel productivity metric (6.3) is similarly interpreted, as the millions of ASKs generated by a barrel of jet fuel. The denominator of this expression need not be barrels of jet fuel, but can also be measured in litres, gallons or metric tonnes depending on the unit of measurement of fuel consumption used by the airline. The final productivity metric (6.4) measures the average number of hours that the fleet is used daily. Obviously, the upper limit of (6.4) is 24 hours and the lower limit is zero, with most airlines operating aircraft at between 8 and 15 hours per day. If an airline operates a relatively large freight business, which is typically when freight revenue as a proportion of total revenue exceeds 15% as is the case for airlines such as Singapore Airlines, Cathay Pacific and Korean Air, the productivity metrics (6.2) and (6.3) should replace Available Seat Kilometres in the numerator with Available Tonne Kilometres. Each of the measures (6.2) to (6.4) of average product attempt to determine the amount of input that is necessary to generate a certain amount of output. Productivity unambiguously increases when the airline generates the same output using fewer inputs and generates more output with the same inputs. Productivity can also increase if the airline uses more inputs and output increases by proportionately more than the increase in inputs. Alternatively, productivity increases when the airline decreases its use of inputs, with output falling by proportionately less than the reduction in inputs. In the case of many resources employed by airlines, the output measure that is relevant in ascertaining productivity is not always ASKs or ATKs. Consider the following examples. An airline adds another cabin crew member to a flight to increase service levels on the flight. Although this decision will not add to available seat kilometres it will add to the customer experience, which will in turn result in more heightened customer satisfaction. Increased customer satisfaction will result in stronger passenger demand, which will generate greater revenue either through stronger seat factors or better yields. In this case the output metric which measures the success or not of adding an extra cabin crew to a flight is not available seat kilometres but a metric that measures customer satisfaction, such as the net promoter score.34 The staff that work in back-office roles in the airline, such as in treasury risk management, government relations, corporate communications, network planning, economics, human resources, and yield management, have little to do with Available Seat Kilometres. Their work, however, is important for informing senior management so that they can make better decisions. Better decisions in turn lead to lower costs and/or stronger revenue. Using an ASK-related productivity metric will not capture these influences and will not capture the productivity of these back-office staff members. Superior measures of productivity include the extent to which this labour has resulted in better decision marking, additional revenue, or cost reduction.

Quiz 6-2: Airline Productivity, Resource Prices, and Unit Cost 1.

An airline is assumed for simplicity to employ only the variable input labour. The number of labour hours worked by its operational labour is described by the variable LO. The number of labour hours worked by its non-operational labour is described by the variable LN. The average wage rate paid by to its total labour force is w per hour. The production relationship between the labour hours employed and available seat kilometres is LO = 0.2 u ASK, where ASKs are defined in millions of available seat kilometres. The fixed cost of the airline business is assumed to be $1b. Use this information to determine the total cost of the airline as a function of the wage rate and ASKs. Also determine the unit cost of the airline by dividing total cost by ASKs.

34 The net promoter score is based on a survey that asks the following question: how likely is it that you will recommend the airline to a friend or colleague? In most cases this question is answered on a 0 to 10 scale. See the article by People Pulse 2019 for further explanation of the net promoter score.

142

Chapter 6

2. Which of the following is a measure of labour productivity? (a) Blocks hours per fleet unit per day. (b) Available seat kilometres per barrel of fuel consumed. (c) Full-time equivalent staff per available tonne kilometre. (d) Available seat kilometres per full time equivalent staff. 3. Which of the following is a measure of fleet productivity? (a) Blocks hours per fleet unit per day. (b) Available seat kilometres per barrel of fuel consumed. (c) Full-time equivalent staff per available tonne kilometre. (d) Available seat kilometres per full-time equivalent staff. 4. Which of the following is a measure of fuel productivity? (a) Blocks hours per fleet unit per day. (b) Available seat kilometres per barrel of fuel consumed. (c) Full-time equivalent staff per available tonne kilometre. (d) Available seat kilometres per full-time equivalent staff. 5. Which of the following represents an improvement in airline productivity? (a) When an airline consumes less fuel because there are fewer flights. (b) When an airline consumes the same fuel for fewer flights. (c) When an airline consumes more fuel for more flights. (d) When an airline consumes the same fuel for more flights. 6. A Vueling Airlines flight from Barcelona Airport (BCN) to the island of Ibiza (IBZ) uses 2 pilots, 5 cabin crew, 1 Airbus A320 aircraft with 180 seats, 20 barrels of jet kerosene, and two airport movements per flight. The block hours per flight is 1 hour. For this flight, determine the ratio of block hours to (1) pilot hours, (2) cabin crew hours, (3) fuel consumption and (4) aircraft movements. Find the average product of fuel, pilot and cabin crew staff, and fleet. Use http://www.gcmap.com/ to find the great circle distance for this flight, which can be used to find the average products. How does your use of the resource inputs change if the flight is a return flight? Explain your answer.

6.4 Airline Fuel Costs 6.4.1 Basic Analytics Fuel costs are a high proportion of total cost of an airline as indicated in section 6.2 above. It is also a complicated component of operating cost that is difficult to understand and to model. For this reason, we will look at this cost category in a little more detail than other cost categories. We start by presenting a simple analytical model of an airline’s fuel cost per available seat kilometre, or FCASK for short. This will provide some insight into the type of variables that cause FCASK, and fuel as a percentage of total cost, to change over time. An airline’s unhedged fuel costs has five main component parts (1) oil prices, (2) the jet-to-crude crack margin or refining margin, (3) the into-plane margin, (4) the US dollar bilateral exchange rate, and (5) fuel consumption. We can describe how these five components are joined together to determine fuel costs by using the following formula: Fuel Cost = (Oil Price + Refining Margin + Into-Plane Margin) u Fuel Consumption u Local Currency Price of 1 US Dollar

(6.5)

where Oil Price is the spot price of crude oil defined in US dollars per barrel, Refining Margin is the difference between the spot price of jet fuel defined in US dollars per barrel and the spot price of crude oil defined in US dollars per barrel (refer also to chapter 12 for a more detailed discussion about the refining margin), the Into-Plane Margin is the price that the airline pays for transporting the jet fuel from the refinery or a storage point into the aircraft (refer also to chapter 12 for a more detailed discussion about the into-plane margin), Local Currency Price of 1 US Dollar is the spot exchange rate between the country in which the airline is domiciled and the US dollar, and Fuel Consumption represents the volume of jet fuel consumed by the airline, which is defined in barrels. Let us simplify (6.5) by using some short-hand notation. First, I will construct the following definition: PJET = Crude Oil Price + Refining Margin + Into-Plane Margin

(6.6)

Equation (6.6) is the into-plane jet fuel price denominated in US dollars per barrel. It is the total cost of buying jet fuel, and having it transported from the refinery to the airport, having it stored at the airport, and pumped into the aircraft. In this case the into-plane jet fuel price is the sum of the spot price of crude oil, the refining margin, and the into-plane margin.

Airline Cost

143

Another way of writing the jet fuel price at (6.6) involves using the crack ratio rather than the refining margin as follows: PJET = Crude Oil Price u Crack Ratio + Into-Plane Margin

(6.7)

We can see in equation (6.7) that the into-plane jet fuel price is the spot price of crude oil multiplied by the crack ratio plus the into-plane margin. The sum of the crude oil price and the refining margin generates the same result as the product of the crude oil price and the crack ratio. Both generate the spot price of jet fuel. The crack ratio is simply another way of writing the refining margin. The crack ratio assumes that the refining margin depends on the level of the spot price of crude oil. When the spot price of crude oil increases, so does the refining margin, and when the spot price of crude oil falls the refining margin falls. The crack ratio is normally a number that is between 1.1 and 1.3. Let us also define the quantity of jet fuel consumed in barrels as QFuel, and the local currency price of 1 US dollar as E. The definition of E is important and can be confusing. An increase in E represents a depreciation in the local currency against the US dollar. It means that it takes more units of local currency to buy 1 US dollar. For example, if the Great British Pound price of 1 US dollar changes from 0.7 to 0.9 this means that the Great British Pound depreciates against the US dollar, because it now costs 0.9 Pound Sterling to buy 1 US dollar where before it costs only 0.7 Pound Sterling. The definition of E is to be distinguished from the definition of 1/E, which is the US dollar price of 1 unit of local currency. An increase in 1/E means that the US dollar is weaker, which represents an appreciation in the local currency. Using (6.6) and the above definitions of E and QFuel, the jet fuel cost of the airline at (6.5) can be written using the following short-hand notation: CFuel = PJET u E u QFuel

(6.8)

Now, let us assume that the fuel consumption of the airline is a linear function of ASKs as follows: QFuel = J u ASK

(6.9)

You will notice that the parameter J is one divided by the fuel productivity of the airline. If we divide both sides of (6.9) by ASKs then on the left-hand side we obtain QFuel/ASK, which is the inverse of fuel productivity described at (6.3) above. An increase in J therefore represents a reduction in fuel productivity. Substituting (6.9) into (6.8) yields the fuel cost function of the airline as a function of the into-plane price of jet fuel, the local currency price of 1 unit of US dollars, the ASKs of the airline and fuel productivity as follows: CFuel = ɀ × P୎୉୘ × E × ASK

(6.10)

If we divide (6.10) by ASKs we obtain the fuel cost per ASK of the airline or FCASK: FCASK =

େూ౫౛ౢ ୅ୗ୏

= ɀ × P୎୉୘ × E

(6.11)

The FCASK of the airline depends on the following critical parameters (1) the into-plane price of jet fuel in US dollars per barrel (PJET), (2) the local currency price of 1 unit of US dollars (E), and (3) the inverse of the fuel productivity of the airline (J). If fuel productivity (1/J) and the exchange rate (E) are reasonably stable over time, then the dominant driver of FCASK over time will be the into-plane jet fuel price. If the into-plane margin is also relatively stable over time, then the dominant driver of FCASK can be narrowed-down even further to the spot price of jet fuel. The following section will investigate the relationship between FCASK and the spot price of jet fuel for a variety of the world’s largest airlines.

6.4.2 Airline Illustrations of Time Series Movements in FCASK In this sub-section we present the time series movements in the FCASK of four different airlines to illustrate the calculation of FCASK, and to demonstrate the importance of the spot price of jet fuel in understanding how FCASK changes over time. The four airlines are the Qantas Group, Singapore Airlines, Turkish Airlines and Delta Air Lines. The movement over time in the Qantas Group FCASK versus the spot price of Gulf Coast jet fuel is given in Figure 610 below, over the timeframe 12 months to June 30, 2000, through to the 12 months to June 30, 2020.

Chapter 6

144

Gulf Coast Jet Fuel Prices (US$/bbl) $140

P-JET

Fuel Cost per ASK (AU c/ASK)

FCASK

3.5 3.14

$120

3

$100

2.5

$80

2.02

$60

2

$60 1.5

$40 $28 $20 1.02

1 Jun-20

Jun-19

Jun-18

Jun-17

Jun-16

Jun-15

Jun-14

Jun-13

Jun-12

Jun-11

Jun-10

Jun-09

Jun-08

Jun-07

Jun-06

Jun-05

Jun-04

Jun-03

Jun-02

Jun-01

Jun-00

Source: Airline Intelligence and Research Database 2021

Fig. 6-10: Qantas Group Fuel Cost Per ASK Versus the Spot Price of Jet Fuel Figure 6-10 indicates that Qantas Group FCASK rises in trend terms from 1.02 Australian cents per ASK in 2000 to 3.14 Australian cents per ASK in 2014, in response to the price of jet fuel increasing from US$28 per barrel to US$122 per barrel. After 2014 the Qantas Group FCASK falls from 3.14 Australian cents per ASK down to 2.02 Australian cents per ASK in 2017, following a decline in the jet fuel price from US$122 per barrel down to US$60 per barrel. While there is a strong correlation between Qantas Group FCASK and the US dollar price of jet fuel between 2000 and 2020, estimated to be 91.4%, there are episodes where the firm FCASK line departs from the dashed jet fuel price line. There are several potential explanations for this gap between the lines. The first is that the analytical model of section 6.4.1 tells us that the drivers of airline FCASK not only include the spot price of jet fuel, but also fuel productivity and the Australian dollar price of 1 US dollar. Variations in either of these variables over time, especially the exchange rate, will influence the gap between the firm and the dashed lines in Figure 6-10. The second driver of the gap between the firm and dashed lines in Figure 6-10 is the gains and losses from fuel hedging. The numerator of the FCASK line that is drawn in Figure 6.10 includes fuel hedge gains and losses, because airlines, including the Qantas Group, generally report their fuel costs as inclusive of these gains and losses. The main impact of the numerator of FCASK including fuel hedge gains and losses is that it causes a lag in the relationship between the spot price of jet fuel and FCASK. It creates a lag because the main aim of fuel hedging is to delay the onset of higher fuel prices. This means that an increase in the spot price of jet fuel in one year may not flow through into a higher FCASK until the following year. The impact of such a lag for the lines drawn in Figure 6-10 is that we would expect to see the peaks and the troughs in the firm line to sit to the right of the peaks and troughs in the dashed line. By incorporating fuel hedge gains and losses in the numerator of FCASK, this also means we would expect to see an asymmetry in the way that FCASK responds to the spot price of jet fuel. This is because the airline will tend to hedge in a way that allows it to buy jet fuel at low prices when the jet fuel price falls but protect it against higher jet fuel prices when the jet fuel price rises. Both the lagged effect and the asymmetric effect are two good reasons why the firm line deviates from the dashed line in Figure 6-10. The third key driver of the gap between the firm and dashed lines in Figure 6-10 is that there are some drivers of jet fuel consumption of the airline that are independent of airline capacity, which is contrary to the assumption that is used in equation (6.9), which assumes for simplicity that jet fuel consumption is wholly attributable to airline capacity. When this assumption is invalid, such factors as the passenger seat factor and the increased use of more fuel-efficient aircraft can influence the extent to which the firm and dashed lines in Figure 6-10 follow each other over time. For example, let us suppose that in any one year the Qantas Group had a much higher than expected seat factor. The higher seat factor leads to heavier aircraft in this year, which in turn leads to higher fuel burn and higher fuel costs. These higher fuel costs are then passed through into higher FCASK for a given spot price of jet fuel. This will in turn cause FCASK to go up even though the spot price of jet fuel remains fixed, or it will cause FCASK to go up at a faster pace than the spot price of jet fuel. In terms of Figure 6-10 it means that the firm line goes up by more than the dashed line goes up. Figure 6-11 below presents a time series of Singapore Airlines FCASK versus the US dollar price of jet fuel over the calendar annual period 2003 to 2019. Like the Qantas Group, there is a high correlation between the spot price of jet fuel and the FCASK of Singapore Airlines, although the strength of the correlation is weaker in the case of Singapore Airlines at 83.8%. This weaker correlation could reflect many forces. The first is that the Singapore dollar versus the US dollar may be more volatile than the Australian dollar versus the US dollar over the period examined. The second is that Singapore Airlines fuel productivity may have changed over time by more than the Qantas Group, reflecting the fact that the airline introduced into its fleet several fuel-efficient Airbus A350 and Boeing B787 Dreamliner aircraft. It

Airline Cost Gulf Coast Jet Fuel Prices (US$/bbl))

P-JET

145 Singapore Airlines Group FCASK (SG c/ASK) 5.8

FCASK

$140 $120

4.8

$100 $80

3.8

$60

2.8

$40 $20

1.8 2019

2018

2017

2016

2015

2014

2013

2012

2011

2010

2009

2008

2007

2006

2005

2004

2003

Source: Airline Intelligence and Research Database 2021

Fig. 6-11: Singapore Airlines Fuel Cost Per ASK Versus the Spot Price of Jet Fuel may also reflect the different levels of success of the Singapore Airlines fuel hedging program compared to that of the Qantas Group. In Figure 6-12 below, we present the movements over time in the FCASK of Turkish Airlines in US cents per available seat kilometre versus the spot price of jet fuel over the calendar annual period 2001 to 2019. Turkish Airlines FCASK (US c/ASK) 3.5

Gulf Coast Jet Prices (US$ / bbl) P-JET

148

FCASK

128

3

108

2.5

88 2

68 48

1.5

28

1 2019

2018

2017

2016

2015

2014

2013

2012

2011

2010

2009

2008

2007

2006

2005

2004

2003

2002

2001

Source: Airline Intelligence and Research Database 2021

Fig. 6-12: Turkish Airlines Fuel Cost Per ASK Versus the Spot Price of Jet Fuel We can see in Figure 6-12 that there is a very close correlation between the dashed jet fuel spot price line and the firm FCASK line, which is estimated to be 97.7% over the period 2001 to 2019. This largely reflects the fact that there is no exchange rate influence in Figure 6-12 because the Turkish Airlines FCASK calculation is in US dollars not Turkish Lira. The fact that there is a strong correlation between FCASK and the jet fuel price without lags would also suggest that the fuel hedging program of Turkish Airlines provides most of its protection over the space of a year, with limited protection beyond a year. Finally, we examine the relationship between FCASM and the spot price of crude oil (as opposed to the jet fuel price) in the case of Delta Air Lines between 1995 and 2020. This is presented in a line graph in Figure 6-13 below along with the spot price of crude oil over the same time period.

Chapter 6

146

Delta FCASK (US c/ASK)

FCASK

WTI Price (US$/bbl)

Crude Price

6

120

5

100

4

80

3

60

2

40

1

20

0

0 2020

2019

2018

2017

2016

2015

2014

2013

2012

2011

2010

2009

2008

2007

2006

2005

2004

2003

2002

2001

2000

1999

1998

1997

1996

1995

Source: Airline Intelligence and Research Database 2021

Fig. 6-13: Delta Air Lines Fuel Cost per ASM Versus the WTI Crude Price We can see in Figure 6-13 that there is a strong, positive correlation between the movement in the crude oil price and the movement in Delta’s FCASM. In fact, the correlation between FCASM and the spot price of crude oil for Delta Air Lines in Figure 6-13 is +97%. It is clear from the graph and the correlation that movements in the crude oil price have had a significant effect on Delta’s fuel costs between 1995 and 2020 and is likely to suggest that the refining margin or the crack ratio has been reasonably stable over this period.

Quiz 6-3: Fuel Costs 1. (a) (b) (c) (d)

What are the components of an airline’s fuel costs? Crude oil, into-plane margin, fuel consumption, US dollar exchange rate, retail margin. Crude oil, jet-crude crack, fuel consumption, Euro exchange rate. Crude oil, jet-crude crack, into-plane margin, fuel consumption, US dollar exchange rate. Crude oil, jet-crude crack, into-plane margin, US dollar exchange rate, airline capacity.

2. (a) (b) (c) (d)

What is the jet-crude crack? The difference between the crude oil price and the jet fuel price. The difference between the jet fuel price and the petrol price. The difference between the jet fuel price and the crude oil price. The difference between the jet fuel price and the into-plane margin.

3. (a) (b) (c) (d)

What is the jet crack ratio? The difference between the spot price of jet fuel and the spot price of crude oil. The difference between the spot price of crude oil and the spot price of jet fuel. The ratio of the spot price of jet fuel and the spot price of crude oil. The ratio of the spot price of crude oil and the spot price of jet fuel.

4. (a) (b) (c) (d)

What is the into-plane margin? The cost to the airline of transporting jet fuel from the on-airport tank farm to the plane. The cost to the airline of transporting jet fuel from the oil refinery to the plane. The cost to the airline of producing jet fuel. The cost of converting crude oil into jet fuel.

5. (a) (b) (c) (d)

Which of the following is NOT a driver of FCASK? The into-plane jet fuel price. The ASKs of the airline. The fuel productivity of the airline. The local currency price of 1 US dollar.

6.

An airline consumes 10,000,000 barrels of jet fuel, the crude oil price is US$100 per barrel, the jet to crude crack ratio is 20%, the into-plane margin is US$10, and the local currency price of US dollars is 2. What is the local currency price of jet fuel? (a) (100 + 10) × 1.2 × 10,000,000 × 2

Airline Cost

147

(ଵ଴଴ାଵ.ଶାଵ଴)×ଵ଴,଴଴଴,଴଴଴

(b) ଶ (c) (100 × 1.2 + 10) × 10,000,000 × 2 (d) 100 × 1.2 × 10,000,000 × 2 7. (a) (b) (c) (d)

An airline’s fuel cost in local currency is $5b, its available seat kilometres is 200b and the seat factor is 75%. What is the airline fuel cost per ASK? 5 ÷ 200 200 y 5 (5 ÷ 200) × 0.75 (200 ÷ 5) × 0.75

An airline’s fuel production is described by the function QFuel = 0.0002 u ASK, where QFuel is the quantity of fuel consumed in barrels and ASK is the airline’s available seat kilometres. The into-plane jet fuel price is described by the following function PJet = 1.17 u POil. Use this information to answer the following 4 questions. 8. What is the level of fuel productivity? (a) 0.0002 (b) 1 y 0.0002 (c) 0.0002 u ASK (d) 0.0002 u QFuel 9. (a) (b) (c) (d)

What is the crack ratio? 0.17 1/1.17 1.17 1.17 u POil

10. (a) (b) (c) (d)

What is the fuel cost of the airline as a function of the oil price and ASKs? 0.0002 u ASK u 1.17 u POil PJet u QFuel POil u QFuel 0.0002 u ASK u 1.17 u PJet

11. (a) (b) (c) (d)

What is the fuel cost of the airline per ASK? 1.17 u POil y 0.0002 0.0002 u 1.17 u PJet 0.0002 u POil y 1.17 0.0002 u 1.17 u POil

12. An airline’s fuel productivity is described by the function QFuel = 0.00015 u ASK. The into-plane jet fuel price is described by the following function PJet = (20 + 1.1 u POil). What is the fuel cost per ASK of the airline as a function of the spot price of oil? 13. Japan Airline’s (JAL’s) fuel productivity is described by the parameter E. The into-plane jet fuel price paid by JAL is US$50 per barrel and the Japanese Yen price of 1 US dollar is 100. Construct an expression for the fuel cost of JAL in Japanese Yen and the fuel cost per ASK in Japanese Yen.

6.5 Airline Labour Costs 6.5.1 Labour Resources at Airlines There are many different types of labour resources that work at airlines. Broadly they can be grouped into two categories – operational labour and non-operational labour. To get a feel for the different types of labour that work for an airline it is useful to start with the key airline non-operational departments because these departments must be filled with people that work for them. These departments include but aren’t limited to those presented in Table 6-5 below. For each of the departments presented in Table 6-5 there are analysts, assistants, managers, senior managers, general managers, and executive general managers. They are tasked with making important decisions for the strategic direction of the airline. The operational staff consist mainly of technical and cabin crew as well as engineers, ground handlers, in-terminal airport staff and caterers. The volume of operational staff will depend on the volume of capacity supplied by the airline. The volume of non-operational staff will not generally depend on the volume of capacity in the short run but will be loosely linked to capacity over a long horizon - if the airline were to supply no capacity, then there would be no need for non-operational staff.

Chapter 6

148

Office of the CEO Freight

Office of the CFO Flight Operations

Legal Safety and Security

Strategy Aircraft Procurement

Engineering Procurement

Corporate Communications Audit Treasury and Risk Management Property Management

Fleet Planning

Human Resources Revenue and Yield Management Environment and Fuel Conservation Shared Services Accounting

Staff Training Superannuation

Government Relations Fleet and Network Planning Industrial Relations Revenue Forecasting Infrastructure Services Taxation Finance

Board Directors

Table 6-5: List of Non-Operational Airline Departments Figure 6-14 below presents a time series of Qantas Group Full-Time Equivalent (FTE) staff between June 2001 and June 2019.

Average Qantas Group FTE Staff 36,000 35,000 34,000 33,000 32,000 31,000 30,000 29,000 28,000

35,339

29,321

28,988

2019

2018

2017

2016

2015

2014

2013

2012

2011

2010

2009

2008

2007

2006

2005

2004

2003

2002

2001

Source: Airline Intelligence and Research Database 2021

Fig. 6-14: Qantas Group Full Time Equivalent Staff As indicated in Figure 6-14, Qantas Group FTE’s dropped dramatically between 2005 and 2015 from a peak of 35,339 as at June 30, 2005, down to 28,988 as at June 30, 2015, a reduction of 6,351 FTEs. As indicated in Figure 6-15 below this decline in FTE’s has coincided with a lift in Qantas Group available seat kilometres, suggesting that much of the reduction in Qantas Group FTEs is attributable to a reduction in non-operational staff. Qantas Group ASKs (m) 160,000

151,430

150,000 140,000 130,000 120,000 110,000 100,000 93,123 90,000

2019

2018

2017

2016

2015

2014

2013

2012

2011

2010

Fig. 6-15: Qantas Group Available Seat Kilometres

2009

2008

2007

2006

2005

2004

2003

2002

2001

Source: Airline Intelligence and Research Database 2021

Airline Cost

149

What is likely to have driven the reduction in non-operational labour? When non-operational labour falls it is generally because the airline is under earnings pressure and believes it needs to take costs out of the business to improve that earnings performance. Often the senior executive will inform key departments that it must cut a certain percentage of FTEs from their business units. This will start with voluntary redundancies and then move onto involuntary redundancies depending on how many employees put their hands up to leave at the first stage. There are many variables that have driven earnings pressure over the past decade and a half, but the most important is the price of oil and thus jet fuel. This is followed by the extent of excess capacity in markets and the subsequent downward pressure on yields. Earnings have also been dented by the large number of constant adverse shocks that have impacted the business including the SARS and Corona viruses, the terrorist events of September 11, the Bali Bombings, and the Volcanic ash disturbances in various parts of the world.

6.5.2 Airline Wage Rates In analytical terms airline manpower costs can be seen as the product of the average wage rate and the number of fulltime equivalent employees: Manpower Costs = Average Wage Rate u FTEs

(6.12)

The average wage rate of the airline, by definition, is equal to the total manpower costs divided by the number of FTE’s. This is not a true average wage rate since it is likely to include non-wage labour costs as well, but most of the content of the Average Wage Rate term on the right-hand side of (6.12) is a mix of the average wage rates of different workers within the airline. Different workers will have different types of arrangements for determining wage rates. Most operational employees will have their trade union determine the wages that they receive. Some of the biggest airline union movements that negotiate wages on behalf of members include the transport workers union (representing workers such as baggage handlers and other ground staff at the airport), the licensed aircraft maintenance engineers (LAME) union, the pilots union, the cabin crew union, and the union representing check-in, finance and administration, and managerial staff. These trade unions usually negotiate wages and conditions on behalf of their members through legally binding agreements. Other employees, usually non-operational employees, will negotiate their wage contracts on their own. The key drivers of movements in wages will be the ability of unions to negotiate higher wages and the capacity of the airline to pay for higher wages. The ability of unions to negotiate higher wages will depend on many forces, including the real purchasing power of their members, which will be driven by inflation relative to nominal wage movements, the negotiation strength and influence of the union itself, which will be a function of the number of members and the importance of those members to the airline’s operational functions and earnings ability, and the productivity of members. The capacity of the airline to pay will clearly depend on the earnings of the airline, both historical and expected. The executives of airlines have many different component parts to their pay. These component parts usually include fixed annual remuneration or FAR, which is simply a given amount of cash that executives receive on a regular basis (weekly, fortnightly, or monthly) and includes compulsory superannuation or pension payments. It also includes a cash bonus, which is a certain percentage of the FAR amount, and represents the cash that the executive receives if the airline meets various financial and other targets. Most airline executives will also be offered a short-term incentive scheme, which represent payment to the executive in the form of shares of the company if the company meets targets. The executive will also be offered a long-term incentive scheme, which represents payment to the executive in the form of shares if the company meets targets and the employee has stayed with the airline for a sufficient period of time. Executive remuneration will also include salary sacrificing, and heavily discounted staff travel, including travel in premium cabins. It would be expected that growth in the FAR component of executive pay, in the main part, will be driven by inflationary pressures, although it may also be influenced by the earnings position of the company. The company will have an average growth in FAR in mind, for example 3%, and will award the better employees with growth in FAR above this and the underperformers with growth below this. If company earnings are weak then the airline may decide to freeze executive wages. The remaining components of executive pay will be heavily influenced by the earnings of the company. Bonus pay will usually be paid after earnings performances are realised and so there is a built-in lag between executive bonuses and earnings.

6.5.3 Comparing Manpower Costs Across Airlines and Redundancies 6.5.3.1 Comparing Manpower Costs It is often difficult to compare manpower costs across airlines. When airlines outsource functions that have heavy labour components, such as ground services, engineering and catering, these costs are not included in labour costs but in outsourcing, material, or other cost categories. The airline itself does not employ the labour but the outsource provider does, which means the FTEs do not sit on the books of the airline but the outsource provider. If airline A takes a view that it is cheaper to outsource a function, or even several functions, such as those presented above but airline B believes that the functions should be in-house then the manpower costs as a proportion of total cost, or manpower costs per ASK

Chapter 6

150

can be much higher at B than at A, even after taking into consideration the fact that the airlines are domiciled in different countries with different wage levels. 6.5.3.2 Redundancy Costs Another increasingly important cost that is associated with the manpower function is redundancy costs. These costs may sit in or outside of manpower costs, and often receive their own categorisation. While manpower costs may increase in the short term because of redundancy payments they will fall over a longer horizon as the airline no longer pays the relevant wages. The employees that the airline targets for redundancy are also an important consideration for unit costs. Often an airline will ask employees to take voluntary redundancy during periods of financial distress. In most instances the employees that put their hand up for voluntary redundancy are those on the verge of retirement and who have been with the company for a long period of time. These employees stand to gain significantly from a redundancy because the redundancy payout is often computed based on the length of time with the company and final salary levels. The employees that typically do not put their hand up for a voluntary redundancy are less secure about their job, have little readily transferable skills and enjoy staff travel benefits. This labour is often less productive labour while those who do put their hand up for voluntary redundancy often have an enormous amount of airline experience or capital that leaves the company when they exit. When the airline experiences a significant brain drain this can lead to poorer management decisions, which either leads to lost revenue opportunities or higher unit costs.

6.5.4 A Simple Model of Airline Labour Costs We assume that there are two sets of airline labour – non-operational (LN) and operational (LO) as described above. The number of operational employees depends on the ASKs of the airline while the number of non-operational employees is independent of the number of ASKs. We can therefore write the number of operational labour units as a function of ASKs as follows: LO = aL u ASK

(6.13)

The total labour of the airline is thus the non-operational labour LNO plus (6.13): L = LNO + aL u ASK

(6.14)

The labour costs of the airline are equal to the average wage rate that the airline pays, w, multiplied by the number of employees (6.14), yielding: CL = w u (LNO + aL u ASK)

(6.15)

Equation (6.15) says that labour costs are a linear function of ASKs. Staff cost per ASK or SCASK is equal to (6.15) divided by the number of ASKs as follows: c୐ =

େై ୅ୗ୏

=

୵୐ొో ୅ୗ୏

+ wa୐

(6.16)

We can see in (6.16) that the key drivers of SCASK over time are the wage rate of employees (w), the productivity of operational employees (aL), the number of non-operational employees (LNO), and the ASKs of the airline. There is pressure for SCASK to increase over time because wages increase over time with inflation, but there is pressure for SCASK to fall over time as the airline expands capacity. The ultimate direction of the SCASK of an airline will depend on the balance of these three forces. Let us illustrate this point by examining the staff cost per ASK of the International Airlines Group (IAG), which consists of the airlines British Airways, Iberia Airways, Aer Lingus, LEVEL and Vueling Airlines – refer to Figure 616 below. We can see in Figure 6-16 that the staff cost per ASK of IAG has been falling over time. This would tend to suggest that the airline’s labour productivity has been increasing, with the airline adding ASKs at a faster pace than it is adding staff. This is consistent with the rising labour productivity of the airline Group as indicated in Figure 6-17 below, with labour productivity rising from 3.45 million ASKs per FTE in 2009 up to 5.14 million ASKs per FTE in 2019.

Airline Cost

151

IAG Staff Cost per ASK (EUR c/ASK) 2.1 1.98

2 1.9 1.8 1.7 1.6 1.5

1.47

1.4 2009

2010

2011

2012

2013

2014

2015

2016

2017

2018

2019

Source: Airlines Intelligence and Research Database 2021

Fig. 6-16: International Airlines Group Staff Cost per Available Seat Kilometre IAG ASK per FTE (m)

5.4 5.2

5.14

5 4.8 4.6 4.4 4.2 4 3.8 3.6 3.4

3.45 2009

2010

2011

2012

2013

2014

2015

2016

2017

2018

2019

Source: Airlines Intelligence and Research Database 2021

Fig. 6-17: International Airlines Group FTE per Available Seat Kilometre

Quiz 6-4: Manpower Costs 1. (a) (b) (c) (d)

Which of the following statements is most correct about airline labour requirements? Operational labour requirements are mostly driven by the earnings performance of the company. Non-operational labour requirements are mostly driven by airline capacity. Operational labour requirements are mostly driven by airline capacity. Non-operational labour requirements are mostly driven by the passengers and freight load.

2. (a) (b) (c) (d)

The FAR in an executive contract is: The fixed “cash” component of the contract. The shares the executive receives if the airline meets short term targets. The shares the executive receives if the airline meets long term targets. The total package salary of the executive.

3. (a) (b) (c) (d)

The FAR in an executive contract stands for: Fixed Annual Remuneration. Fixed Annual Reward. Flexible Annual Remuneration. Flexible Annual Reward.

4.

The wage rate outcomes for airline employees are determined by:

Chapter 6

152

(a) (b) (c) (d)

Inflation. Productivity. The earnings performance of the company and the ability of the company to pay. All the above.

5. (a) (b) (c) (d)

Which of the following explains why the staff cost per ASK of an airline may fall over time? The wage of airline staff increases with inflation. The ASKs of the airline grow at a faster pace than non-operational staff. The wage of airline staff increases with labour productivity. The airline’s labour productivity falls over time.

6.

The Qantas Group employs 5,000 non-operational staff. It pays $110,000 per annum in wages on average to non-operational staff and $75,000 per annum to operational staff. The labour productivity of the airline is 5m available seat kilometres per operational labour unit. Use this information to determine the staff cost per available seat kilometre of the Qantas Group. Why does staff cost per ASK fall as ASKs increase?

6.6 Aircraft Capital Costs 6.6.1 Owned Aircraft Airlines operate aircraft that are either owned by the airline or leased from aircraft lessors. When the airline owns the aircraft the price that the airline pays for the aircraft is allocated to the costs in the airline’s profit and loss statement over the life of the aircraft via the depreciation line in that statement. The simplest way to think of the amount that sits in the new aircraft depreciation cost line in the profit and loss statement is to divide the amount paid for the aircraft by the life of the aircraft: Aircraft Depreciation Cost =

୅୫୭୳୬୲ ୔ୟ୧ୢ ୤୭୰ ୲୦ୣ ୅୧୰ୡ୰ୟ୤୲ ୐୧୤ୣ ୭୤ ୲୦ୣ ୅୧୰ୡ୰ୟ୤୲

(6.17)

The aircraft depreciation method described by (6.17) is referred to as straight-line depreciation, which is a popular method used by airlines to determine the depreciation cost of their aircraft investments. For example, the price of a new Boeing 787-800 Dreamliner aircraft is US$248.3 million. The life of the aircraft is 20 years. The aircraft depreciation cost that sits in the profit and loss statement of the airline for the life of the aircraft asset using (6.17) is: Aircraft Depreciation CostB787-800 =

ଶସ଼,ଷ଴଴,଴଴଴ ଶ଴

= US$12,415,000

Another way to estimate the depreciation cost is to simply multiply the value of the aircraft by a rate of depreciation. This method can be applied to both new and old aircraft. The aircraft depreciation cost in this case is: Aircraft Depreciation Cost = Value of Aircraft × Depreciation Rate

(6.18)

For example, if the rate of depreciation is 3.5% for our Boeing 787-800 aircraft, then the depreciation cost that sits in the profit and loss statement of the airline is: Aircraft Depreciation CostB787-800 = 3.5% × 248,300,000 = US$8,690,500 After the depreciation cost is accounted for, the value of the aircraft asset is now worth the original value less the depreciation expense: New Value of Aircraft AssetB787-800 = 248,300,000 – 8,690,500 = US$239,609,500 This new value of the aircraft asset is also referred to as the book value of the asset, or the written down value. This method of computing the depreciation expense of owned aircraft is called the diminishing value method. It is given this name because the aircraft’s depreciation expense starts out relatively high early in the life of the asset and then falls or diminishes. This is demonstrated in Table 6-6 which sets out the depreciation of our Boeing 787-800, in the third column entitled “Depreciation Expense”. In Table 6-6 the depreciation expense is highest in the 1st year at US$8.69m, falling by the 20th year of the aircraft asset to US$4.42m. Similarly, the book value of the asset falls from US$248.3m in period 0 to US$121.77 by year 20. As you would expect there are some complications associated with both the numerator and the denominator of (6.17) and the two components of (6.18) – nothing is simple in aviation! Let us now discuss some of these complications. In the case of the numerator of (6.17) and the value of aircraft in (6.18), the major complication is associated with the fact that airlines pay aircraft manufacturers such as Airbus, Boeing, and Embraer in US dollars. This means that the amount paid for the aircraft in the currency of the airline will depend on the exchange rate between the currency of the airline or the local currency and the US dollar. For example, a Japanese carrier such as All Nippon Airways or Japan

Airline Cost

Year 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20

Written-down Value of Aircraft Asset (US$ m) $248.3 $239.61 = $248.3 - $8.69 $231.22 = $239.61 - $8.39 $223.13 = $231.22 – $8.09 $215.32 = $223.13 - $7.81 $207.78 = $215.32 - $7.54 $200.51 = $207.78 - $7.27 $193.49 = $200.51 - $7.02 $186.72 = $193.49 - $6.77 $180.19 = $186.72 - $6.54 $173.88 = 180.19 - $6.31 $167.79 = $173.88 - $6.09 $161.92 = $167.79 - $5.87 $156.25 = $161.92 - $5.67 $150.79 = $156.25 - $5.47 $145.51 = $150.79 - $5.28 $140.42 = $145.51 - $5.09 $135.50 = $140.42 - $4.91 $130.76 = $135.50 - $4.74 $126.18 = $130.76 - $4.58 $121.77 = $126.18 - $4.42

153

Depreciation Expense (US$ m) $8.69 = 3.5% u 248.3 $8.39 = 3.5% u 239.61 $8.09 = 3.5% u $231.22 $7.81 = 3.5% u $223.13 $7.54 = 3.5% u $215.32 $7.27 = 3.5% u $207.78 $7.02 = 3.5% u $200.51 $6.77 = 3.5% u $193.49 $6.54 = 3.5% u $186.72 $6.31 = 3.5% u $180.19 $6.09 = 3.5% u $173.88 $5.87 = 3.5% u $167.79 $5.67 = 3.5% u $161.92 $5.47 = 3.5% u $156.25 $5.28 = 3.5% u $150.79 $5.09 = 3.5% u $145.51 $4.91 = 3.5% u $140.42 $4.74 = 3.5% u $135.50 $4.58 = 3.5% u $130.76 $4.42 = 3.5% u $126.18

Table 6-6: Diminishing Value Method Depreciation Schedule for a Boeing B787-800 Airlines will pay for new Boeing aircraft, such as the B787, in US dollars. This means that if the Japanese Yen price of 1 US dollar is higher, which means that the Yen is weaker against the US dollar, it will cost the airline more in Japanese Yen to buy aircraft from Boeing or Airbus. This in turn means that the depreciation expense associated with the aircraft will also increase in response to the depreciation of the Yen against the US dollar. When airlines buy aircraft, it is not as simple as registering an interest in an aircraft with the aircraft manufacturer, handing over millions of dollars in one transaction, receiving the aircraft from the manufacturer and then flying the aircraft in operations. The payments that the airline makes to the aircraft manufacturer takes place in stages. The first stage involves the payment of an initial deposit. The second stage involves the payment of pre-delivery payments. And the last stage involves complete settlement of the aircraft at the time the airline takes delivery of the aircraft. At each point in these various payment stages, the exchange rate can be different, which means the airline faces the risk that its currency is weaker at each of these payment stages. It also means that when the airline takes receipt of the aircraft, the value of that aircraft in the currency of the airline will depend on several different historical exchange rates. To see this, suppose that All Nippon Airways buys a Boeing 787-800 aircraft for US$243.8 million. It pays a deposit on January 1, 2018, of 5% of this value. It then pays 6 pre-delivery payments of 1% of the purchase price every 6 months, and it pays the amount outstanding after these payments when the aircraft is delivered on December 31, 2021. This timeline of payments is described in Figure 6-18, with the US dollar payments appearing underneath the timeline and the Japanese Yen price of 1 US dollar appearing above the timeline in Figure 6-18. 1 Jan-18 112 JPY = 1 US$

30 Jun-18 111 JPY = 1 US$

31 Dec-18 110 JPY = 1 US$

30 Jun-19 31 Dec-19 30 Jun-20 31 Dec-20 109 JPY 109 JPY 108 JPY = 104 JPY = 1 US$ 1 US$ = 1 US$ = 1 US$

US$12.19

US$2.438

US$2.438

US$2.438

US$2.438 US$2.438

US$2.438

31 Dec-21 100 JPY = 1 US$

US$217

Fig. 6-18: Ann Nippon Airways Boeing 787-800 Payment Schedule and Exchange Rate Movements These US dollar payments are converted into Japanese Yen using the spot Japanese Yen to US dollar exchange rate at the time the payment is made. We can see in Figure 6-18 that the exchange rate that is used to translate the required payment in US dollars into Japanese Yen is different for 7 of the 8 payments. The exchange rate when the deposit is paid is 1 US$ equals 112 Japanese Yen (the first exchange rate figure presented above the timeline in Figure 6-18). This drops to 1 US$ equals 111 Japanese Yen by the time the first pre-delivery payment is made (the second exchange rate figure in Figure 6-18). By the time the airline makes its final pre-delivery payment the Japanese Yen price of 1 US$ drops to 104 and it drops even further to 100 Yen by the time of the last settlement payment. It follows that the total

Chapter 6

154

Japanese Yen value of the aircraft when All Nippon takes receipt of the aircraft will depend on 7 different historical exchange rates. The total Japanese Yen value of the aircraft when the airline takes receipt of the Boeing 787-800 is: Value of B787-800 = 12.19u112 + 2.438u(111 + 110 + 109+ 109 + 108 + 104) + 100u217 = 24,652m JPY Over a 20-year life of the asset, the straight-line depreciation expense in each year of the aircraft using (6.17) is: Aircraft Depreciation Cost =

ଶସ,଺ହଶ ଶ଴

= 1,232.6m JPY

Let us now suppose that the Japanese Yen versus US dollar exchange rate remained at the 1 January 2018 value of 112 for all payments in Figure 6-18. In this case the payment that All Nippon Airways would make to Boeing converted into millions of Japanese Yen becomes: Value of Dreamliner B787-800 = 243.8 u 112 = 27,306m JPY The depreciation expense in each year of the life of the asset in millions of Yen in this case is: Aircraft Depreciation Cost =

ଶ଻,ଷ଴଺ ଶ଴

= 1,365.3m JPY

The depreciation expense is therefore higher in the world where the exchange rate is fixed at 112 because over the period January 1, 2018, through to December 31, 2021, the Japanese Yen price of 1 US dollar falls, or the Japanese Yen appreciates. The other key complication associated with the value of the aircraft in the depreciation expense formula is that the true or market value of the aircraft may not reflect the book value of the aircraft. The true value of the asset is the value that the aircraft would fetch if it were sold in the second-hand aircraft market. The price of aircraft in the second-hand aircraft market is determined by the demand for and the supply of aircraft in that market. If the demand for air travel is in a weakened state, or there is a significant increase in the number of airlines and leasing companies who wish to sell aircraft into the second-hand market, this will result in a relatively low price in the second-hand market for aircraft. This second-hand price may be below the book value of the same vintage aircraft that sits on the balance sheet of airlines. If this is the case, airlines may need to “write-off” the difference between the value of the asset on the airline’s balance sheet and the value of the aircraft in the second-hand aircraft market. This write-off amount will sit within the nonoperating cost items in the profit and loss statement of the airline – refer also to section 6.2 above where we discussed the cost of asset impairment. Another complication associated with determining the depreciation expense of aircraft is in determining the life of the aircraft asset, and thus determining the depreciation rate for the aircraft. According to IATA 2020, which analyses the depreciation policies of a number of major airlines, aircraft assets are depreciated over 15 to 25 years, with 20 years as the midpoint. Airlines can extend the life of the aircraft asset by overhauling different parts of the aircraft. For example, the airline may overhaul the airframe, it may refurbish the cabin, and it may overhaul the aircraft’s engines. The airline will need to make an assessment about how such modifications add to the useful life of the aircraft asset.

6.6.2 Leased Aircraft An airline may also rent or lease aircraft from an aircraft leasing company, such as AerCap or General Electric Capital Aviation Services, or from another airline. The airline in this case is called the lessee and the leasing company is called the lessor. The airline pays to the lessor a lease premium, which gives the airline the right to operate the aircraft. This lease premium will depend on many different factors, such as the type of lease arrangement that is in place (finance or operating lease), the price that the lessor paid for the aircraft, the cost of the finance that the lessor used to purchase the aircraft, any costs associated with the lessor bearing exchange rate risk, and the tenure of the lease arrangement. The lease premium will also depend on the state of the demand for aircraft leases, with strong demand for aircraft leases enabling lessors to charge a relatively high lease premium, and weak demand for aircraft leases placing downward pressure on premiums. The lease premium was once a component of the operating cost line of the profit and loss statement of the airline. Changes to accounting rules, however, mean that airlines no longer include a lease premium in their profit and loss statement. Instead, the lease premium forms part of the depreciation expense line of the profit and loss statement. This change stems from new accounting rules which now treat leased aircraft as “right-of-use” assets. This means that the accounting rules treat the leased aircraft as an asset of the airline rather than an asset of the lessor. This change in treatment follows from the fact that the airline lessee has exclusive use of the asset. As the aircraft is treated as an asset of the airline then the cost of using that asset is depreciated and enters the profit and loss statement of the airline as a depreciation expense rather than a lease premium expense. The leased asset also enters the non-current assets part of the airline balance sheet as a right-of-use asset where before it wasn’t a part of the balance sheet. This has significantly affected the asset base of many airlines that rely heavily on leased aircraft.

Airline Cost

155

To illustrate the altered treatment of leased assets as right-of-use assets, consider the balance sheet of Singapore Airlines in the September quarter in 2018, 2019 and 2020 – refer to Table 6-7 below. STATEMENTS OF FINANCIAL POSITION AT 30 SEPTEMBER 2020 (in $ million) 30-Sep 2020 Equity Attributable to Owners of the Company Share Capital Mandatory convertible bonds Treasury shares Other reserves Non-controlling Interests Total Equity Deferred Account Deferred Taxation Long-term Lease Liabilities Borrowings Other Long-Term Liabilities Provisions Defined Benefit Plans Represented by: Property, Plant and Equipment Right-of-Use Assets Intangible Assets Associated Companies Joint Venture Companies Long-Term Investments Other Long-Term Assets CURRENT ASSETS Derivative assets Inventories Trade Debtors Amounts owing by subsidiary companies Deposits and other debtors Prepayments Other short-term assets Investments Cash and bank balances Assets held for sale Less: CURRENT LIABILITIES Borrowings Lease liabilities Current tax payable Trade and other creditors Amounts owing to subsidiary companies Sales in advance of carriage Deferred revenue Deferred account Derivative liabilities Provisions NET CURRENT ASSETS/(LIABILITIES)

The Group 30-Sep 2019

30-Sep 2018

7,180.2 3,496.1 (133.2) 4,708.4 15,251.5 393.0 15,644.5 32.1 1,139.2 1,229.6 8,486.6 1,716.8 953.7 109.9 29,312.4

1,856.1

1,856.1

(156.0) 10,098.7 11,798.8 399.1 12,197.9 42.9 1,843.8 1,618.7 5,808.2 338.0 942.1 102.5 22,894.1

(171.5) 12,367.4 14,052.0 369.8 14,421.8 99.3 2,211.1 4,364.9

24,005.0 1,127.8 319.9 829.4 202.3 50.1 692.3

24,224.7 1,627.9 463.7 824.2 185.9 337.0 380.3

20,366.6 421.9 954.2 148.7 341.3 1,588.3

17.5 206.5 790.6

191.8 234.1 1,390.8

9.5 205.1 1,518.2

190.4 104.4 30.3 528.3 7,058.5 14.0 8,940.5

77.4 172.2 49.4 127.0 1,307.9

118.5 185.7 35.3 72.2 1,982.5

3,550.6

4,935.3

1,005.1 427.8 69.6 2,306.6

852.5 490.9 77.9 3,100.7

148.7 92.1 2,889.6

792.0 879.4 25.6 1,011.1 337.7 2,085.6

2,962.3 667.4 27.0 123.6 397.9 (5,149.6)

2,637.1 580.0 47.9 26.5 407.1 (1,893.7)

774.5 104.0 21,975.6

Source: Airline Intelligence and Research Database 2021

Table 6-7: Singapore Airlines Balance Sheet September 2018, 2019, and 2020 We can see the Right-of-Use assets line under the Property, Plant and Equipment line in the non-current assets part of the balance sheet of the airline. We can see that Right-of-Use assets were recorded as zero as at September 30, 2018,

Chapter 6

156

but recorded as SG$1.628b as at September 30, 2019, and SG$1.128b as at September 30, 2020. At the same time the depreciation expense of the airline increased from SG$324.5m over the September quarter 2018 up to SG$508.7m over the September quarter 2019, and SG$1,090.9m over the September quarter 2020. The increase in the depreciation line of Singapore Airlines reflects the changing accounting treatment of leased aircraft. To determine the depreciation expense associated with right-of-use assets the airline will determine the present discounted value of the lease payments that it is contracted to pay to the lessor under the lease. The present discounted value of lease payments is equal to the sum of the discounted value of each lease payment. By discounting the lease payments, we recognise that lease payments today are worth more than the same lease payments made in the future. For example, suppose that an airline is required to pay a lease payment of R to the lessor every year over a T year lease of the aircraft. The present discounted value (PV) of these lease payments is determined as follows: PV of Lease Payments = R +

ୖ ଵା୰

ୖ

ୖ

ୖ

+ (ଵା୰)మ + (ଵା୰)య + …… + (ଵା୰)೅షభ

(6.19)

where r is referred to as the discount rate, which is often approximated by the weighted average cost of capital of the airline.35 The terms (1 + r)i in the denominator of our PV of lease payment expressions (6.19) represents the discount factor. These denominator terms convert the lease payments from a future value to a present value. We divide each of our lease payments by (1 + r)i because the future value of a particular payment is worth less than the present value of the same payment. We can think of the PV of the lease payments on the left-hand side of (6.19) as the capital value of the leased asset. Once we have this capital value, we can simply multiply it by a depreciation rate to find the notional depreciation expense which is used in place of a lease premium in the profit and loss statement. Now that we have simple models of the owned and leased aircraft costs of the airline, we can present a simple model of the aircraft capital costs of the airline. Let us suppose that the airline has a fleet of FO aircraft that it owns and a fleet of FL aircraft that it leases. The owned aircraft are valued at ܽ௜ை on average for each aircraft and the leased aircraft are valued at ܽ௜௅ on average for each aircraft based on the PV of lease payments at (6.19). The aircraft depreciation rate that is applied to both leased and owned aircraft is G. Using this information, the aircraft capital cost of the airline, which includes the depreciation of its owned aircraft, and the notional depreciation of right-of-use aircraft is: ో

ై

୊ a୐୧ ቁ Aircraft Capital Cost = ߜ ቀσ୊୧ୀଵ a୓୧ + σ୧ୀଵ

(6.20)

Equation (6.20) applies a fixed depreciation rate, G, to the value of owned and leased aircraft, which is the component in brackets on the right-hand side of (6.20). This generates an estimate of the costs associated with operating both owned and leased aircraft across its network.

Quiz 6-5: Aircraft Capital Costs 1.

An airline buys a Boeing 737 Max 9 aircraft for US$130m. The airline expects the aircraft to have a life of 20 years. What is the depreciation expense associated with this aircraft in the case of the straight-line depreciation method in millions of US dollars? (a) 130 ଵ (b)

(c) (d)

35

ଶ଴ ଶ଴

ଵଷ଴ ଵଷ଴ ଶ଴

2.

For the Boeing 737 Max 9 aircraft in question 1., assume that the method that is used to determine the depreciation expense of the aircraft is the diminishing value method with a 10% rate of depreciation. Set up a table like Table 6-6 above, including the year of operation of the aircraft, the book value, and the depreciation expense. Graph the depreciation expense over the 20-year life of the aircraft using a line graph and explain its shape.

3. (a) (b) (c) (d)

What is a pre-delivery payment? The price of an aircraft determined at the time of signing the aircraft delivery contract. Final payment for an aircraft on delivery. A series of deposit payments made to aircraft manufacturers prior to the delivery of the aircraft. The payment made at the time an aircraft deal is struck and before delivery.

The weighted average cost of capital is the weighted average of the cost of debt finance and the cost of equity finance. The weight that is applied to the cost of debt is the ratio of debt to debt plus equity, and the weight that is applied to the cost of equity is the ratio of equity to debt plus equity. The weighted average cost of capital represents the cost incurred by the airline in using a combination of borrowed funds (debt) and funds raised from shareholders (equity) to purchase income-generating assets, such as aircraft.

Airline Cost

4. (a) (b) (c) (d)

157

Why will a weaker local currency against the US dollar lead to higher aircraft depreciation costs? Because we observe empirically that higher aircraft prices coincide with a weaker US dollar. Because airlines usually pay aircraft manufacturers in local currency. Because airlines usually pay aircraft manufacturers in US dollars. Because a weaker US dollar reduces the input prices of aircraft manufacturers.

5. Explain why an airline’s aircraft depreciation expense depends on long lags of the value of the local currency versus the US dollar. 6. What do we mean by the fair market value of an aircraft asset? Why might this differ from the book value of an aircraft asset? 7. When will an airline need to write-off the value of an aircraft asset? (a) When the second-hand price of aircraft is falling. (b) When the price at which an aircraft could be sold in the second-hand market is higher than the value of the aircraft on the balance sheet of the airline. (c) When the US dollar is stronger. (d) When the value of the aircraft asset on the balance sheet of the airline is worth less than the value at which the aircraft could be sold in the second-hand aircraft market. 8. What is generally accepted as the average useful life of an aircraft asset? (a) 10 years (b) 15 years (c) 20 years (d) 25 years 9. What is meant by a right-of-use asset? (a) When two or more airlines share the right to use an asset. (b) When an airline leases an aircraft asset over a relatively long horizon and has exclusive use of it over the term of the lease. (c) When the government grants regulatory approval for the airline to operate an aircraft. (d) When one airline subleases its owned aircraft to another airline.

6.7 Airport Charges 6.7.1 Airport Charges and the Building Block Approach Airlines are required to compensate airports for the costs associated with the provision of aeronautical services, such as use of the runway, taxiways, apron area, gate, and terminal. The price that airlines pay to airports for the use of these services is often referred to as an airport charge. As many airports around the world operate as geographic natural monopolies, meaning that the next closest competing airport is a sizeable distance to the airport in question, competition authorities around the world often take a close look at the way that airports set airport charges. These competition authorities often require or recommend airports to set airport charges using an average cost pricing basis. This approach says that airports set airport charges equal to the average cost of airport production. The average cost of airport production is equal to the total cost of providing aeronautical services by the airport plus a suitable return on assets all divided by an activity driver. For most relatively large commercial airports, the primary activity driver is the number of passenger movements. The relevant average cost that is used to determine airport charges in this case is the total airport cost divided by the number of passenger movements. This approach to setting airport charges as total airport costs divided by passenger numbers is often referred to as a building-block approach because it builds-up the costs of each of the services that the airport provides and then divides by passenger numbers. The costs that are included in the numerator of the building block expression include capital costs and operational costs. The capital costs represent the return on capital of the airport plus the depreciation of relevant airport assets. These costs are generally independent of the activity driver, in this case the number of passengers. The operational costs include the costs associated with providing security, maintenance and repair, and utility costs. These costs generally depend on the number of passengers. In the case of relatively small airports that are largely serviced by regional aircraft, such as the ATR, Embraer, Saab, Fokker or De Havilland aircraft, the activity driver is usually maximum take-off weight or MTOW. In the case of using MTOW as the activity driver in the building-block approach, the airport charge is total airport production costs divided by MTOW. Not only is MTOW used at small regional airports, but it is also often the basis for setting airport charges that are paid by small regional airlines when they use the aeronautical services of large airports. For example, as at November 2019 Sydney Airport charged $3.78 per tonne for take-offs and landings in the case of services provided by airlines flying within NSW (which is the state of Australia in which Sydney Airport is located). These intra-state services are largely supplied by small aircraft and regional airlines (such as QantasLink, Virgin Regional and Rex Express at the

Chapter 6

158

time of writing). In the case of domestic services at Sydney terminal 2, which largely involves interstate jet (B737 and A320) aircraft services, Sydney Airport charged $8.15 on a per passenger arrival and departure basis for the supply of terminal services and $3.91 per passenger in the case of runway services. This is an example of a large airport setting the basis for airport charges differently for large jet aircraft compared to small propellor aircraft.

6.7.2 Single-Till Versus Dual-Till Approaches to Airport Charges The building-block approach is applied either as a single-till or a dual-till approach. Under a dual-till approach, the airport must separate its costs into those attributable to the provision of aeronautical services and those attributable to the provision of non-aeronautical goods and services. Non-aeronautical goods and services include the products sold by retail merchants within the airport terminal and outside the terminal but within the airport precinct, the use of the airport car park by passengers, access to the airport by taxis, ride-share vehicles, busses and trains, and hire car operations at the airport. The aeronautical costs are then divided by an activity driver such as passenger numbers to obtain dual-till airport charges. We can describe this by using the following formula: Dual-Till Airport Charge =

୅୧୰୮୭୰୲ ୅ୣ୰୭୬ୟ୳୲୧ୡୟ୪ େ୭ୱ୲ୱ ୔ୟୱୱୣ୬୥ୣ୰ ୒୳୫ୠୣ୰ୱ

(6.21)

Under a single-till approach, the building block model is applied by taking the costs associated with aeronautical activities and subtracting from this the profit that is made from non-aeronautical activities and dividing the result through by an activity driver, such as passenger numbers. This is shown in the formula below: Single-Till Airport Charge =

୅୧୰୮୭୰୲ ୅ୣ୰୭୬ୟ୳୲୧ୡୟ୪ େ୭ୱ୲ୱି୔୰୭୤୧୲ ୤୰୭୫ ୒୭୬ି୅ୣ୰୭୬ୟ୳୲୧ୡୟ୪ ୅ୡ୲୧୴୧୲୧ୣୱ ୔ୟୱୱୣ୬୥ୣ୰ ୒୳୫ୠୣ୰ୱ

(6.22)

We can see by comparing (6.21) and (6.22) that aeronautical charges are higher under the dual-till approach compared to the single-till approach if non-aeronautical profit is greater than zero. Non-aeronautical profit is usually greater than zero, which is why airports prefer the dual-till approach and airlines prefer the single-till approach. To demonstrate the two approaches, let us consider the Airport Monitoring report of the Australian Competition and Consumer Commission (ACCC) 2019. In the Airport Monitoring report, the ACCC presents operational and financial statistics pertaining to Perth International Airport, which is the major airport in the capital city Perth, Western Australia on page 108. Over the financial year ending June 30, 2019, the airport had carried 14.6m passengers. The aeronautical operating expense of the airport is estimated by the ACCC to be A$145.7m. The total non-current aeronautical asset base of the airport is estimated to be A$977.7. The property, plant and equipment component of these non-current assets was estimated to be A$962.5. If we assume that these assets depreciate at the rate 4% this generates a depreciation expense of A$38.5m. Perth Airport’s current assets as at June 30, 2019, was valued at A$147.6m according to the Perth Airport annual report 2019. We assume that half of these assets are attributable to the aeronautical side of the business, which amounts to A$73.8. The total aeronautical asset base of Perth Airport is therefore: Perth Airport Aeronautical Asset Base = A$977.7 + A$73.8 = A$1,051.5 The building-block approach also includes the return on capital of the airport. The return on capital is computed by multiplying the aeronautical asset base of the airport by the Weighted Average Cost of Capital of the airport or the WACC, which is assumed to be 10%. This yields a capital cost equal to: Capital Cost = 10% u A$1,051.5 = A$105.15m The aeronautical costs of the airport that are determined under the dual-till building-block approach is the sum of the operating costs, the costs associated with the depreciation of aeronautical assets, and the aeronautical asset capital cost, which gives: Dual Till Total Aeronautical Costs = A$145.7 + A$38.5 + A$105.15 = A$289.35m The dual-till building block approach to estimating the airport charge using (6.21) is then the aeronautical costs divided by the number of passengers, which gives: Dual Till Airport Charge at Perth =

ଶ଼ଽ.ଷହ ଵସ.଺

= ‫ܣ‬$19.82

The figure A$19.82 is estimated to be the average aeronautical cost of the airport. Under the single-till approach, we subtract the non-aeronautical operating profit from aeronautical costs and divide the result by passenger movements. According to the Australian Competition and Consumer Commission 2019 the operating profit of the non-aeronautical Perth business was A$161.7m in 2019. If we subtract this operating profit from the dual till aeronautical costs, then we obtain:

Airline Cost

159

Total Aeronautical Costs – Non-Aeronautical Operating Profit = A$289.35m – A$161.7 = A$127.65 If we divide this net cost figure by the number of passengers, we obtain the single-till airport charge using (6.22): Single-Till Airport Charge =

ଵଶ଻.଺ହ ଵସ.଺

= ‫ܣ‬$8.74

The dual-till charge is twice as much as the single-till charge, which is clearly why Perth Airport would prefer to apply the dual-till approach. The rationale behind using the single-till approach is largely that the non-aeronautical demand for an airport is complementary to aeronautical demand. As such, the profit that an airport enjoys from its non-aeronautical activities is largely driven by its aeronautical business. When an airline brings capacity into an airport, not only does it benefit the airport from the aeronautical revenue that it earns, but it also benefits the airport from the non-aeronautical revenue that it earns. Airlines argue that since they are instrumental in providing non-aeronautical benefits to the airport then this should be factored into any calculation of the airport charge.

6.7.3 Reporting Airport Charge Costs When airlines report the costs associated with consuming aeronautical services at the airport, it is often not made explicit or isolated in their external reporting. These costs are often hidden because they are combined with other cost categories such as terminal, route navigation and en-route charges as well as handling charges. Airport charges can represent a non-trivial component of airline costs, particularly short-haul, low-cost carriers, and therefore it is important that we understand exactly what these costs might be. In the case of Singapore Airlines for example, their airport charges are hidden in the cost line “landing, parking, and overflying charges” in its external reporting. This cost category represented 8.4% of Singapore Airline’s total non-fuel cost over the 3 months to September 2019. Over the same period, Norwegian Air Shuttle (a major low-cost carrier in Europe operating out of Oslo) reported that airport and air traffic control expenses were 19.5% of total non-fuel costs while Ryanair reported that airport and handling charges represented 26.6% of total non-fuel cost over the September 2019 quarter (Airline Intelligence and Research Database 2021). Even if airport charges are mixed with other cost lines, we can divide these mixed costs by passenger numbers to obtain a maximum or ceiling level of airport charges on a per passenger basis. To understand what I mean by this, consider Air France/KLM, which has a cost line in its reporting that is landing fees and en-route charges.36 Landing fees, which are a large component of airport charges, are mixed with enroute charges, or the costs associated with getting guidance from air traffic control whilst in the air on the flight path from one airport to another. Landing fees are likely to be the most significant component of this mixed cost category. If we divide this cost by the number of passengers carried, we obtain an upper limit on the likely estimate of the average landing charge per passenger paid by Air France/KLM to airports. A time series of this estimated charge per passenger is presented in Figure 6-19 below over the annual period 2004 to 2019. Air France/KLM Landing & En Route Fees (€ per Passenger) 25

24.64

24 23 22

22.06

22.69

21 20.33

20

2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 Source: Airline Intelligence and Research 2021

Fig. 6-19: Air France/KLM Landing Fees and En-route Charges per Passenger

36

Airport charges are often a mix of fees associated with landing on the runway, using the taxiways and apron area, as well as fees for using the gates and terminals, and fees for using aircraft parking areas. It may also include utility fees and fees associated with the provision of ground handling services. Some airlines, such as Air France/KLM, will appear to use the words “landing charges” as synonymous with airport charges, but landing charges strictly speaking only refer to the charges associated with using the runway, taxiway and apron areas of airports.

Chapter 6

160

We can see in Figure 6-19 that the average landing fee and en-route charge per passenger paid by Air France/KLM between 2004 and 2019 varies between €20.33 and €24.64. This is the maximum possible estimate that we would expect for the average airport charge that Air France/KLM pays, although the actual average charge is likely to be materially lower than this.

Quiz 6-6: Airport Charges 1.

The following is a link to the financial report of Turkish Airlines for the 9 months to September 2020 https://investor.turkishairlines.com/documents/financial-results/30_09_2020-usd-ifrs-rapor-ing.pdf. On page 43, the airline presents its results for Airport Expenses over the period 1 January to 30 September 2020. What percentage of the total cost of sales is airport expenses for Turkish Airlines over the 9 months to 30 September?

2. (a) (b) (c) (d)

Which activity driver is generally the most important in understanding movements in airport charges? Fleet units. Passengers Carried. Seats Carried. Available Freight Tonne Kilometres.

3. What is the dual-till approach to setting airport charges? (a) Costs are separated into aeronautical and non-aeronautical, and the airport charge is set by dividing aeronautical costs by passenger movements. (b) Costs are separated into aeronautical and non-aeronautical, and the airport charge is set by dividing nonaeronautical costs by passenger movements. (c) Costs are separated into aeronautical and non-aeronautical, and the airport charge is set by subtracting from aeronautical costs the profit made on non-aeronautical services and dividing the result by passenger movements. (d) Aeronautical and non-aeronautical costs are added together and divided by passenger movements. 4. What is the single-till approach to setting airport charges? (a) Costs are separated into aeronautical and non-aeronautical, and the airport charge is set by dividing aeronautical costs by passenger movements. (b) Costs are separated into aeronautical and non-aeronautical, and the airport charge is set by dividing nonaeronautical costs by passenger movements. (c) Costs are separated into aeronautical and non-aeronautical, and the airport charge is set by subtracting from aeronautical costs the profit made on non-aeronautical services and dividing the result by passenger movements. (d) Aeronautical and non-aeronautical costs are added together and divided by passenger movements. 5.

The following information pertains to Brisbane International Airport over the 12 months to June 30, 2019. Brisbane Airport is in the southern part of the state of Queensland, Australia. Use this information to estimate the dual-till airport charge and the single-till airport charge for Brisbane Airport over the 12 months to June 30, 2019. Metric Passenger Numbers Aeronautical Operating Costs Non-Aeronautical Operating Profit Aeronautical Property Plant and Equipment Rate of Depreciation of Property, Plant and Equipment Non-current Assets of the Aeronautical Business Current Assets of the Aeronautical Business WACC

2018-19 Value 24.0m A$213.5 A$308.3 A$2,487.6 4% A$2,877.2 A$133.7 10%

6.8 Maintenance Costs There are three broad sets of maintenance costs. The first is referred to as station engineering or line maintenance, the second involves fixing an aircraft fault when it is reported by crew, and the third involves a major maintenance check. Line maintenance is a brief check of the aircraft while it is waiting at the gate between its arrival time and its departure time. This involves inspecting parts of the aircraft such as wheels, brakes, and oil levels. It will also involve brief running repairs that the aircraft sensors indicate the aircraft requires. Most aircraft receive around 12 hours of line maintenance per week. The maintenance associated with fixing a fault depends on the nature of the fault. This may be a quick fix that

Airline Cost

161

involves hours of maintenance effort or a much longer fix involving days of maintenance effort. It may also involve a part being flown into the location of the aircraft from another port to enable the fault to be fixed. There are three major aircraft maintenance checks which bundle together hundreds of tasks. The first major maintenance check is the A-check. The A-check occurs every 8 to 10 weeks. During this check, the maintenance engineers will change the filters, lubricate key systems such as the hydraulics in the control surfaces that steer the aircraft, and they will also inspect all the emergency equipment on the aircraft, such as the inflatable slides. A typical A-check on a Boeing 737 aircraft will take somewhere between 6 and 24 hours. The next major maintenance check is called the C-check. The C-check takes place every 1 and a half years to 2 years depending on the aircraft type and takes around 3 weeks to perform. Often when an aircraft is undertaking a C-check the airline will also upgrade its cabin interior at the same time so that the aircraft is not losing any further time out of operation. The last major maintenance check is the D-check. The D-check is also referred to as the C4 or the C8 check depending on the aircraft that is being maintained. The D-check is performed every 6 years and the entire aircraft is essentially dismantled and put back together. Every item in the cabin area is removed, including the seats, the toilets, the galleys, and the overhead bins, so that engineers can inspect the metal skin of the aircraft, both inside and out. The D-check will also require the engines to be taken off the aircraft so they too can be tested and the aircraft skin around the engines can be inspected. The D maintenance check will also require the engineers to remove and overhaul the landing gear. To enable this the aircraft is supported on massive jacks that can withstand the weight of the aircraft. The D-check also requires taking apart all the aircraft systems. These systems are checked, repaired, or replaced and reinstalled. The D-check is an extremely expensive exercise, often costing millions of dollars and takes between 3 to 6 weeks. At the completion of the D-check the aircraft is almost like a brand-new aircraft. With improvements in aircraft technology, aircraft are designed so that they require less maintenance. For example, the Boeing B787 Dreamliner only needs a D check every 12 years rather than every 6 years, which is the requirement for older aircraft such as the Boeing 747 or 767.37 It is clear from the description of the major maintenance checks above that the maintenance expense of the airline will mostly depend on the age of the aircraft, or the number of hours or kilometres flown by the aircraft. It will also depend on the size of the aircraft. It follows that the maintenance expense of the airline can be treated as a simple linear function of the available seat kilometres that are flown, which can be written as follows: CM = cM u ASK

(6.23)

where cM represents maintenance cost per ASK or MASK for short.

Quiz 6-7: Maintenance Costs 1. (a) (b) (c) (d)

What is line maintenance? A major maintenance check that involves pulling the entire aircraft apart and pulling it back together again. A maintenance check that involves an engineer examining the aircraft for small lines or cracks in the fuselage. A brief check or ‘go around’ of the aircraft while it is waiting at the gate. It involves engineers fixing a fault that has been reported by pilots on its way to its destination.

2. (a) (b) (c) (d)

Which of the major maintenance checks takes the longest? A-check. B-check. C-check. D-check.

3. (a) (b) (c) (d)

Which type of maintenance check would involve the airline pulling all engines off the aircraft? A-check. B-check. Line maintenance. D-check.

4. (a) (b) (c) (d) (e)

Which of the following are most likely to be activity drivers of the timing of maintenance checks? The age of the aircraft. The number of flying hours of the aircraft. The number of aircraft movements. All the above. (a) and (b) only.

37 If you would like to read more about aircraft major maintenance checks, Qantas Newsroom 2016 has published an excellent readerfriendly news article on the topic.

Chapter 6

162

6.9 Ground Handling Ground handling is any aircraft service that takes place below the wing of the aircraft when the aircraft is on the ground and parked at a terminal gate. It includes services such as cabin services, which involves cleaning the cabin and replenishing onboard consumables such as tissues, soap, toilet paper and inflight magazines. It also includes catering services, which involves unloading unused food and drink from the aircraft and loading fresh food and drink for passengers and crew. Another ground handling service is ramp services, which includes any service that takes place on the ramp or apron area, including aircraft marshalling (guiding the aircraft into the gate and out from the gate), aircraft towing by pushback tractors, lavatory drainage, water cartage, baggage handling, baggage sorting and transportation, freight handling, refuelling, ground power, airstairs or aerobridges, de-icing, and anti-icing. Ground handling also involves passenger services, which includes the provision of an airport sales desk, check-in counter services, providing gate-arrival services, lost and found services, staffing transfer counters, customer service counters, airline lounges and special passenger and VIP services. Airlines use one of four different types of ground handling services. Firstly, they can use their own ground handling teams, which will usually be the case for large airlines operating at relatively large local airports. Secondly, they can use the ground handling services provided by airports, for which they pay the airport a ground handling charge. Thirdly, they can use dedicated ground handling companies, which we will discuss in more detail below. Finally, they can use the ground handling services provided by other airlines. Airlines may in fact use a combination of all four types of ground handling services depending on the location of the airport, whether the relevant flight is domestic or international, the frequency of services to the airport, and the existence of dedicated ground handling companies at the airport. The major non-airline and non-airport ground handling companies include Swissport, which is the world’s largest provider of cargo and aircraft ground handling, servicing 835 client-companies and handling 230 million passengers and 3.9 million flights per year (Swissport 2020). Menzies Aviation is another major non-airline ground handling company that has been in the ground handling business since 1833 and is present at 202 locations in over 34 countries across 6 continents and serves 200 million passengers and 1.2 million aircraft per year (Menzies Aviation 2020). The ground handling company Dnata has over 46,000 employees globally, operates in 35 countries, is present at 126 airports and serves over 320m airline passengers per year (Dnata 2020). To assess the key costs associated with ground handling and the key activity drivers, I present below the revenue and operating costs and the operating statistics of Dnata over the 12 months to March 31, 2020, which is information that is presented in the Emirates Group 2019-20 annual report – refer to Table 6-8 below. As indicated in Table 6-8, the most important cost line of Dnata is staff cost, which in 2019-20 was 41.2% of total cost. Labour is used for most ground handling activities such as handling baggage and freight, cleaning the aircraft, providing check-in services to passengers, and marshalling aircraft. The second most important cost associated with providing ground handling services at airports is referred to by Dnata as direct costs. This will include the cost of energy that is used to power vehicles, the cost of replenishing the consumables on the aircraft and the cost of water that is used on the aircraft. These direct airport costs represent 9.6% of total Dnata costs. Another important cost category is depreciation and amortisation. Ground handling involves the use of a diverse range of equipment, including but not limited to refueler trucks and their pumps, filters and hoses, tugs and tractors, ground power units, which supplies power to the aircraft parked on the ground, buses to move passengers from the terminal to the aircraft or between terminals, container loaders and transporters, potable water trucks, lavatory service vehicles, catering vehicles, belt loaders, passenger boarding stairs, an de-icing and anti-icing vehicles. The wear and tear of these pieces of equipment forms the company’s depreciation costs. Dnata’s depreciation and amortisation cost was 6% of total cost in 2019-20. Most of these key ground handling costs will depend on the number of aircraft movements and the number of passengers on the aircraft, which will drive the volume of baggage that must be moved, the number of passengers that must be checked-in, the number of meals that must be catered for and the number of consumables that must be replenished. To model ground handling costs, the simplest way this is done is to assume a constant cost per passenger moved, cG and then multiply this by the number of passengers moved as follows: CG = cG u PAX

(6.24)

Figure 6-20 presents the handling cost per passenger of the Singapore Airlines Group at a quarterly frequency between March 2012 and December 2019. Figure 6-20 indicates that handling cost per passenger in the December quarter 2019 was SG$47, which is around the same level as it was 7 years prior. The handling cost per passenger in the case of the Singapore Airlines Group cycles up and down around a mean of around the SG$50 mark, with these significant cycles likely to be driven by movements in the exchange rate. With all Singapore Group’s services involving international travel, half of the airline’s ground handling activity will involve foreign airports, the cost of which is likely to be denominated in foreign currency. This means that the airline Group’s ground handling costs will be heavily affected by the cycle in the Singapore dollar exchange rate against other currencies.

Airline Cost

163

2019-20 Revenue in AED m International Airport Operations Travel Services UAE Airport operations Inflight Catering Other Services Total

2018-19

3,940 3,537 3,171 3,313 262 14,233

Operating Costs AED m Employee costs Direct costs Travel services Airport Operations Inflight Catering Other Depreciation and amortisation Facilities related expenditure Sales and marketing expenses Information technology costs Impairment of intangible assets Corporate overheads Total Operating Costs

5,875

5,386

2,534 1,364 1,352 32 853 501 321 320 193 908 14,253

2,476 1,350 1,070 67 459 788 370 246 78 851 13,141

Operating Statistics International Aircraft Handled International Cargo Handled (‘000 tonnes) UAE airports aircraft handled UAE airports cargo handled (‘000 tonnes) Meals Uplifted (m) Average Employee Strength

492,657 2,231 188,210 698 93.5 46,211

488,225 2,364 210,514 727 70.9 45,004

Source: Emirates Group Annual Report 2020

Table 6-8: Dnata Revenue and Costs

Singapore Airlines Handling Costs per Passenger (SGD) 58 $56

56 54 52 50

$49

48 $47

46 44

$45 Dec-19 Sep-19 Jun-19 Mar-19 Dec-18 Sep-18 Jun-18 Mar-18 Dec-17 Sep-17 Jun-17 Mar-17 Dec-16 Sep-16 Jun-16 Mar-16 Dec-15 Sep-15 Jun-15 Mar-15 Dec-14 Sep-14 Jun-14 Mar-14 Dec-13 Sep-13 Jun-13 Mar-13 Dec-12 Sep-12 Jun-12 Mar-12

Source: Airline Intelligence and Research Database 2021

Fig. 6-20: Singapore Airlines Handling Charges per Passenger

Chapter 6

164

Quiz 6-8: Ground Handling Costs 1. (a) (b) (c) (d)

Which of the following would not be considered a part of ground handling costs? Buying jet fuel. The cost of refuelling the aircraft. The labour costs associated with replacing toilet paper on the aircraft. The cost of employing labour to add baggage to the aircraft and take baggage from the aircraft.

2.

Name three major non-airline ground handling companies in global aviation.

3. (a) (b) (c)

What is the ground handling service referred to as aircraft marshalling? When aircraft are told by air traffic control into which gate they are required to take the aircraft. When chocks are added to the wheels of aircraft once they are stationary at the gate. It is a managerial position within the ground handling group that involves directing ground handling labour to perform various tasks. (d) A person who guides the aircraft into the gate parking position using glow sticks or flags.

4. (a) (b) (c) (d)

What are the different types of ground handling service arrangements? The airline uses its own ground handling crew and equipment. The airline uses the ground handling crew and equipment supplied by the airport. The airline uses the ground handling crew of another airline. All the above.

5. (a) (b) (c) (d)

Which of the following would be considered a key activity driver of ground handling costs? The number of aircraft movements. The flight hours. The number of pilots onboard. The wages paid to ground handling labour.

6.10 Airline Cost Functions and Runs As we saw in chapter 4, section 4.1 there are three key runs that airlines face – the short run, the medium run and the long run. In this section we examine how we model the costs of the airline over these three different runs.

6.10.1 Total Long Run Cost Function We saw in sections 6.4 through to 6.9, at equations (6.10), (6.15), and (6.20) to (6.24) that in the case of the 6 most important cost categories, which are likely to cover between 70% and 80% of an airline’s total cost, the three most important drivers of airline costs are passenger movements, available seat kilometres and the number and value of fleet units. We can summarise this by stating the long run cost function of the airline in the following way: CLR = cPAX u PAX + cASK u ASK + cFleet u Fleet + K

(6.25)

Equation (6.25) describes the total cost of the airline in a situation in which the airline has sufficient time to vary all three of the key activity drivers – PAX, ASKs and Fleet. The term K captures any other cost that is not influenced by these three driver variables, such as the airline’s overhead. As the number of fleet units can only be altered by the airline over a long horizon, ASKs can only be altered over medium and long run horizons and PAX can be altered over all horizons then as long as the fleet can be changed in (6.25), then (6.25) represents a long run cost expression.

6.10.2 Medium Run Cost Function In the medium run the airline does not have enough time to alter the number of fleet units but it does have enough time to alter the ASKs and PAX. In this case the number of fleet units is fixed. Let us define the stock of fixed fleet units of the airline as ‫ܨ‬. Substituting this definition into equation (6.25) and we obtain the medium run cost function: CMR = cPAX u PAX + cASK u ASK + cFleet u F + K

(6.26)

The term cFleet u ‫ ܨ‬+ K represents the fixed cost of the airline in the medium run, consisting of the long run fixed cost, K plus the medium run fixed cost, cFleet u ‫ ܨ‬which is determined by the fixed fleet of the airline. The remaining terms represent the variable costs of the airline in the medium run.

Airline Cost

165

We would expect that there is a relationship between the number of available seat kilometres and the number of passengers over the medium run. In other words, if the airline increases capacity this in turn will lead to more seats and therefore more passengers being carried. We can write this in the following way using a linear relationship: PAX = a0 + a1ASK

(6.27)

Substituting (6.27) into (6.26) and we obtain the simplified medium run cost function as follows: CMR = c u ASK + KMR

(6.28)

where c { cPAX u a1 + cASK and KMR { cFleet u ‫ ܨ‬+ K + cPAX u a0. Equation (6.28) essentially says that over a medium run horizon the cost function of the airline has a component that varies with the number of ASKs, c u ASK and a component that does not vary with the number of ASKs, K*. The component that varies with the number of ASKs will include the airline’s fuel costs, its cabin and pilot crew salaries and other expenses, airport and terminal navigation charges, enroute charges, maintenance costs and ground handling costs. These will usually represent more than half of the airline’s cost base. The way that c is calculated in (6.28) is by identifying those costs that vary with available seat kilometres and dividing these through by available seat kilometres. A simple way of identifying the ASK-varying costs involves subtracting from total operating cost depreciation and lease costs, or aircraft capital costs, and any other overhead costs, including non-operational labour costs and property rental costs. We draw the medium run cost function (6.28) in Figure 6-21 below. Medium Run Airline Costs

C = c*uASK + K*

C = cuASK + K*

K* Available Seat Kilometres

0 Fig. 6-21: Illustration of the Medium Run Cost Function

We can see in Figure 6-21 that the medium run cost function starts at K* on the vertical axis when ASKs equal zero and then rises as ASKs increase. The medium run cost function pivots upward, as indicated by the dashed line, if there is an increase in variable costs per ASK or c. This will be the case, for example, if fuel prices increase, if the local currency depreciates against the US dollar and pilot or cabin crew wages increase. The medium run cost function shifts upward if the costs that are independent of ASKs in the medium run increase. This will occur, for example, when the airline increases its operating fleet units, or it hires more non-operational labour.

6.10.3 Short Run Function When an airline decides to alter the passenger seat factor in the short run, this can alter some of the airline’s costs. Put another way, on a flight from Hong Kong to Chengdu in an A320 aircraft with 180 seats, the flight will have higher operating costs when the passenger seat factor is 90% compared to when it is 60%. The operating costs are higher for the flight with the higher passenger seat factor because there are costs that go up as more passengers are carried on the flight. This will typically include costs such as airport charges, terminal navigation charges, passenger in-air meals and beverages, merchant service fees, cleaning the inside of the aircraft post-flight (which is part of ground handling costs), baggage and check-in desk handling (again, a part of ground handling costs), and a component of fuel costs. The category a “component of fuel costs” requires further explanation. Most of the fuel that is burnt on a particular flight is attributable to the weight of the empty aircraft hull, the distance that is flown by the aircraft on the flight, and the age of the aircraft. There is a non-trivial percentage of fuel that is burnt on a flight, however, that is attributable to

Chapter 6

166

the weight that is added to the aircraft by passengers and freight. This additional, load-related fuel burn will depend on the length of the flight and could vary between 5% and 15% of the fuel consumed on a flight. If we take our medium run cost function (6.26) and fix the level of capacity at ‫ ܭܵܣ‬then we obtain the short run total cost function as follows: CSR = cPAX u PAX + cASK u ‫ ܭܵܣ‬+ cFleet u F + K

(6.29)

In the case of the short run cost function (6.29), there is a variable component that depends on the number of passengers, cPAX u PAX, and a fixed component that does not vary with the number of passengers, cASK u ‫ ܭܵܣ‬+ cFleet u ‫ ܨ‬+ K. We can simplify the expression at (6.29) by simplifying the fixed cost expression as follows: CSR = cPAX u PAX + KSR

(6.30)

where KSR = cASK u ‫ ܭܵܣ‬+ cFleet u ‫ ܨ‬+ K. The term cPAX refers to the costs that vary with the number of passengers carried in the short term divided by the number of passengers carried. The fixed costs include the major cost categories that vary with airline capacity and the number and value of the airline’s fleet units, such as the component of fuel costs that is unrelated to the aircraft’s load, flight crew costs (including pilot and cabin crew costs), depreciation, amortisation, and lease costs (or aircraft capital costs), and maintenance costs. These fixed costs will represent 75% to 95% of the total costs of the airline in the short run. As the number of passengers carried in the short term is just equal to the passenger seat factor times the number of fixed seats then we can write the short run cost function of the airline at (6.30) as: CSR = cPAX u PSF u ܵҧ + KSR

(6.31)

If we were to graph this function with cost on the vertical axis and the passenger seat factor on the horizontal axis it simply looks like an increasing straight line that starts at a point on the vertical axis above the origin, and then increases at the rate cPAXuܵҧ as the passenger seat factor increases – refer to Figure 6-22 below. Short Run Airline Costs C = cPAXuPSFuܵҧ + KSR

KSR

0

1

Passenger Seat

Fig. 6-22: An Illustration of the Short Run Cost Function The key to building an estimate of the cost function in Figure 6-22 is in identifying those costs that are likely to be higher if the airline increases the number of passengers that it carries on a flight. For airline analysts who work within an airline, there will be an area within the airline that is able to identify these key cost categories and the likely volume drivers of those categories. This information can be used to determine the passenger-varying costs and the remaining costs that are invariant to changes in passengers carried. For analysts that work outside of the airline business, the task is more difficult because most airlines do not provide enough information in their external reporting to identify the costs that vary with passenger volumes and those that don’t. And it is difficult to identify those costs using statistical techniques, for example, using multiple regression techniques. This is because it is difficult to disentangle the separate effects of capacity and passengers carried on costs, because passengers carried are heavily correlated with capacity. There are a few carriers, however, that do provide enough detailed information to enable an estimate of the percentage of costs that are likely to vary with passengers carried for a given level of capacity. If we multiply this percentage by the estimate of the total cost of the flight and divide through by the number of passengers, then we obtain an estimate of the costs that vary with passengers carried per passenger. The airlines that provide a breakdown of passenger varying costs, and their estimate of the percentage of costs that vary with the passenger seat factor, are given in Table 6-9 below.

Airline Cost

Airline Singapore Airlines Copa Airlines Cebu Pacific Airlines Delta Air Lines GOL Air Lines Hawaiian Airlines Pegasus Airlines Norwegian Air Shuttle Turkish Airlines Asiana Airlines El Al Airlines

Passenger Varying Cost % Total Cost 13.3% 16.2% 15.2% 7.1% 7.2% 11.2% 11.1% 10.5% 9.6% 8.5% 12.1%

167

Time Period 12 months to June 2018 12 months to December 2016 12 months to June 2018 12 months to September 2018 12 months to June 2018 12 months to September 2018 12 months to December 2014 12 months to September 2018 12 months to December 2014 12 months to December 2016 12 months to December 2017

Source: Airline Intelligence and Research Database 2021

Table 6-9: Estimates of Passenger Varying Cost as % of Total Cost Table 6-9 indicates that the short run costs that vary with passengers is likely to range from 7.1% up to 16.2% of total operating cost. As airport and terminal navigation charges are likely to be the most significant component of the costs that vary with passengers carried, then passenger varying costs per passenger will depend heavily on the city pairs and airports that define the route that is being analysed. It will also depend on whether the route is domestic or international, with airport charges generally higher for international services compared to domestic services. Airport charges also tend to be a higher proportion of total cost for low-cost carriers than for full-service airlines and so this will need to be factored into any calculations.

Quiz 6-9: Cost Functions and Runs 1. Which of the following costs are likely to vary with passengers carried and the passenger seat factor in the short term? (a) Cabin crew costs. (b) Flight crew costs. (c) Airport charges. (d) Ground handling. (e) Aircraft depreciation. (f) Operating lease costs. (g) Terminal navigation costs. (h) Route navigation costs. (i) Catering costs. (j) Maintenance costs. (k) Interest charges on airline debt. (l) The CEO’s salary. 2. An airline has a total cost per passenger of $250. It carries 2.5m passengers per year and it knows that 12% of its costs in the short run vary with the number of passengers. What is the passenger-varying cost per passenger of the airline? (a) 0.12 u 250 (b) 0.12 u 250 y 2.5 (c) 250 y 2.5 (d) 0.12 u 2.50 + 250 3. The following is the income statement and key operating data of the Brazilian low-cost carrier Azul Airlines over the 3 months to September 2020. Income Statement (R$ million) OPERATING REVENUES Passenger Cargo and other revenues Total operating revenues OPERATING EXPENSES Aircraft Fuel Salaries, wages, and benefits Depreciation and amortization Landing fees

3Q20 624.5 180.8 805.3 226.1 309.6 445.9 73.8

Chapter 6

168

Traffic and customer servicing Sales and marketing Maintenance materials and repairs Other operating expenses Total Operating expenses Operating income

46.4 60.2 111.4 (220.5) 1,053.0 (247.7)

Operating Data ASKs (m) Domestic International Revenue Passengers (‘000) Block Hours Departures Operating Passenger Aircraft

3Q20 3,240 2,967 273 2,380 45,325 27,313 139

You are to use this information to arrive at the following estimates. (a) An estimate of the passenger varying costs, and the passenger varying costs per passenger. Assume that 10% of fuel costs depend on the load on the aircraft and landing fees are imposed on a per passenger basis. Use this information to estimate the short run cost function by analysing each cost category and determining whether that category will vary with the number of passengers when capacity and fleet units are fixed. Graph the short run cost function. (b) An estimate of the ASK-varying costs, and the ASK-varying cost per ASK. Assume that 70% of salaries, wages and benefits are attributable to operational capacity. Like part (a), you will need to analyse each cost category and determine whether that cost category is likely to change if ASKs increase, and the fleet units are fixed. Graph the medium run cost function.

6.11 Non-Fuel Cost Per ASK 6.11.1 What is Non-Fuel Cost per ASK? Non-fuel cost per ASK, or NFCASK for short, for a particular airline is defined as EBIT-level cost excluding fuel divided by the available seat kilometres or ASKs supplied by the airline (see also the definition provided in chapter 2). We can represent this by the following formula: Non-Fuel Cost per ASK (NFCASK) =

୓୮ୣ୰ୟ୲୧୬୥ େ୭ୱ୲ୱି୊୳ୣ୪ େ୭ୱ୲ୱ ୅୴ୟ୧୪ୟୠ୪ୣ ୗୣୟ୲ ୏୧୪୭୫ୣ୲୰ୣୱ

(6.33)

In the case of U.S. airlines, and other airlines that use miles as opposed to kilometres in their reporting metrics, the nonfuel unit cost metric that is used is non-fuel cost per available seat mile or NFCASM. This can be represented by the following formula: Non-Fuel Cost per ASM (NFCASM) =

୓୮ୣ୰ୟ୲୧୬୥ େ୭ୱ୲ୱି୊୳ୣ୪ େ୭ୱ୲ୱ ୅୴ୟ୧୪ୟୠ୪ୣ ୗୣୟ୲ ୑୧୪ୣୱ

(6.34)

In this section we wish to understand how NFCASK changes over time, and why NFCASK is different across airlines. NFCASK is an important unit cost metric and one that is popularly used and analysed by airlines, because it allows airlines to analyse unit cost without the influence of the spot price of jet fuel, which has a significant influence on CASK. As a first step in understanding NFCASK, we examine the changes in the NFCASK over time for four different airlines based in different parts of the world. Figure 6-23 below presents the time series movements in Qantas Group non-fuel cost per ASK between June 2000 and June 2020. The time series in Figure 6-23 is measured at a yearly frequency (12 months to June 30). If we decompose the time series movements into its main component parts, we see that there is an underlying upward trend between 2000 and 2014 with significant cyclical movement around that trend. Qantas NFCASK rose from 1.02 Australian cents per ASK in 2000 up to 3.14 Australian cents per ASK between 2000 and 2014. For the following three years there was a significant downward movement, resulting in a NFCASK low of 2.02 Australian cents per ASK by 2017, after which NFCASK has risen, finishing at 2.59 Australian cents by 2020. While inflationary pressures and higher wage rates will have affected NFCASK in an upward direction over this period there has also been significant downward pressure on NFCASK during certain periods over the 20-year horizon that is examined. The downward pressure on NFCASK is usually brought about by three variables - a stronger local currency unit (Australian dollar in this example) against the US dollar, the existence of returns to scale, which is the result of more ASKs being distributed over a certain level of costs that are invariant to ASKs, and higher levels of fuel, labour, and fleet productivity.

Airline Cost

169

Qantas NonFuel Cost per ASK (c/ASK) 3.7 3.14

3.2

2.59

2.7 2.2 2.02

1.7 1.2

1.02

0.7 Jun-20

Jun-19

Jun-18

Jun-17

Jun-16

Jun-15

Jun-14

Jun-13

Jun-12

Jun-11

Jun-10

Jun-09

Jun-08

Jun-07

Jun-06

Jun-05

Jun-04

Jun-03

Jun-02

Jun-01

Jun-00

Source: Airline Intelligence and Research Database 2021

Fig. 6-23: Qantas Group Non-Fuel Cost Per ASK Time Series As indicated in Figure 6-24 below, Singapore Airlines Group’s non-fuel cost per ASK has trended downward between 2003 and 2019. Singapore Airlines Group NFCASK (SG c/ASK)

8.5

8.21 8

7.83

7.5 7.20

7 6.5

6.27

6

2019

2018

2017

2016

2015

2014

2013

2012

2011

2010

2009

2008

2007

2006

2005

2004

2003

Source: Airline Intelligence and Research Database 2021

Fig. 6-24: Singapore Airlines Non-Fuel Cost Per ASK Time Series Singapore Airlines Group NFCASK has declined from 8.21 Singapore cents per available seat kilometre in calendar 2003 down to 6.27 Singapore cents per ASK in 2019. Singapore Airline Group’s NFCASK has cycled around this downward trend, with upward cycles between 2006 and 2010 and later between 2014 and 2016. The downward trend is likely to be attributable to the strength of the Singapore dollar against the U.S. dollar reducing the cost of aircraft depreciation and lease costs as well as the cost of other goods and services bought in a foreign currency. It is also likely to be attributable to an increase in labour, fuel, and fleet productivity. In Figure 6-25 we see that Southwest Airline’s NFCASM has increased in trend terms over a 3-decade horizon, from a low of 5.3 cents per ASK in 1990 up to 9.6 cents per ASK in 2019, with an acceleration in cost from 2008 onwards. The trend increase in the NFCASM of the airline will be attributable to general inflationary pressures. Resource costs such as wages, airport and navigation charges, the price of aircraft, property rental and lease rates, and utility and energy costs often increase over time with the CPI. As Southwest is a U.S. carrier then exchange rate movements will have had little impact on the long-term direction of Southwest NFCASM. While productivity improvements may have helped to keep the rate of NFCASM inflation relatively subdued over the period 1992 to 2007, it appears to have not been enough to generate strong NFCASM inflation pressure thereafter.

Chapter 6

170

Southwest Airlines NFCASM (US c/ASM) 10.6 9.6

9.6 8.6 7.6 6.6

6.4

5.6 5.3

4.6

2019 2018 2017 2016 2015 2014 2013 2012 2011 2010 2009 2008 2007 2006 2005 2004 2003 2002 2001 2000 1999 1998 1997 1996 1995 1994 1993 1992 1991 1990 Source: Airline Intelligence and Research Database 2021

Fig. 6-25: Southwest Airlines Non-Fuel Cost Per ASK Time Series Figure 6-26 presents the movements in non-fuel cost per ASK of Air France/KLM over the period 2004 to 2019. Air France/KLM NFCASK (Euro c/ASK)

7.5 7.4 7.3

7.30 7.28

7.2 7.1 7 6.9 6.8

6.78

6.7 2019

2018

2017

2016

2015

2014

2013

2012

2011

2010

2009

2008

2007

2006

2005

2004

Source: Airline Intelligence and Research Database 2021

Fig. 6-26: Air France/KLM Non-Fuel Cost Per ASK Time Series We can see in this case that NFCASK exhibits two distinct trends – a downward trend between 2004 and 2014 and an upward trend thereafter. This is likely to be driven by the cyclical movements in the US dollar against the Euro over these two periods.

6.11.2 Analytical Representation of NFCASK In this section we wish to understand the types of variables that are likely to drive non-fuel cost per ASK over time, and the types of variables that are likely to result in one airline having a different non-fuel cost per ASK than other airlines. The way that we approach this question of understanding the key drivers is by using an analytical approach. Let us re-examine the long run costs in equation (6.25). If we subtract the fuel costs at (6.10) from (6.25) we obtain the long run total cost function excluding fuel: NFCLR = CLR – CFuel = cPAX u PAX + cASK u ASK + cFleet u Fleet + K - P୎୉୘ × E × ASK × ɀ

(6.35)

We can simplify (6.35) by altering the coefficient attached to ASKs as follows: ‫כ‬ u ASK + cFleet u Fleet + K NFCLR = CLR – CFuel = cPAX u PAX + c୅ୗ୏

(6.36)

Airline Cost

171

‫כ‬ where ܿ஺ௌ௄ { cASK - P୎୉୘ × E × ɀ. To determine long run non-fuel cost per ASK we divide (6.36) by ASKs as follows: ୒୊େై౎ ୅ୗ୏

=

ୡౌఽ౔ ×୔୅ଡ଼ ୅ୗ୏

+

ୡూౢ౛౛౪ ×୊୪ୣୣ୲ ୅ୗ୏

If we define fleet productivity as

ଵ ఉ

+ =

୏ ୅ୗ୏

‫כ‬ + c୅ୗ୏

஺ௌ௄ ி௟௘௘௧

(6.37)

, we recognise that ASK = SEAT u PASL and we assume that the other long

run costs K = 0, then we can re-write (6.37) as: ୒୊େై౎ ୅ୗ୏

=

ୡౌఽ౔ ×୔ୗ୊ ୔୅ୗ୐

‫כ‬ + c୊୪ୣୣ୲ × Ⱦ + c୅ୗ୏

(6.38)

Equation (6.38) tells us that long run non-fuel cost per ASK will depend on passenger-driven costs per passenger (cPAX), the passenger seat factor (PSF), the passenger average sector length (PASL), fleet productivity (E), the average aircraft capital cost of the airline per fleet unit (cFleet) and ASK driven costs per ASK excluding fuel. We will use information about these 6 factors to understand why NFCASK changes over time, and why it will be different for different airlines.

6.11.3 What Drives Non-Fuel Cost Per ASK over Time? Section 6.11.2 sets out 6 key variables that determine non-fuel cost per ASK. We use these 6 variables to determine which forces are likely to drive changes in the non-fuel cost per ASK over time. 6.11.3.1 Inflation Inflation drives changes in NFCASK over time because it influences the passenger-driven costs per passenger, cPAX, the ‫כ‬ and the fleet driven costs per fleet unit, cFleet. Inflation may influence these three ASK-driven non-fuel cost per ASK, ܿ஺ௌ௄ parameters in different ways. The first is the impact of inflationary pressures on the wages of operational labour, which includes ground handling labour wage rates which will form a part of cPAX and the cabin crew and pilot wage rates which ‫כ‬ . The second is the impact of inflationary pressures on airport and air services costs and the passwill form a part of ܿ஺ௌ௄ through of these costs into higher landing fees and route navigation charges which impacts cPAX. The third is the impact of inflationary pressures on aircraft prices and depreciation, and the premium that is paid for operating lease costs which forms a part of cFleet. For large western economies, the general price level typically increases over time at an average rate of between 2% and 3%. This tends to force the trend in NFCASK over a long period of time to be positive. 6.11.3.2 Exchange Rate There are many different non-fuel cost items that are potentially denominated in US dollars, and denominated in other currencies aside from US dollars, that will be affected by exchange rate changes. As a result, the exchange rate is likely ‫כ‬ and cFleet in (6.38). The specific way that the exchange rate influences these variables to influence the variables cPAX, ܿ஺ௌ௄ include the following. Firstly, aircraft are purchased in US dollars, they are leased in US dollars and aircraft spare parts are bought in US dollars, so that cFleet is affected by exchange rate changes. Secondly, engineering costs that are ‫כ‬ . outsourced to overseas jurisdictions often require payment in US dollars, which means the exchange rate affects ܿ஺ௌ௄ Thirdly, overseas port costs such as airport charges, overfly charges, terminal navigation charges and ground handling ‫כ‬ fees are denominated in US dollars and in other currencies, which means the exchange rate will influence cPAX and ܿ஺ௌ௄ in the case of international operations. There are also airline costs that are denominated in local currency that will be affected by exchange rate movements. In these cases, local suppliers to airlines denominate contracts in local currency even though they make their own payments in US dollars for the products they are supplying. These suppliers assume all exchange rate risk and decide whether they will pass exchange rate changes through into the prices they charge airline customers. Examples of products that airlines pay for in local currency that may be affected by exchange rate movements include consulting services, back-office furniture, IT equipment and software, and land-based vehicles that are purchased by the airline (such as the executive motor vehicle fleet). 6.11.3.3 Fleet Productivity An increase in fleet productivity means that the airline is using fewer aircraft to generate the same level of output, or it is using the same fleet of aircraft to generate higher levels of output. An improvement in fleet productivity lowers NFCASK because an increase in output for the same fleet means that fleet related costs are unchanged at the same time as ASKs increase, resulting in aircraft capital costs per ASK falling.

Chapter 6

172

6.11.3.4 Passenger Average Sector Length The most important of the six drivers of NFCASK is the passenger average sector length. Equation (6.38) tells us that the higher the passenger average sector length, the lower non-fuel cost per ASK. The rationale behind this result is that there are some non-fuel costs that do not depend on the distance that the aircraft travels, such as airport and terminal navigation charges and passenger airmeals amongst other costs. When we divide these costs by a metric that does depend on the distance that the aircraft travels, such as ASKs, it generates a term that falls as the distance the aircraft travel increases. The extent to which non-fuel cost per ASK is lower as the passenger average sector length is higher, depends on the proportion of total long run costs that do not depend on the distance that the aircraft travels. For most airlines this will be somewhere between 20% and 30% of total costs. 6.11.3.5 Cost Reduction and Product Improvement Programs When airline earnings are under pressure, airlines often initiate structural programs to reduce costs. This will involve finding productivity improvements across the business, aggressively pursuing supplier pricing improvements through procurement strategies, and generally making the business leaner. If an airline requires significant cost savings, then it needs to take chunks of cost out of the major cost lines, notably manpower, depreciation, and amortisation and airport charges. Non-operational labour costs are usually the first port of call for airlines when attempting to take costs out of the business when it is under earnings pressure. It is also the case that airlines will try and reduce the complexity of its fleet. This involves reducing the number of fleet types, which reduces the costs associated with training cabin and flight crew, as well as reducing the costs associated with the need to buy and store spare parts.

6.12 Adjusting Non-Fuel Cost per Available Seat Kilometre for Sector Length Airlines and airline analysts often wish to compare the non-fuel costs of one airline to other airline peers. This is most often performed by comparing NFCASK. The ultimate objective of comparing costs across airlines is to determine the extent to which the difference in product offering by the two airlines matches the extent to which the costs are different, since an airline that offers a superior quality air travel service will naturally have higher costs. To identify the component of NFCASK that differs from that of other airlines because of product differences it is necessary to purge from the differences in NFCASK all the other forces that may cause the differences in cost. The most important of these is the passenger average sector length. It is problematic comparing the NFCASK of two airlines if the two airlines have materially different average sector lengths because we know from the analysis of section 6.11 that an airline with a high average sector length will naturally have a lower NFCASK. We adjust the NFCASK of an airline for the average sector length by performing the following steps: (1) choose a set of airlines that you wish to benchmark, (2) find the average of the average sector length of these airlines, (3) find the percentage difference between the average sector length found at step (2) and the average sector length of the individual airlines in the benchmarking exercise, (4) multiply the number found at step (3) for each airline by an estimate of the proportion of costs that do not vary with distance, which I usually set at between 0.2 and 0.3, (5) find 1 plus the calculation at step (4) for each airline in the dataset, and (6) multiply the estimates at step (5) by the estimates of nonfuel cost per ASK for each of the corresponding airlines. The estimated number at step (6) is the non-fuel cost per ASK adjusted for sector length. Figures 6-27 and 6-28 below present the NFCASK for a set of full-service and low-cost airlines respectively domiciled in Asia in calendar 2014. One could look at these Figure 6-29 and 6-30 graphs and make the mistake in believing that the Japanese airlines are the highest non-fuel cost per ASK full-service carriers in the market and Garuda the lowest non-fuel cost per ASK full-service carrier. And similarly, Jet Airways Lite, the former Indian low-cost carrier, has the highest non-fuel cost per ASK of the low-cost carriers and Air Asia Malaysia has the lowest non-fuel cost of the low-cost carriers. Until one makes an adjustment for sector length, however, it is not possible to hold this view although it may indeed be the case that these airlines have the highest and lowest NFCASK. Figures 6-29 and 6-30 below provide some insight into the adjustments to NFCASK that needs to be made so that non-fuel cost per ASK can be compared across airlines. We can see in Figure 6-29 that NFCASK for the full-service airlines is generally lower when the average sector length is higher. For airlines that are positioned above the firm line in Figure 6-29, they tend to have higher non-fuel cost per ASK after taking into consideration an adjustment for sector length. This includes the Japanese carriers, ANA and JAL, Qantas, and Virgin Australia. The airlines that lie below the firm line have lower non-fuel costs per ASK than expected after adjusting for sector length, and includes Garuda, Singapore, China Southern, Air China and Jet Airways. For the airlines that lie above the firm line this may be for a variety of reasons, including the airline is domiciled in an economy that has higher labour costs than most other economies in Asia (JAL, Qantas and Virgin Australia are examples of such carriers), the airlines have lower levels of resource productivity than other airlines, these airlines operate to much higher load factors, which drives up passenger driven costs, and the airline may be generally mismanaged, with too many overheads and other inefficient practises.

Airline Cost

173

Non-Fuel Cost per ASK CY14 (US c/ASK) 9.06

8.50

7.91

7.01

6.28

6.23

5.75

5.48

5.45

5.37

5.26

EMIRATES

AIR CHINA

THAI AIRWAYS

CHINA SOUTHERN

JET AIRWAYS

CATHAY PACIFIC

AIR NZ

CHINA EASTERN

VIRGIN AUSTRALIA

QANTAS

JAL

ANA

3.16

2.49

GARUDA

4.79

SINGAPORE AIRLINES

10 9 8 7 6 5 4 3 2 1 0

Source: Airline Intelligence and Research Database, 2021

Fig. 6-27: Non-Fuel Cost Per ASK in CY14 – Full-Service Airlines in Asia

2.70

2.69

CEBU

TIGER AIRLINES

2.05

AIR ASIA MALAYSIA

2.75

AIR ASIA THAILAND

JET AIRWAYS LITE

2.78

AIR ASIA INDONESIA

Non-Fuel Cost per ASK CY14 (US c/ASK) 5.38 6 5 4 3 2 1 0

Source: Airline Intelligence and Research Database 2021

Fig. 6-28: Non-Fuel Cost Per ASK in CY14 – Low-Cost Airlines in Asia The same rationale applies to Figure 6-30 below in the low-cost carrier case. For the low-cost carriers in Figure 630 Jet Airways Lite appears to have significantly higher NFCASK on a sector length adjusted basis while Air Asia Malaysia and Cebu Pacific tend to have lower NFCASK than expected. When benchmarking non-fuel costs per ASK we would therefore argue that Cebu Pacific and Air Asia Malaysia have the lowest costs in Asia while Jet Airways lite has the highest.

Quiz 6-10: Non-Fuel Cost per ASK 1. Why is non-fuel cost per ASK an important metric for comparison with other airlines when an airline competes in the highly price sensitive segment of the market? (a) Because in the price sensitive segment of the market, it is imperative to keep prices low which requires the airline to keep costs low. (b) Because in the price sensitive segment of the market, it is imperative to keep prices high which requires the airline to keep costs low. (c) Because in the price sensitive segment of the market, it is imperative to keep prices high which requires the airline to keep costs high. (d) Because in the price sensitive segment of the market, what is of most importance is the product offered by the airline, which in turn requires the airline to keep non-fuel costs low.

Chapter 6

174

Non-Fuel Cost per ASK (US c/ASK) 10 ANA 9 JAL 8 7 6

Virgin Air NZ Cathay Jet Airways Air China Thai

China East China South

5

Qantas

Emirates

4 3 2

Singapore

Garuda

1 0 0

1,000

2,000

3,000

4,000

5,000

6,000

7,000

8,000

9,000

Average Sector Length (km) Source: Airline Intelligence and Research Database 2021

Fig. 6-29: Non-Fuel Cost Per ASK Versus Average Sector Length in CY14 – Full-Service Airlines in Asia Non-fuel Cost per ASK (US c/ASK) 6 Jet Airways Lite 5 4 AA Thailand

Cebu

3

Tiger

AA Indonesia 2

AA Malaysia

1 0 700

900

1,100

1,300

1,500

1,700

1,900

2,100

Average Sector Length (km) Source: Airline Intelligence and Research Database 2021

Fig. 6-30: Non-Fuel Cost Per ASK Versus Average Sector Length in CY14 – Low-Cost Airlines in Asia 2. What is the most likely explanation for a falling non-fuel cost per ASK? (a) Employing more labour. (b) Higher wages and inflation. (c) Higher levels of productivity. (d) A depreciation in the local currency unit against the US dollar. 3. When comparing the non-fuel cost per ASK of one airline to that of another airline it is important to adjust for the average sector length of the airlines because: (a) There are some costs that do not vary with distance. (b) There are no costs that vary with distance. (c) There are some costs that are increasing in distance. (d) There are some costs that are decreasing in distance.

Airline Cost

175

4. What is the relationship between non-fuel cost per ASK and the average sector length? (a) If airline A is the same as airline B in every single respect except that airline A has a higher average sector length than B than airline A will have a lower non-fuel cost per ASK. (b) If airline A is the same as airline B in every single respect except that airline A has a longer average sector length than B than airline A will have the same non-fuel cost per ASK. (c) If airline A is the same as airline B in every single respect except that airline A has a lower average sector length than B than airline A will have a lower non-fuel cost per ASK. (d) If airline A is the same as airline B in every single respect including that airline A has the same average sector length as B, then airline A will have a lower non-fuel cost per ASK. Use the following calendar 2016 information for various airlines to answer the next 3 questions. Airline

Non-Fuel Cost in USD m $8,520 $7,535 $3,274 $2,546

Qantas Group Singapore Group Virgin Australia Air New Zealand Average

Available Seat Kilometres m 149,733 141,844 40,460 41,108

Average Sector Length km 2,304 4,254 1,612 2,205 2,594

5. Singapore Airlines Group non-fuel cost per ASK in US cents is equal to: ଻,ହଷହ×ଵ଴଴ (a) (b) (c) (d)

ଵସଵ,଼ସସ ଵସଵ,଼ସସ×ଵ଴଴ ଻,ହଷହ ଵସଵ,଼ସସ

଻,ହଷହ×ଵ଴଴ ଻,ହଷହ×ଵ଴଴ ସଶହସ

6. Singapore Airlines average sector length differs from the average by: (ସ,ଶହସିଶ,ହଽସ) (a) (b) (c) (d)

ଶ,ହଽସ (ଶ,ହଽସିସ,ଶହସ)

ଶ,ହଽସ ଶ,ହଽସ ସ,ଶହସ

(ସ,ଶହସିଶ,ହଽସ)

ଶ,ଷ଴ସାସ,ଶହସାଵ,଺ଵଶାଶ,ଶ଴ହ

7. Singapore Airlines non-fuel cost per ASK adjusted for sector length is: (ସ,ଶହସିଶ,ହଽସ) ଻,ହଷହ×ଵ଴଴ (a) ቂ1 + ቃ× (b) ቂ1 + (c) ቂ1 െ (d) ቂ1 െ

ଶ,ହଽସ ଵସଵ,଼ସସ (ସ,ଶହସିଶ,ହଽସ)×଴.ଶହ ଻,ହଷହ×ଵ଴଴

ଶ,ହଽସ (ସ,ଶହସିଶ,ହଽସ)

ቃ×

ቃ × 0.25 ×

ଶ,ହଽସ (ସ,ଶହସିଶ,ହଽସ)×଴.ଶହ ଶ,ହଽସ

ଵସଵ,଼ସସ ଻,ହଷହ×ଵ଴଴ ଵସଵ,଼ସସ ଻,ହଷହ×ଵ଴଴

ቃ × 0.25 ×

ଵସଵ,଼ସସ

6.13 Airline Economies of Scale and Scope 6.13.1 Airline Economies of Scale Economies of scale in simple practical terms represents the reduction in the average cost of the airline (essentially NFCASK) as the airline gets bigger. The simplest way to think about airline economies of scale is that there are a range of costs that do not vary as the airline grows, which means that as ASKs increase, these fixed costs per ASK must fall. Examples of these permanently fixed costs include the salaries paid to the senior executives of the airline, the lease payments paid for the property that houses the back-office or non-operational staff, and the utility, stationary, IT and other expenses that are incurred to support non-operational staff. There are also other cost benefits associated with an airline increasing in size. Bigger airlines purchase more aircraft and other goods and services in bulk. This enables them to secure bulk purchase or volume discounts. These volume discounts pass-through into lower airline unit costs. These discounts could relate to a range of goods and services including but not limited to aircraft and engine purchases and thus the depreciation line of the airline, spare parts, food and beverages for passengers, accommodation for technical and cabin crew, utility payments, printing and stationary, furniture for the head office, photocopying and scanning equipment, IT and telecommunications equipment, and smart phone contracts, and land transport (executive fleet, buses, cars, airport transport equipment).

176

Chapter 6

Bigger airlines often have more stable earnings streams because their network coverage is wider, they have greater control over market fares, they have stronger alliance relationships and greater diversification of their business investments. A less volatile earnings stream translates into a higher credit rating and lower risk premium, which lowers interest rates that large airlines pay on debt, and it lowers the returns that are required to be paid to shareholders. Large airlines with a strong reputation and a greater ability to offer non-wage benefits, such as a wider staff travel network, are also more likely to attract labour at lower wage rates. This in turn leads to lower staff cost per ASK.

6.13.2 Airline Economies of Scope Economies of scope represents the reduction in unit cost that an airline can enjoy because it has costs that are common to two or more flying brands, or two or more business units. For example, flying groups such as the International Airlines Group, the Singapore Airlines Group and the Qantas Group have both full-service and low-cost airline flying brands. There are some common costs that can be shared across those flying brands. For example, Treasury Risk Management can perform its functions for the airline group rather than individual flying brands within the group. This functional area would be in place even if the airline group flew only a single brand. The costs associated with funding this area can therefore be shared across all flying brands across the group. As another example, an airline may wish to set-up a dedicated air freight business. Rather than building a team with completely new management for the air freight business, the airline sets up a team from existing management, which is tasked with managing the freight business along with other tasks for the passenger airline business. The cost of this labour is shared across the passenger business and the freight business.38

Quiz 6-11: Economies of Scale and Scope 1. What is airline economies of scale? (a) When airline costs fall over a short horizon as it grows bigger. (b) When an airline’s fixed costs fall as output increases. (c) When an airline has common costs that can be allocated across 2 or more brands or flying segments. (d) When airline costs fall over a long horizon as it grows bigger. 2. What is airline economies of scope? (a) When an airline has common costs that can be allocated across 2 or more brands or flying segments. (b) When an airline’s unit cost falls as the airline gets bigger. (c) When productivity increases because of the more efficient use of resources. (d) When an airline’s unit cost is lower because of the ability to obtain volume discounts.

38

If you wish more information about economies of scope, consult the gold-standard book by Tirole 2001 on pages 16-17 and 20.

CHAPTER 7 MAXIMISING SHORT RUN AIRLINE PROFIT

The goal of most airlines, particularly airlines that have private owners, is to maximise profit. By maximising profit the airline maximises the money that is returned to airline owners. Given the consistently below average earnings performance of the airline sector over the past few decades, it would seem reasonable to form the view that airlines are not likely to be setting key operational levers at optimal levels. One such operational lever is the passenger seat factor. The passenger seat factor is a short run operational lever. This means that it can be varied over a short time horizon, which is a period of up to 12 months. This is to be distinguished from the frequency of airline services and the number and type of fleet units that are operated by the airline, which take longer than 12 months to vary in any substantial way and are thus fixed in the short run. The objective of this chapter is to determine the passenger seat factor, and therefore the average airfare, that maximises profit, when the frequency of airline services and the fleet are fixed. We refer to this as the short run optimisation problem of the airline. The key to solving the short run optimisation problem of the airline is to properly construct the short run airline profit function. The short run airline profit function describes the operating profit of the airline as a function of the passenger seat factor. Most of the hard work in constructing the short run airline profit function has already been done in this textbook, with chapter 4 presenting readers with the short run airline revenue function and chapter 6 the short run airline cost function. In this chapter we will bring the revenue and cost functions together to formulate the short run airline profit function. We will show that the short run airline profit function increases at a decreasing rate as the passenger seat factor increases, eventually reaches a maximum and then begins to fall as the passenger seat factor is increased too far. The objective of building a model of short run profit in this way is to identify the passenger seat factor that corresponds with the peak of the short run profit function. We will demonstrate in this chapter that the passenger seat factor that corresponds with maximum short run profit occurs when marginal revenue is equal to marginal cost. This is a very well-known profit maximising rule in economics. An airline should continue increasing the passenger seat factor if the addition to operating revenue exceeds the addition to operating cost, in which case marginal revenue exceeds marginal cost, because this adds to profit. It should do this up until marginal revenue just equals marginal cost because this is the point at which the change in profit is equal to zero or profit is maximised. This analysis is of limited use to an airline if it cannot be put into practise. It is for this reason that the analytical methods that are developed in this chapter are illustrated using actual airline data. I will also show how the methods are implemented in practise, and how the techniques can be developed in Microsoft Excel. After reading through this chapter, you will be able to determine if an airline is setting passenger seat factors at profit maximising levels, and if they are not, the extent to which airline profit can be increased if its passenger seat factor is increased or decreased to profit maximising levels.

7.1 Linear Average Airfare Function 7.1.1 Theory To analyse the short run airline profit function, we subtract the short run airline cost function from the short run airline revenue function. The short run airline cost function from chapter 6, equation (6.31) is repeated below for your convenience: CSR = cPAX u PSF u Sത + KSR

(7.1)

The short run airline revenue function from chapter 4, equation (4.7), is repeated below for your convenience: RSR = a u PSF u Sത + b u PSF2 u Sത

(7.2)

If we subtract our short run cost function at (7.1) from the short run revenue function at (7.2) then we obtain the short run airline profit function: Short Run Profit = a u PSF u Sത + b u PSF2 u Sത - cPAX u PSF u Sത - KSR

(7.3)

Chapter 7

178

The short run profit function (7.3) describes how the airline’s profit changes on a particular route when the airline targets a higher passenger seat factor, and the capacity and fleet units of the airline are fixed. We present a schematic diagram of how the short run profit function varies with the passenger seat factor in Figure 7-1 below. Short Run Airline Revenue/Cost

Profit

C = cPAXuPSFu‫܁‬ത + KSR

B Loss

A R=(auPSF + buPSF2)uSത

KSR Loss

0

PSFA

PSFB

1

Passenger Seat Factor

Fig. 7-1: An Illustration of the Short Run Profit Function In Figure 7-1 the short run revenue function is the concave down parabola and the total cost function is the upward sloping straight line. The area between the revenue curve and the cost line represents the profit of the airline. There are two areas of loss indicated in Figure 7-1 – for a passenger seat factor that is too low (below PSFA) and for a passenger seat factor that is too high (above PSFB). The loss at a low passenger seat factor occurs because the airline cannot recover the high fixed costs with a low volume of passengers. The loss at the higher passenger seat factor stems from the fact that the airline has allowed the passenger seat factor to rise too far, resulting in an average airfare that has fallen too far to recover unit costs. The sweet spot for the airline is somewhere between points A and B in Figure 7-1, where the revenue curve lies above the cost line, and the airline is making money. Where does the airline choose a passenger seat factor between the two points A and B? To answer this question, we need to use some simple calculus.39 The passenger seat factor which generates the maximum level of short run profit is found by differentiating (7.3) with respect to the passenger seat factor and setting the result equal to zero as follows: ୢ୔୰୭୤୧୲ ୢ୔ୗ୊

= a + 2 × b × PSF െ c௉஺௑ = 0

(7.4)

The result (7.4) is called the first order condition. The component on the right-hand side of (7.4), a0 + 2a1PSF, is called marginal revenue. This measures the addition to revenue when the airline decides to sell one more seat on the aircraft. The component on the right-hand side of (7.4), cPAX, is called the marginal cost. This measures the addition to total cost when the airline decides to sell one more seat on the aircraft. The first order condition (7.4) says that the airline sets marginal revenue equal to marginal cost. If we solve (7.4) for the PSF, we obtain the passenger seat factor which maximises short run profit: PSF ‫= כ‬

ୡುಲ೉ ି௔

(7.5)

ଶ×ୠ

Equation (7.5) says the following in words:

39

In this section, the only calculus you will need to know is how to differentiate a quadratic function of a single variable, which is hopefully something you attempted at high school. If you need a refresher, see Haeussler and Paul 1987 chapter 11. There is also a good chapter on parabolas and quadratic functions in chapter 4 which may be helpful. The quadratic function takes the form Y = aX2 ୢଢ଼ ୢଢ଼ + bX + c. The first derivative of this function is ୢଡ଼ = 2aX + b. To find the maximum or minimum turning point we set ୢଡ଼ = 0 and solve for X, which gives X* =

ିୠ ଶୟ

.

Maximising Short Run Airline Profit

Optimal Passenger Seat Factor =

179

୑ୟ୰୥୧୬ୟ୪ ୔ୟୱୱୣ୬୥ୣ୰ େ୭ୱ୲ିୖୣୱୣ୰୴ୟ୲୧୭୬ ୅୧୰୤ୟ୰ୣ ଶ×ୗୣ୬ୱ୧୲୧୴୧୲୷ ୭୤ ୅୴ୣ୰ୟ୥ୣ ୅୧୰୤ୟ୰ୣ ୲୭ ୲୦ୣ ୔ୟୱୱୣ୬୥ୣ୰ ୗୣୟ୲ ୊ୟୡ୲୭୰

The optimal passenger seat factor at (7.5) will depend on the extent to which the average airfare falls as the seat factor goes up (b), the extent to which the operating cost of the airline goes up as the passenger seat factor goes up (cPAX), and the reservation average airfare, (a). ି௔ You will note that the passenger seat factor which maximises revenue is PSF ‫= ככ‬ . The profit maximising ଶ×௕

௖

passenger seat factor is therefore lower than the revenue maximising passenger seat factor by the quantity ುಲ೉. The ଶ×௕ extent to which the profit maximising seat factor is lower than the revenue maximising seat factor clearly depends on the marginal passenger cost. In the case of profit maximisation, the airline takes into consideration the fact that it costs more to increase the passenger seat factor, which reduces profit. Figure 7-2 below presents the short run profit maximising seat factor in graphical form. SR Revenue/Costs C = cPAXuPSFu‫܁‬ത + FC A

B

2

R=(a0uPSF + a1u PSF )uSത Fixed Cost

Max Profit

Max Revenue

0

1

Passenger Seat Factor

Fig. 7-2: Profit Maximising Passenger Seat Factor The short run profit maximising seat factor occurs when we take the firm cost line and shift it up until it is just tangential to the revenue curve. This occurs at point A in the graph. We then project this down to the horizontal axis and this gives us the profit maximising passenger seat factor. Point A in the graph is the maximum possible distance between the revenue curve and the total cost line. The revenue maximising passenger seat factor occurs where the tangent to the revenue curve has a slope of zero in Figure 7-2, which occurs at the point B where the dashed tangent line has a slope of zero or is horizontal. You will note that the passenger seat factor at the profit maximising point A is lower than the passenger seat factor where revenue is maximised at point B. To determine the profit maximising average airfare, we simply substitute the expression at (7.5) into the average airfare function (4.5) (in chapter 4), which gives: Average Airfare = a + b × ቀ

ୡౌఽ౔ ିୟ ଶ×ୠ

ቁ=

ୡౌఽ౔ ାୟ ଶ

(7.6)

The optimal average airfare is the average of the reservation airfare and the variable cost per passenger of the airline. The maximum profit that the airline can earn in the short run is found by substituting the average airfare and the passenger seat factor at the optimum into (7.3). This is left as an exercise for the interested reader, which I am sure you are.

7.1.2 Airline Illustration – Aegean Airlines Aegean Airlines is a full-service carrier based in Athens, Greece and is the national carrier of Greece. It is also the largest Greek airline by passengers, network size and number of fleet units. Its main destinations from Athens include the major capital cities in Europe, such as, but not limited to, Paris, Rome, Berlin, Moscow, Madrid, Geneva, and London (Aegean Airlines Destinations 2021). An analyst for Aegean Airlines has estimated the following linear relationship between the average airfare and the passenger seat factor per flight: P = 150 - 80 u PSF The average seat count on Aegean aircraft is 150. This implies that the short run revenue function of the airline per flight is:

Chapter 7

180

RSR = 150 u PSF u 150 – 80 u PSF2 u 150 = 22,500 u PSF – 12,000 u PSF2 The passenger-varying cost per passenger of the airline is €20 and the fixed cost per flight is €7,500. This implies the following short run cost function per flight: CSR = 20 u Passengers + 7,500 The number of passengers is equal to the passenger seat factor multiplied by the average seat count, or Passengers = 150 u PSF. Substituting this into the short run cost function yields: CSR = 20 u (PSF u 150) + 7,500 = 3,000 u PSF + 7,500 Subtracting the short run cost function from the short run revenue function generates the Aegean Airlines short run profit function: SSR = 22,500 u PSF – 12,000 u PSF2 - 3,000 u PSF - 7,500 = 19,500 u PSF – 12,000 u PSF2 – 7,500 Let us now draw the short run revenue, cost, and profit functions for the average Aegean Airlines flight. We start-off in Figure 7-3 below by drawing the average airfare as a function of the passenger seat factor and underneath this the short run revenue function. We can see that the average airfare function of Aegean Airlines falls as the passenger seat factor increases towards 100%, with a reservation airfare of €150 and an average airfare equal to €70 when the seats on the average flight are fully occupied. This gives rise to the revenue curve in the bottom segment of Figure 7-3, which starts to bend as the passenger seat factor increases, reaching a maximum at a passenger seat factor of around 94% and then begins to turn down for a passenger seat factor greater than 94%. Figure 7-4 below presents the short run cost function in the top of the graph and the revenue and cost functions together in the bottom component of the graph. We can see that the cost function starts at the short run fixed cost level of €7,500 per flight, and then increases by €20 for every additional passenger on the flight up until the plane is full at a passenger seat factor of 100%. The bottom component of Figure 7-4 places the cost function and the revenue function in the same graph. We can see that the cost function meets the revenue function in two places, and the revenue function rises above the cost function for a passenger seat factor exceeding 63%. To find the passenger seat factor that maximises short run profit, we take the cost line in the bottom component of Figure 7-4, and we shift that cost line upward until it is just tangential to the revenue curve, which occurs at the filled circle. This tangent cost line is the dashed line in Figure 7-4. The point of tangency is the point which generates the maximum difference between the revenue curve and the cost line. If we take the point of tangency at the filled circle and project this down to the horizontal axis, then we obtain the passenger seat factor for Aegean Airlines which maximises short run profit. Reading this off the bottom component of Figure 7-4 this is just above 81%. Let us now demonstrate how this profit maximising passenger seat factor can be found by using calculus. If we differentiate our Aegean Airlines short run profit function with respect to the passenger seat factor, we obtain the following equation: dɎୗୖ = 19,500 െ 24,000 × PSF = 0 dPSF If we solve this equation for the passenger seat factor, we obtain the passenger seat factor at which the slope of the profit function is equal to zero: PSF ‫= כ‬

ଵଽ,ହ଴଴ ଶସ,଴଴଴

= 81.25%

We can also see this result by graphing the profit function for Aegean Airlines directly – see Figure 7-5 below. We can see in Figure 7-5 that the passenger seat factor which is associated with the uppermost point on the profit curve is 81.25%. Let us now demonstrate how we obtain this profit maximising passenger seat factor when we use formula (7.5). From the average airfare function, we know that the reservation average airfare is a = €150, the marginal passenger cost is cPAX = €20 and the sensitivity of the average airfare to the PSF is b = -80. Substituting these parameters into (7.5) and we obtain: Passenger Seat Factor ‫= כ‬

௖ି௔ ଶ×௕

=

ଶ଴ିଵହ଴ ିଶ×଼଴

= 81.25%

We can see that the uppermost point on the Aegean Airlines short run profit function occurs at a passenger seat factor of 81.25%. This generates a maximum level of profit of around €422 per flight. This is found by taking the uppermost point of the profit function, which is indicated by a small circle, and projecting this point horizontally across to the

Maximising Short Run Airline Profit

181

Aegean Airlines Average Airfare (Euro) 160 150 140 130 120 110 100 90 80 70 98 95 92 90 87 84 81 78 76 73 70 67 64 62 59 56 53 50 48 45 42 39 36 34 31 28 25 22 20 17 14 11 8 6 3 0

Passenger Seat Factor (%) Aegean Airlines Revenue (Euro) 12,088 10,088 8,088 6,088 4,088 2,088 88 98 95 92 90 87 84 81 78 76 73 70 67 64 62 59 56 53 50 48 45 42 39 36 34 31 28 25 22 20 17 14 11 8 6 3 0

Passenger Seat Factor (%) Fig. 7-3: Average Airfare and Short Run Revenue Functions Aegean Airlines vertical axis. In 2019 the airline operated 115,765 flights so that a profit of €422 per flight amounts to an annual profit of almost €50m at these flight numbers. The maximum level of profit is found analytically by substituting the optimal passenger seat factor of 81.25% back into the profit function as follows: Ɏ‫כ‬ୗୖ = 19,500 u 0.8125 – 12,000 u 0.81252 – 7,500 = €421.875 The average airfare at the optimum is found by substituting the passenger seat factor into the average airfare function, which is equal to: P* = 150 - 80 u 0.8125 = €85 This can also be found by using formula (7.6) as follows: Average Airfare =

ଶ଴ାଵହ଴ ଶ

= €85

The Aegean Airlines passenger seat factor in 2019 was 83.94% while the airline’s revenue per passenger was found to be €87.30 (Aegean Airlines Financial Results 2020, 2-5). The airline is therefore operating at parameters in 2019 that appear to be reasonably close to the short profit maximising findings in the numerical example of this section.

Chapter 7

182

Aegean Airlines Short Run Cost Per Flight (Euro) 10900 10400 9900 9400 8900 8400 7900 7400 99 96 93 90 87 84 81 78 75 73 70 67 64 61 58 55 52 49 46 44 41 38 35 32 29 26 23 20 17 15 12 9 6 3 0

Passenger Seat Factor (%) Aegean Airlines Revenue/Cost (Euro) 12,088 10,088 8,088 6,088 4,088 2,088 88 98 95 92 90 87 84 81 78 76 73 70 67 64 62 59 56 53 50 48 45 42 39 36 34 31 28 25 22 20 17 14 11 8 6 3 0

Passenger Seat Factor (%) Fig. 7-4: Short Run Cost and Revenue Functions Aegean Airlines

Quiz 7-1. Short Run Profit 1.

In the case of the first order condition (7.4), which is repeated below: a + 2 × b × Passenger Seat Factor െ c௉஺௑ = 0 Identify the marginal revenue and the marginal cost components. In the case of the marginal revenue component, identify the positive volume effect of an increase in the passenger seat factor and the negative price effect.

2.

Using equation (7.5), what is the profit maximising passenger seat factor if the reservation average airfare is $350, the passenger varying cost per passenger is $100 and the sensitivity of the average airfare to the passenger seat factor is -300? What is the profit maximising average airfare?

Maximising Short Run Airline Profit

183

Aegean Airlines Profit (Euro) 1,000

€422

0 -1,000 -2,000 -3,000 -4,000 -5,000 -6,000 -7,000

81.25%

-8,000

99 96 93 90 87 84 81 77 74 71 68 65 62 59 56 53 50 47 43 40 37 34 31 28 25 22 19 16 12 9 6 3 0

Passenger Seat Factor (%) Fig. 7-5: Short Run Profit Function Aegean Airlines 3.

Air Arabia is a low-cost carrier located in the United Arab Emirates, with headquarters at Sharjah International Airport (SHJ). The airline operates to 170 destinations in the Middle East, North Africa, Asia, and Europe (Air Arabia Destinations 2021). The average airfare function of the airline is P = 650 – 400 u PSF, where P is denominated in United Arab Emirates Dirham or AED. The passenger varying cost per passenger of the airline is AED 30. The fixed cost per flight of the airline is AED 30,000. Air Arabia operates a 180 seat A320 on its flights. Use this information to answer the following questions. (a) Find the average airfare charged by Air Arabia at a passenger seat factor of 50% and a passenger seat factor of 80%. (b) What is the reservation airfare? (c) Interpret the coefficient attached to the PSF in the average airfare function. (d) Find the short run revenue function. Determine the passenger seat factor which maximises the short run revenue function. (e) What is the short run cost function? (f) What is the short run profit function? (g) Find the passenger seat factor which maximises short run profit. Compare this to your answer at (d) above. (h) Find the average airfare at the short run profit maximising passenger seat factor. Compare this to the average airfare at the revenue maximising passenger seat factor. (i) Find the maximum value of short run profit. (j) Find the elasticity of the average airfare to the passenger seat factor at the profit maximising passenger seat factor. Interpret the elasticity.

7.2 Cobb Douglas Average Airfare Function 7.2.1 Theory When the average airfare function is described by a Cobb-Douglas function, the profit function is the short run revenue function described by equation (4.16) in chapter 4 minus the short run cost function (7.1). The short run Cobb Douglas revenue function is repeated below for your convenience: RSR = A u PSF1-H u Sത

(7.7)

Subtracting the cost function (7.1) from the revenue function (7.7) generates the following short run profit function: Short Run Profit Function = A u PSF1-H u Sത - cPAX u PSF u ܵҧ – KSR

(7.8)

We can see how profit is determined in this case in Figure 7-6 below. Unlike the linear average airfare function example, in the case of the Cobb-Douglas average airfare function example the revenue curve increases at a decreasing rate but never falls. This means that there is less chance that the linear cost function meets the revenue curve at a high point for the passenger seat factor. In this case there is only one loss making zone, which is to the left of point A in the diagram in the case in which the passenger seat factor is too low.

Chapter 7

184

Short Run Airline Revenue/Costs A u PSF1-H u ‫܁‬ത

C = cPAXuPSFu‫܁‬ത + KSR

Profit

A KSR

Loss

0

1

Passenger Seat Factor

Fig. 7-6: The Short Run Profit Function with Cobb-Douglas Average Airfare What is the short run profit maximising passenger seat factor? In Figure 7-6, it is described by the maximum distance between the revenue curve and the straight cost line, which is somewhere to the right of point A. To answer this question analytically however, we must use calculus. We first differentiate the profit function (7.8) with respect to the passenger seat factor and set the result equal to zero as follows: ୢ୔୰୭୤୧୲ ୢ୔ୗ୊

= (1 െ ɂ) × A × PSF ିக െ c୔୅ଡ଼ = 0

(7.9)

The condition (7.9) is a marginal revenue equals marginal cost condition, which we can write as: (1 െ ɂ) × A × PSF ିக = c୔୅ଡ଼

(7.10)

The left-hand side of (7.10) is marginal revenue and the right-hand side is marginal cost. For an equilibrium to take place the left-hand side of (7.10) must be positive because we know that marginal cost, cPAX is positive. This in turn requires that H, which is less than zero, is less than 1 in absolute terms. This means that the elasticity of the average airfare to a change in the passenger seat factor is inelastic. If we solve (7.10) for the passenger seat factor, we obtain the passenger seat factor that maximises short run profit when the average airfare function is Cobb-Douglas: ୡ

ౌఽ౔ ቃ Passenger Seat Factor ‫ = כ‬ቂ(ଵିக)×୅

షభ ഄ

(7.11)

Equation (7.11) can also be written in words as follows: ୑ୟ୰୥୧୬ୟ୪ ୔ୟୱୱୣ୬୥ୣ୰ େ୭ୱ୲

Optimal Passenger Seat Factor = ቂ(ଵା୅୧୰୤ୟ୰ୣ

୉୪ୟୱ୲୧ୡ୧୲୷ )×୓୰୥ୟ୬୧ୡ ୈୣ୫ୟ୬ୢ

ቃ

భ ఽ౟౨౜౗౨౛ ుౢ౗౩౪౟ౙ౟౪౯

The optimal passenger seat factor depends on the level of organic demand (A), the marginal passenger cost (cPAX) and the elasticity of the average airfare to a change in the passenger seat factor (H). The profit maximising average airfare is found by substituting (7.11) into the Cobb-Douglas average airfare function which gives: ୡ

಍ ಍

ౌఽ౔ ቃ = Average Airfare‫ = כ‬A × ቂ(ଵିக)×୅

ୡౌఽ౔ ଵିக

(7.12)

The average airfare at the optimum at (7.12) is therefore the marginal passenger cost divided by 1 plus the elasticity of the average airfare to the passenger seat factor. The optimum can also be shown graphically – refer to Figure 7-17 below.

Maximising Short Run Airline Profit

185

Short Run Airline Revenue/Costs RSR = A u PSF1-H u ‫܁‬ത

Profit

KSR

C = cuPSFu‫܁‬ത + KSR

A Loss

0

PSF*

1

Passenger Seat Factor

Fig. 7-7: Optimal Short Run Profit with Exponential Average Airfare The optimum in Figure 7-17 is the point where the slope of the cost line (dashed line) is tangential to the revenue curve, generating the profit maximising passenger seat factor PSF*.

7.2.2 Aeroflot Illustration Aeroflot is the flag carrier and largest airline of Russia. It is headquartered in Moscow with a hub at Sheremetyevo International Airport (SVO). Aeroflot flies to 146 destinations in 52 countries (Aeroflot Destinations 2021). The airline’s passenger revenue per passenger or average airfare on a representative flight is described by the following function: P = 124 u PSF-0.7 where P is defined in US dollars. The average seat count for Aeroflot flights is 165 seats. This means that the short run revenue function of the airline per flight is the passenger revenue per passenger function times the number of passengers as follows: RSR = 124 u PSF-0.7 u PSF u 165 = 20,460 u PSF0.3 The airline’s passenger varying cost per passenger is US$45 and the fixed cost per flight is US$10,000. This means that the short run total cost function per flight is: CSR = 45 u Passengers + 10,000 = 45 u PSF u 165 + 10,000 = 7,425 u PSF + 10,000 The short run profit function is the short run revenue function less the short run cost function, which yields: SSR = 20,460 u PSF0.3 – 7,425 u PSF – 10,000 The short run revenue, cost and profit functions are presented in Figure 7-8 below. We can see in Figure 7-8 that the top component presents the revenue curve with the straight cost line superimposed on the revenue curve. The gap between the revenue curve and the cost line represents the Aeroflot profit per representative flight. Below the 12% passenger seat factor mark, the airline makes a loss on the flight because it does not sell enough tickets to passengers. If we shift the cost line upward keeping the slope unchanged until it is just tangential to the revenue curve, we obtain the point where Aeroflot maximises profit - refer to the dashed line in the top component of Figure 7-8. This occurs at a passenger seat factor that is in the high 70’s. If we examine the bottom component of Figure 7-8, we see that this tangency point coincides with the uppermost point on the profit curve. Let us now be more specific about the passenger seat factor which maximises Aeroflot short run profit by examining the first order condition for profit maximisation. The first order condition is found by differentiating the short run profit function with respect to the passenger seat factor and setting the result equal to zero as follows: ୢ஠౏౎ ୢ୔ୗ୊

= 0.3 × 20,460 × PSF ି଴.଻ െ 7425 = 0

Chapter 7

186

Aeroflot Revenue and Cost (US$ m) 25,000 Revene

Cost

20,000

15,000

10,000

5,000

0 99% 96% 93% 90% 87% 84% 81% 78% 75% 72% 69% 66% 63% 60% 57% 54% 51% 48% 45% 42% 39% 36% 33% 30% 27% 24% 21% 18% 15% 12% 9% 6% 3% 0%

Passenger Seat Factor Aeroflot Profit (US$ m) 4,000 2,000 0 -2,000 -4,000 -6,000 -8,000 -10,000 -12,000 99% 96% 93% 90% 87% 84% 81% 78% 75% 72% 69% 66% 63% 60% 57% 54% 51% 48% 45% 42% 39% 36% 33% 30% 27% 24% 21% 18% 15% 12% 9% 6% 3% 0%

Passenger Seat Factor Fig. 7-8: Short Run Revenue, Cost and Profit Functions for Aeroflot Solving this expression for the passenger seat factor yields the profit maximising passenger seat factor for Aeroflot on an average flight: PSF ‫ = כ‬ቀ

଻,ସଶହ

షభ బ.ళ

ቁ ଴.ଷ×ଶ଴,଺ସ଴

= 77.1%

This result implies that the passenger seat factor which maximises the short run profit of the average Aeroflot flight is 77.1%, which is consistent with the uppermost point of the profit function in Figure 7-8. If we substitute this optimal seat factor into the average airfare function, we obtain the average airfare at the optimum, which is: P* = 124 u (PSF*)-0.7 = 124 u (0.771)-0.7 = US$148.76 If we substitute the optimal passenger seat factor into the short run profit function, we obtain the maximum amount of profit that can be earned by Aeroflot per flight, which is US$3,200. Aeroflot flew almost 450,000 flights over the 12 months to February 2020 (Aeroflot Traffic Statistics 2020). If the airline would have earned US$3,200 on each of those flights it would have earned over the space of a year US$1.4b in profit, which I suspect would have made its shareholders very happy.

Maximising Short Run Airline Profit

187

Quiz 7-2 Short Run Profit Function with Cobb Douglas Average Airfare Function Consider the following Cobb-Douglas average airfare function for the short haul operations of the Irish national carrier Aer Lingus P = 165 u PSF-0.8, where P is defined in Euro. Aer Lingus is a wholly owned subsidiary of the International Airlines Group, which includes the flying brands British Airways, Iberia, Vueling and LEVEL. Aer Lingus headquarters is on the grounds of Dublin Airport, with the airline flying to 93 destinations throughout Asia, Europe, and North America, with its short haul operations to Europe including destinations in France, Germany, Greece, Italy, the Netherlands, Croatia, Portugal, Switzerland, Spain, and the UK (Aer Lingus Destinations 2021). The passenger varying cost per passenger of the airline in the case of its short haul operations is €40 and fixed cost per flight averages €7,500. The average seat count of the aircraft operated by Aer Lingus in the case of its short haul operations is 180. Use this information to answer the following questions. (a) What is the airfare elasticity with respect to the passenger seat factor? How is this interpreted? (b) What is the average airfare at a passenger seat factor of 85%? (c) What is the short run passenger revenue function? (d) What is the short run total cost function? (e) What is the short run profit function? Graph the short run profit function using Microsoft Excel. (f) What is the short run profit maximising passenger seat factor? (g) What is the short run profit maximising average airfare? (h) What is the maximum value of short run profit?

7.3 Break-even Passenger Seat Factor 7.3.1 Traditional Calculation and its Problems There are a handful of airlines around the world that publish breakeven seat factor statistics, including Singapore Airlines, LATAM Airlines, Emirates, Copa Airlines (Panama) and Azul Airlines (Brazil).40 For those airlines that do not publish breakeven seat factors in their external reporting, it is likely that they compute the statistic for their own internal planning. The breakeven seat factor reported by the above airlines is derived from the following identity: Profit = Passenger Revenue Per RPK × Passenger Seat Factor × ASK - CASK × ASK

(7.14)

where RPK = revenue passenger kilometre, ASK = Available Seat Kilometre and CASK = cost per available seat kilometre. Setting (7.14) equal to zero and re-arranging for the Passenger Seat Factor yields the formula for the traditional breakeven passenger seat factor calculation used and reported by airlines: PSF୆୉ =

େ୅ୗ୏

(7.15)

ୖୣ୴ୣ୬୳ୣ ୔ୣ୰ ୖ୔୏

Equation (7.15) says that to compute the breakeven passenger seat factor the airline simply divides the cost per available seat kilometre by the passenger revenue per revenue passenger kilometre. Figure 7-9 below describes the breakeven seat factor at (7.15) in a graphical way. The breakeven seat factor (PSFBE) in Figure 7-9 is determined at the point of intersection of the cost line and the revenue line. This approach implicitly assumes that costs are invariant to the seat factor (the cost line is horizontal) and revenue increases in a straight line with the seat factor. These two assumptions are incorrect and provide misleading inferences. To see the potential misleading inferences, consider the advice the breakeven equation provides in response to an increase in CASK. Specifically, if CASK goes up, equation (7.15) tells us that the passenger seat factor should be increased because the numerator goes up, but the denominator remains unchanged. In the case of Figure 7-9, an increase in CASK from the firm horizontal line to the dashed horizontal line leads to a change in the point of intersection of the revenue line and the horizontal cost line, which in turn increases the breakeven passenger seat factor from PSFBE to PSFBE*. Since a higher seat factor coincides with a lower yield, then the advice that (7.15) and Figure 7-9 provides is that in response to a higher CASK the airline should lower its yield, which is inconsistent with basic economic principles.

7.3.2 Singapore Airlines We can see the problem of higher unit costs resulting in a higher breakeven seat factor at play by considering the movements in the externally published Singapore Airlines breakeven seat factor – refer to Figure 7-10 below.41 The quarterly movement in the Singapore Airlines breakeven seat factor over the 12 years to March 2016 shares a +84% correlation with the price of jet fuel lagged one year (or four quarters, which is why the t-4 reference is used in the Figure 40

In the case of Emirates for example, you will see the breakeven seat factor calculated in its 2018-19 annual report on page 89 (Emirates Airlines Annual Report 2019). 41 Singapore Airlines stopped publishing its breakeven seat factor in its quarterly News Release bulletin in the September quarter of 2018.

Chapter 7

188

Revenue (R), Cost (C)

R =Yield × SF × ASK

C* = c* x ASK

C = c x ASK

0

PSFBE*

PSFBE

Passenger Seat Factor

1

Fig. 7-9: Interpretation of the Traditional Breakeven Seat Factor Calculation SQ Passenger Breakeven Seat Factor (%) 95 90 85 80 75 70 65 60

BE SF

Price of Singapore Jet Kerosene (SJK) (US$/bbl) 200

SJK (t-4)

150 100 50 0 Mar-16

Sep-15

Mar-15

Sep-14

Mar-14

Sep-13

Mar-13

Sep-12

Mar-12

Sep-11

Mar-11

Sep-10

Mar-10

Sep-09

Mar-09

Sep-08

Mar-08

Sep-07

Mar-07

Sep-06

Mar-06

Sep-05

Mar-05

Sep-04

Mar-04

Source: Airline Intelligence and Research Database 2021

Fig. 7-10: Breakeven Seat Factor Reported by Singapore Airlines 7-10 legend for Singapore Jet Kerosene or SJK). The higher price of jet kerosene has led Singapore Airlines to increase its estimate of the breakeven seat factor because it uses equation (7.15) to determine this statistic, which means that when costs go up the breakeven seat factor estimate must also go up. This is the case even though the response to higher unit costs that maximises profit for the airline is to increase fares, which reduces the passenger seat factor. The information that is provided by the breakeven seat factor is therefore inconsistent with basic economic principles. The main intuitive problem with the calculation at (7.15) is that by using CASK, the airline is using information about a long run cost metric to make an assessment about a performance metric that is best used for short term planning. Rather than using CASK, the airline needs to use the part of CASK that relates to short run variation in cost, which is a topic to which we turn in section 7.3.3 below.

7.3.3 A Superior Breakeven Formula The first step in modelling the breakeven seat factor as a short-term performance indicator is to construct a more appropriate cost representation, including a component of cost that varies with the seat factor as presented at equation (7.1) and repeated for your convenience below: CSR = cPAX × PSF × ܵҧ+ KSR

(7.16)

Under the assumption that the average airfare does not depend on the passenger seat factor, the short run profit function is:

Maximising Short Run Airline Profit

189

SSR = Average Airfare u PSF u Sത – cPAX u PSF u Sത - KSR

(7.17)

The breakeven passenger seat factor is found by setting (7.17) equal to 0 and rearranging for the passenger seat factor as follows: ୊୧୶ୣୢ େ୭ୱ୲

PSF୆୉ = (୅୴ୣ୰ୟ୥ୣ

(7.18)

୅୧୰୤ୟ୰ୣିୡౌఽ౔ )×ୗത

The intuition behind (7.18) is represented in Figure 7-11 below, where the breakeven point occurs when a positively sloped cost line (C) that starts above the origin cuts a positively sloped revenue line (R) that starts at the origin. Revenue (R), Cost (C)

R =Fare × PSF × Seats

C = cPAX x PSF u Seats + KSR

KSR

0

1

PSFBE

Seat Factor

Fig. 7-11: Short Run Breakeven Seat Factor To compute the passenger seat factor (7.18), the airline must split its costs into those that vary with the seat factor in the short run and those that do not. The breakeven seat factor (7.18) is a far superior measure than (7.15) because it explicitly takes into consideration the fact that variations in the seat factor will affect a component of total cost not all total cost.

7.3.4 Multiple Breakeven Seat Factors The breakeven seat factor at (7.18) still has the problem that higher costs cause a higher breakeven seat factor. It has this problem because it does not take into consideration the inverse relationship between the average airfare and the passenger seat factor. If we take this relationship into consideration, the revenue lines in Figures 7-9 and 7-11 are no longer straight lines but curves – this can be seen in Figure 7-12 below. Revenue (R), Cost (C) CSR = cPAXuPSFuSeats + KSR

R =Fare × PSF × Seats

0

PSFBE1

PSFBE2

Fig. 7-12: Short Run Breakeven Seat Factor with Bend in the Revenue Curve

1

Seat Factor

Chapter 7

190

You can see in Figure 7-12 that if an airline believes that a higher seat factor will coincide with a lower average airfare, then it must also believe that there are likely to be two breakeven seat factors (PSFBE1 and PSFBE2). The logic behind this finding is as follows. A relatively low passenger seat factor PSFBE1 will allow the airline to breakeven despite it resulting in relatively low revenue, other things being equal, because it will be offset by a relatively high yield (which generates relatively high revenue) and relatively low variable costs. A relatively high seat factor PSFBE2 will allow the airline to breakeven even though it results in relatively high sales volumes, other things being equal, because it will be offset by a relatively low yield (which generates relatively low revenue) and relatively high variable costs. The higher are the fixed costs of the airline relative to the total costs (that is, the higher the cost line sits in the graph), the closer together will be the breakeven points. You will also notice that if the cost line is shifted upward to the dashed line in Figure 7-12, the breakeven seat factor increases for the lower breakeven seat factor and decreases for the higher breakeven seat factor. The intuition here is that for the higher breakeven seat factor, an increase in cost must be met with a decrease in the seat factor because it is this response that will increase revenue (the airline is in the inelastic zone of the revenue curve, which means that higher revenue can only be achieved by raising fares and reducing the seat factor). In the case of the lower breakeven seat factor, an increase in cost must be met by an increase in the seat factor because the airline is in the elastic zone of the revenue curve. To raise revenue in this case the airline must reduce fares and increase the seat factor.

7.3.5 Breakeven Seat Factor when the Average Airfare is a Cobb-Douglas Function When the average airfare is a Cobb-Douglas function of the passenger seat factor, we know that the revenue function increases at a decreasing rate, but it does not bend over. In other words, revenue never falls when the seat factor goes up. As indicated in Figure 7-13 below, when the revenue curve does not actually bend over it is less likely that the total cost function will meet the revenue function in two places. This will depend on the slope of the cost line, or the passenger varying cost per passenger. The greater is the slope of the cost line, as indicated by the dashed line in Figure 7-13, the more likely it is that the cost line meets the revenue curve at two points. Revenue(R), Cost(C)

R =A u SF1+H

CSR = KSR + c x SF x ASK

0

PSFBE1 PSFBE2

PSFBE3 1

Passenger Seat Factor

Fig. 7-13: Short Run Breakeven Seat Factor with Cobb-Douglas Average Airfare Function

7.3.6 Illustration – All Nippon Airways An average airfare equation was estimated for All Nippon Airways domestic services between 2003 and 2017 on a calendar annual basis. All Nippon Airways is one of two large full-service carriers that operate out of Japan – the other full-service carrier being Japan Airlines. All Nippon Airways has headquarters in the Shi dome City Centre in the Minato ward of Tokyo. It operates domestic and internationals services to 97 destinations (ANA Destinations 2021). The average domestic airfare equation for All Nippon Airways (ANA) was estimated by an airline analyst, with a summary of the estimated equation given as follows: Average Airfare = 21,758 – 8,097.532 u PSF Under the assumption that the airline flies a 180 seat A320-200 aircraft on a representative domestic route, the passenger revenue function for the airline as a function of the PSF is the airfare function given above multiplied by the number of seats multiplied by the PSF as follows: Passenger Revenue = 3,916,440 u PSF – 1,457,556 u PSF2 The passenger varying costs of the airline per passenger is 2,222 Japanese Yen, while the fixed cost for the flight is 1.5m Japanese Yen. This implies the following short run cost function for ANA for the route in question:

Maximising Short Run Airline Profit

191

CSR = 2,222 u 180 u PSF + 1,500,000 = 399,960 u PSF + 1,500,000 The short run profit of the airline is short run passenger revenue less short run cost, which is: Short Run Profit = 3,916,440 u PSF – 1,457,556 u PSF2 - 399,960 u PSF - 1,500,000 If we graph the passenger revenue function (firm curve) and the cost function (dashed line) together in Microsoft Excel, we obtain Figure 7-14 below. ANA DOM. Revenue/Cost (JPY) $3,000,000 Revenue

$2,500,000

Cost

$2,000,000 $1,500,000 $1,000,000 $500,000 $0 99% 96% 93% 90% 87% 84% 81% 78% 75% 72% 69% 66% 63% 60% 57% 54% 51% 48% 45% 42% 39% 36% 33% 30% 27% 24% 21% 18% 15% 12% 9% 6% 3% 0%

PSF Fig. 7-14: Short Run Profit ANA Domestic Flight with 180 Seats We can see in Figure 7-14 that there is a single breakeven point that is less than a passenger seat factor of 100%. The breakeven point occurs at a passenger seat factor of around 55%. Let us now show how we obtain the breakeven passenger seat factor. If we set the short run profit function equal to zero, we obtain: 0 = 3,916,440 u PSF – 1,457,556 u PSF2 - 399,960 u PSF - 1,500,000 If we simplify this expression this yields the following quadratic equation: – 1,457,556 u PSF2 + 3,516,480 u PSF – 1,500,000 = 0 We wish to find the values of PSF that satisfy this quadratic equation. If it makes it easier, we can also write this in terms of X as follows: – 1,457,556 u X2 + 3,516,480 u X – 1,500,000 = 0 To solve this quadratic equation, we use the quadratic formula: X=

ି௕±ඥ௕మ ିସ௔௖ ଶ௔

=

ିଷ,ହଵ଺,ସ଼଴±ඥଷ,ହଵ଺,ସ଼଴మ ିସ(ିଵ,ସହ଻,ହହ଺)(ିଵ,ହ଴଴,଴଴଴) ଶ(ିଵ,ସହ଻,ହହ଺)

where the a-term in the above formula is the coefficient attached to X2, the b-term in the above formula is the coefficient attached to X and the c-term is the term that is independent of X. The solution to the quadratic formula is X = 0.5536 or X = 1.5956 which correspond to the breakeven seat factors 55.36% and 159.56%, with the first breakeven seat factor also identified in Figure 7-14. Let us now show you how to obtain these results using Microsoft Excel. First, open-up a blank worksheet in Microsoft Excel. In cells A1, A2 and A3 insert the text a, b and c. In cell B1 type in the value -1,457,556, in cell B2 type in the value 3,516,480 and in cell B3 type in the value -1,500,000. Next, in cell A5 type in the text “PSF BE1” and in cell A6 type in the text “PSF BE2”. In cells B5 and B6 insert the following formulae into Microsoft Excel: In cell B5 Ÿ “=(-1*B2+(((B2^2)-4*(B1)*(B3))^0.5))/(2*B1)” In cell B6 Ÿ “=(-1*B2-(((B2^2)-4*(B1)*(B3))^0.5))/(2*B1)”

192

Chapter 7

The entry in cell B5 will generate the first breakeven seat factor of 55.36% and the entry in cell B6 will generate the second breakeven seat factor of 159.56%. There is another way of finding the solution to our quadratic equation in Microsoft Excel using the Goal Seek function. To use the Goal Seek function, continue working in the same worksheet as our quadratic formula worksheet from above. In cell A8 type in the text “PSF”, in cell B8 type in the text “Quadratic”, and in cell B9 type in the formula for the quadratic which is “=B1*(A9^2)+B2*A9+B3”. In cell A9 type in the number 0.5. Enter your cursor into a blank part of the worksheet, such as cell A11. Then click on “Data” in the main menu, and then “What-if Analysis”. Choose the option “Goal Seek..”. In the Goal Seek dialogue box in the “Set cell:” box choose the cell $B$9, in the “To value” box type in the number 0, and in the “By changing cell:” box type in the cell A9. Then press OK. Microsoft Excel will then find the value in cell A9 that generates a value of 0 in cell B9. The number you should get in cell A9 is the breakeven solution 0.5536. If we now alter the initial value in cell A9 to 2, then the solution we obtain when we perform the Goal Seek in Microsoft Excel is 1.5956. The results tell us that any passenger seat factor above 56% results in short run profit for ANA because revenue increases by more than cost beyond this point. To find the breakeven average airfare we substitute the breakeven passenger seat factor into the average airfare expression above to obtain: Average Airfaret = 21,758 – 8,097.532 u 0.5536 = 17,275 Yen This is the average airfare which generates a profit of zero for ANA in the domestic market.

Quiz 7-3. Breakeven Seat Factors 1. Air New Zealand flies from Auckland to Queenstown with revenue per RPK of 15.5 New Zealand cents and cost per available seat kilometre of 12.5 New Zealand cents. Find the breakeven seat factor for the Air New Zealand flight using the traditional method. What happens to the breakeven seat factor estimate if the cost per available seat kilometre increases to 13 New Zealand cents? What is the problem with this answer? 2. Which of the following represents a critical problem with the traditional method of finding the breakeven seat factor? (a) A concave down revenue parabola is assumed with the traditional method. (b) The traditional method assumes all costs are invariant to the seat factor. (c) The traditional method assumes all costs are variable to the seat factor. (d) The traditional method assumes the revenue function comes from a Cobb-Douglas average airfare function. 3. Return to the Rex Express Airlines example where the estimated average airfare function for a representative flight is Average Airfaret = 258.4281 – 379.369 u Passenger Seat Factort + 1625.749 u GDP Growtht + 0.7926 u Cost Per Passengert. Assume in this average airfare function that economic growth is -1.5% (input into the function GDP Growth = -0.015) and the cost per passenger is $190. Also assume that the private operating cost of the airline is described by the function C = 1,020 u Passenger Seat Factor + $3,500. (a) Use this information to find the preferred short run breakeven passenger seat factor(s). (b) Now assume that the passenger varying cost per passenger seat factor increases from 1,020 to 1,100. How does this affect your estimate of the breakeven passenger seat factor(s)?

CHAPTER 8 MAXIMISING MEDIUM RUN AIRLINE PROFIT

When an airline decides to increase capacity, it does so under the belief that this will increase profitability. We know from chapter 6 that when an airline increases capacity its costs must increase, because there are many major inputs into production, such as fuel and labour, of which the airline must consume more if it is to increase capacity. We know from chapter 5, however, that when an airline increases capacity, this may lead to an increase or decrease in revenue depending on the extent to which airline yields fall in response to the increase in capacity. For an increase in capacity to result in an increase in profit, the airline therefore needs revenue to increase, and revenue to increase by more than costs increase. If revenue were to decrease in response to an increase in capacity, then profit must fall because revenue decreases and cost increases, which must reduce profit. The aim of this chapter is to understand the relationship between capacity and profit by understanding the extent to which revenue and costs change in response to an increase in capacity. We do this by analysing profit as a function of capacity, which we call the medium run profit function. We will show that the medium run profit function, like the medium run revenue function in chapter 5, increases at a decreasing rate as capacity increases from low levels, eventually reaches a maximum at the profit maximising level of capacity, and then begins to fall as capacity is increased beyond optimal levels. Airlines wish to set capacity at levels which maximise profit. We will show in this chapter that this point occurs when a small increase in capacity leads to an increase in revenue that is just equal to the increase in cost. Economists call this the marginal revenue equals marginal cost condition. To identify this point, it is necessary that the airline identifies and quantifies two critical parameters - the extent to which yields fall as capacity increases, which is formalised by the elasticity of airline yields to capacity or the own yield elasticity, and the proportion of total operating costs that depend on capacity. The marginal revenue equals marginal cost condition is more complicated in the case where airlines face competition, which is true along most routes. When competitors are present, an increase in an airline’s capacity draws a reaction from competitor airlines, which in turn affects the yield and revenue of the airline in question, called the own airline. The marginal revenue of the own airline in this case depends not only on its own capacity but also on the capacity of its competitors. This in turn means that the profit maximising level of capacity depends not only on the own yield elasticity but two further elasticities - the elasticity which measures how competitor capacity reacts to the capacity of the own airline, and the elasticity of the own airline’s yield to a change in the capacity of competitors, called the cross-yield elasticity. One of the reasons why airlines all over the world find it difficult to consistently make money is because they find it difficult to understand and to quantify the bends in the relationship between profit and capacity. It is hoped that after reading this chapter airline analysts and strategists will have a better understanding of the bends in the profit function, and the critical parameters that are required to determine the profit maximising level of capacity.

8.1 Medium Run Airline Profit Maximisation – Linear PRASK Function In this section we combine the medium run airline revenue analysis of chapter 5 and the airline cost analysis of chapter 6 to determine the medium run profit function of the airline. Remember that the medium run is a length of time that is long enough for the airline to strategically alter its level of capacity. Capacity can be altered by changing the frequency of services, the size of the aircraft, or the average distance over which the aircraft operates. Because capacity is allowed to vary over the medium run, the medium run profit function expresses profit as a function of capacity. The analysis of the sections below demonstrates how to derive the medium run profit function and to find the profit maximising level of capacity when PRASK is a linear function of capacity.

8.1.1 Theory and Analytics In chapter 5, section 5.1 we developed the medium run passenger revenue function for the airline as a function of capacity when PRASK is a linear function of ASKs. This is described by equation (5.7) and repeated below for your convenience: RMR = PRASK u ASK = a0 u ASK + a1 u ASK2

(8.1)

In chapter 6, section 6.10.2 at equation (6.28) we developed the cost function for an airline in the medium run. This medium run cost function is repeated below for your convenience:

Chapter 8

194

CMR = cuASK + K

(8.2)

The medium run airline profit function in the case of the linear PRASK function is the revenue function (8.1) less the cost function (8.2), which gives:42 SMR = [a0 + a1uASK]uASK - cuASK – K

(8.3)

If we expand the term in square brackets on the right-hand side of (8.3) we see that the medium run airline profit function is a concave down parabola in ASKs: SMR = a0uASK + a1uASK2 - cuASK – K = a1uASK2 + (a0 – c)uASK - K

(8.4)

We know that the profit function parabola is concave down because the coefficient attached to ASK2 in (8.4), a1, is less than zero, and this is the case because we know that an increase in ASKs leads to lower PRASK in the linear PRASK function. Let us now graph the medium run profit function at (8.4) so that we have an idea about what it may look like. This is done schematically in Figure 8.1 below. Profit A

B

C

0

ASK* =

௖ି௔బ ଶ௔భ

ASK Profit Function

-K

Fig. 8-1: Medium Run Profit Function In Figure 8-1 we see that the profit function in ASKs starts on the vertical axis below the origin at the level of fixed cost K. It then increases as ASKs increase, moving through point C, up until the maximum turning point at A, before falling thereafter through point B. Point C to the left of the optimal point occurs where the level of capacity is too low, and yields are too high. The airline can make more money by expanding capacity through this point because this leads to an increase in revenue that exceeds the increase in costs despite yields falling. At a point such as B, which is to the right of the optimum, the airline is in a position of oversupplying ASKs. In this case yields are too low. The airline has pushed yields to the point at which increasing capacity requires yields to fall too far, resulting in additional revenue that is exceeded by the additional cost. To find the level of ASKs that maximise medium run profit we need to use calculus. This involves differentiating the medium run profit function (8.4) with respect to ASKs and setting the result equal to zero, which yields: ୢ஠౉౎ ୢ୅ୗ୏

= 2aଵ ASK + a଴ െ c = 0

(8.5)

The condition at (8.5) is a marginal ASK revenue equals marginal ASK cost condition, which we simplify to marginal revenue (MR) equals marginal cost (MC). MR is the change in revenue in response to a small increase in ASKs. MC is the change in cost in response to a small increase in ASKs. The MR component of (8.5) is 2a1ASK + a0, and the MC component is c.

42

In the analysis to follow in this chapter we assume that passenger revenue is equal to total revenue. There is no treatment of freight revenue and other revenue. For many airlines, such as low-cost carriers, this is a reasonable assumption. Alternatively, we can think of freight and other revenue as ‘fixed revenue’ which we can deduct from fixed costs to obtain net fixed costs. The net fixed cost can be represented by the K term in equation (8.3).

Maximising Medium Run Airline Profit

195

The fact that MC is positive (c > 0) means that MR must also be positive if condition (8.5) is to hold. This places a restriction on the possible values that ASK can take on. To see this, we start by specifying the condition required for MR to be positive, which is: MR = 2a1ASK + a0 > 0

(8.6)

Equation (8.6) implies the following restriction on ASKs: ASK
~a2~, whereas in the Cournot model a1 = a2. Similarly, the way that airline 2’s passenger supply affects airline 2’s price is stronger than the way that airline 1’s passenger supply affects airline’s two’s price. The model generates these different own and cross-price effects because the two airlines are assumed to produce differentiated goods. While this is more realistic, it does come at a significant cost, which is that the analysis is significantly more complicated as we will show shortly. Substituting the linear functions (9.41) and (9.42) into the first order conditions (9.39) and (9.40) yields the first order conditions under the assumption of the linear airfare functions: ୢ஠భ ୢ୕భ

= a଴ + 2aଵ Qଵ + aଶ Qଶ െ cଵ = 0

(9.43)

Chapter 9

234 ୢ஠మ ୢ୕మ

= b଴ + bଵ Qଵ + 2bଶ Qଶ െ cଶ = 0

(9.44)

To find the levels of Q1 and Q2 which maximise profit for both airlines, we need to solve the two equations (9.43) and (9.44) simultaneously in the two unknowns Q1 and Q2. The easiest way to solve equations (9.43) and (9.44) in my opinion is to set up this two-equation system in matrix form. The first step in this process is to shift all terms that are independent of the number of passengers to the right-hand side of the equals sign as follows: 2a1Q1 + a2Q2 = c1 – a0 b1Q1 + 2b2Q2 = c2 – b0

(9.45) (9.46)

The matrix version of the system of equations at (9.45) and (9.46) can then be written as: cଵ െ a଴ a ଶ Qଵ ൨ ൤ ൨ = ቂc െ b ቃ 2bଶ Qଶ ଶ ଴

2a ൤ ଵ bଵ

(9.47)

2ܽ ܽଶ Let us now explain the contents of the matrix system (9.47). The matrix ൤ ଵ ൨ is a (2u2) square matrix consisting ܾଵ 2ܾଶ of the coefficients attached to the output variables on the left-hand side of equations (9.45) and (9.46). The (2u1) vector ܿଵ െ ܽ଴ ܳ ൤ ଵ ൨ is a vector of the passenger outputs of airlines 1 and 2. The (2u1) vector ቂܿ െ ܾ ቃ is a vector of the variables that ܳଶ ଶ ଴ are independent of the output variables on the right-hand side of (9.45) and (9.46).49 We can simplify the matrix notation at (9.47) even further by writing it in the following short-hand way: AQ = b

(9.48)

where the letters in bold in (9.48) indicate that the relevant variable represents a matrix. The A matrix is the (2u2) matrix ܽଶ 2ܽ ܳ ൨, the variable Q stands for the (2u1) vector of passenger supply variables ൤ ଵ ൨ and the (2u1) of coefficients ൤ ଵ ܳଶ ܾଵ 2ܾଶ ܿଵ െ ܽ଴ vector b stands for the parameters that are independent of output, ቂܿ െ ܾ ቃ. To solve (9.48) we need to pre-multiply the ଶ ଴ right and left-hand side of the matrix system (9.48) by the inverse of matrix A, which is A-1. If we do this, we obtain the solution in matrix form: Q = A-1b

(9.49)

In my opinion the easiest way to find the solution (9.49) is to use a neat little matrix rule called Cramer’s Rule.50 The solution using Cramer’s Rule involves finding the determinants of the various matrices contained in (9.48) in the following way: Q‫כ‬ଵ =

Q‫כ‬ଶ =

ୡభ ିୟబ ୟమ ቚୡ ିୠ ቚ మ బ ଶୠమ ଶୟభ ୟమ ฬ ฬ ୠభ ଶୠమ

=

ଶ(ୡభ ିୟబ )ୠమ ି(ୡమ ିୠబ )ୟమ

ଶୟభ ୡభ ିୟబ ฬ ୠభ ୡమ ିୠబ ଶୟభ ୟమ ฬ ฬ ୠభ ଶୠమ

=

ଶ(ୡమ ିୠబ )ୟభ ି(ୡభ ିୟబ )ୠభ

ฬ

ସୟభ ୠమ ିୟమ ୠభ

ସୟభ ୠమ ିୟమ ୠభ

(9.50)

(9.51)

The numerator and denominator terms in (9.50) and (9.51) are referred to as determinants in matrix algebra. A determinant is a particular numerical value that is representative of a square matrix. The determinant in the denominator is the same in both equations. It is the determinant of the (2x2) matrix consisting of the coefficients attached to the output variables. It is found by using the following formula: 2a ฬ ଵ bଵ

aଶ ฬ = 2a1u2b2 – a2ub1 2bଶ

(9.52)

The denominator determinant is found by multiplying the entry in row 1 column 1 of the matrix, 2a1, by the entry in row 2 column 2, 2b2, and subtracting from this the entry in row 1 column 2, a2, times the entry in row 2 column 1, b1. Another way of saying this is that we multiply the main diagonal entries, which are the entries along the diagonal that

49 If you need help on matrices see Chiang and Wainwright 1985, 48-81. Matrices are covered in lots of other good mathematical economics books. 50 Chiang and Wainwright 1985, 103-107 gives an excellent explanation of Cramer’s Rule.

Monopoly and Oligopoly Airline Competition

235

runs from left to right as indicated by the firm ellipse and subtract from this the product of the off-main diagonal entries, which are indicated by the dashed ellipse. The numerator determinants in the solutions (9.50) and (9.51) are different. The numerator determinant for the solution to Q1 is formed by replacing the first column of the matrix in the denominator of (9.50) with the (2 x 1) vector of the independent variables as follows: cଵ െ a଴ ቚc െ b ଶ ଴

aଶ 2bଶ ቚ = (c1 – a0)u2b2 – a2(c2 – b0)

(9.53)

You will see that the first column of (9.53) as indicated by the circle is the (2u1) vector of the independent variables, ܿଵ െ ܽ଴ ቂܿ െ ܾ ቃ. To find the determinant we simply use the same rule as applied to (9.52) which is to multiply the mainଶ ଴ diagonal entries and subtract from this the product of the off-main diagonal entries. The numerator determinant for the solution to Q2 is formed by replacing the second column of the matrix in the denominator of (9.51) with the (2 x 1) vector of independent variables as follows: 2a ฬ ଵ bଵ

cଵ െ a଴ ฬ = 2a1(c2 – b0) – b1(c1 – a0) cଶ െ b଴

(9.54)

We can see that the output of the two airlines is significantly more complex than in the model in which the air travel services of the two airlines are identical. The output of each airline in this case depends on the marginal passenger cost of the two airlines, the sensitivity of an airline’s average airfare to the airlines own output, and the sensitivity of the airline’s average airfare to the competitor’s output. The market level of passengers is determined by adding together the output of airline’s 1 and 2 at (9.50) and (9.51) as follows: Q* = Q‫כ‬ଵ + Q‫כ‬ଶ =

ଶ(ୡభ ିୟబ )ୠమ ି(ୡమ ିୠబ )ୟమ ାଶ(ୡమ ିୠబ )ୟభ ି(ୡభ ିୟబ )ୠభ ସୟభ ୠమ ିୟమ ୠభ

=

(ୡభ ିୟబ )(ଶୠమ ିୠభ )ା(ୡమ ିୠబ )(ଶୟభ ିୟమ ) ସୟభ ୠమ ିୟమ ୠభ

(9.55)

The market level of passengers in this case depends on airline costs and the way that changes in passenger output affect the average airfares. If we substitute both output solutions (9.50) and (9.51) into the average airfare functions (9.41) and (9.42) we obtain the solutions for the average airfares of airline’s 1 and 2, which are: Pଵ‫ = כ‬a଴ + aଵ ቂ

ଶ(ୡభ ିୟబ )ୠమ ି(ୡమ ିୠబ )ୟమ

Pଶ‫ = כ‬b଴ + bଵ ቂ

ସୟభ ୠమ ିୟమ ୠభ

ቃ + aଶ ቂ

ଶ(ୡభ ିୟబ )ୠమ ି(ୡమ ିୠబ )ୟమ ସୟభ ୠమ ିୟమ ୠభ

ଶ(ୡమ ିୠబ )ୟభ ି(ୡభ ିୟబ )ୠభ ସୟభ ୠమ ିୟమ ୠభ

ቃ + bଶ ቂ

ቃ

ଶ(ୡమ ିୠబ )ୟభ ି(ୡభ ିୟబ )ୠభ ସୟభ ୠమ ିୟమ ୠభ

ቃ

These average airfare solutions simplify to: Pଵ‫ = כ‬a଴ + ቂ

(ୡభ ିୟబ )(ଶୟభ ୠమ ିୟమ ୠభ )ା(ୡమ ିୠబ )ୟభ ୟమ

Pଶ‫ = כ‬b଴ + ቂ

ସୟభ ୠమ ିୟమ ୠభ

ቃ

(ୡభ ିୟబ )ୠమ ୠభ ା(ୡమ ିୠబ )(ଶୟభ ୠమ ିୠభ ୟమ ) ସୟభ ୠమ ିୟమ ୠభ

ቃ

(9.56) (9.57)

These are obviously complicated solutions - and this is just the case of two airlines. The main point is for you to understand just how complex it is when you add the assumption that two airlines offer to the market differentiated products. The solutions for price depend on the strength of the impact of one airline’s output on the demand and price of the competing airline, which are the a and b parameters along with the marginal cost of the two airlines. These parameters interact in a complex way. One of the ways that we like to use the pricing formula (9.56) and (9.57) is to investigate the impact of an increase in both airline costs on the average airfares that are charged by the airlines. Let us assume that there is an increase in the cost of both airlines by dc1 = dc2 = dc. This could come about, for example, because of an increase in fuel costs which raises the costs of all airlines along a route. The change in the average airfares at (9.56) and (9.57) are found by differentiating the average airfare solution with respect to both cost terms as follows: ୢ୔‫כ‬భ ୢୡ ୢ୔‫כ‬మ ୢୡ

=

ଶୟభ ୠమ ିୟమ ୠభ ାୟభ ୟమ

=

ଶୟభ ୠమ ିୠభ ୟమ ାୠమ ୠభ

ସୟభ ୠమ ିୟమ ୠభ

ସୟభ ୠమ ିୟమ ୠభ

(9.58) (9.59)

We can see that both (9.58) and (9.59) are relatively complicated expressions. By making a simplifying assumption however, we can see how both expressions are likely to be less than 1 meaning that the airlines will pass on some of the

Chapter 9

236

higher costs into higher airfares but not all the costs. If we assume that a2 = b2 and a1 = b1 which means that an increase in Q1 affects both P1 and P2 with the same downward force and an increase in Q2 affects P1 and P2 with the same downward force, then (9.58) and (9.59) become: ୢ୔‫כ‬భ ୢୡ ୢ୔‫כ‬మ ୢୡ

ଷୟ ୠ ିୟ ୠ

(9.60)

= ସୟభ ୠమ ିୟమ ୠభ భ మ

మ భ

ଷୟ ୠ ିୠ ୟ

(9.61)

= ସୟభ ୠమ ିୟ భୠమ భ మ

మ భ

We can see that the price response in both cases is identical and equal to some positive value between 0 and 1 because both expressions must be positive, and the numerator is smaller than the denominator.

9.5.2 Numerical Example It may be easier for you to understand our differentiated Cournot model if we replace the a’s and the b’s with numbers. In the AKL-CHC route example of section 9.4.2 we assumed that competition on the route was a homogeneous duopoly. We now assume that competition between NZ and JQ on the route is a differentiated duopoly, which is closer to the reality of competition on this route. The pricing functions for the two airlines on the route are: PJQ = 300 – QJQ - 0.5QNZ PNZ = 500 – 0.5QJQ – QNZ

(9.62) (9.63)

These equations are equivalent to (9.41) and (9.42) in section 9.5.1 above. Assuming the cost functions are the same as in our section 9.4.2 numerical illustration, then the profit functions of our two airlines are found by multiplying QJQ by (9.62) and subtracting JQ’s cost function and multiplying (9.63) by QNZ and subtracting NZ’s cost function as follows: Ɏ୎୕ = ൣ300 െ Q୎୕ െ 0.5Q୒୞ ൧ × Q ୎୕ െ 50Q୎୕ െ 150,000 Ɏ୒୞ = ൣ500 െ 0.5Q୎୕ െ Q୒୞ ൧ × Q ୒୞ െ 60Q୒୞ െ 150,000

(9.64) (9.65)

The first order conditions for a maximum are found by differentiating (9.64) with respect to QJQ and (9.65) with respect to QNZ and setting the respective equations equal to zero, which gives: ୢ஠ె్ ୢ୕ె్ ୢ஠ొౖ ୢ୕ొ్

= 300 െ Q୎୕ െ 0.5Q୒୞ െ Q୎୕ െ 50 = 0

(9.66)

= 500 െ 0.5Q୎୕ െ Q୒୞ െ Q୒୞ െ 60 = 0

(9.67)

These two conditions are equivalent to (9.43) and (9.44) in section 9.5.1. Let us now set the system (9.66) and (9.67) up using matrix algebra: ቂ

െ2 െ0.5 Q୎୕ 50 െ 300 ൨=ቂ ቃ൤ ቃ െ0.5 െ2 Q୒୞ 60 െ 500

(9.68)

Solving (9.68) using Cramer’s Rule gives: Q‫= ୕୎כ‬ Q‫୒כ‬୞ =

ିଶହ଴ ିସସ଴ ିଶ ቚ ି଴.ହ ିଶ ቚ ି଴.ହ ିଶ ቚ ି଴.ହ

ቚ

ି଴.ହ ቚ ହ଴଴ିଶଶ଴ ଶ଼଴ ିଶ ି଴.ହ = ସି଴.ଶହ = ଷ.଻ହ = 74.7 ቚ ିଶ ିଶହ଴ ቚ ଼଼଴ିଵଶହ ିସସ଴ = 201.33 ି଴.ହ = ଷ.଻ହ ቚ ିଶ

(‘000)

(9.69)

(‘000)

(9.70)

Substituting the solutions for output (9.69) and (9.70) into the pricing functions (9.62) and (9.63) yields the market clearing average airfares of Jetstar and Air New Zealand on AKL-CHC: PJQ = 300 – 74.7 - 0.5(201.33) = 124.64 PNZ = 500 – 0.5(74.7) – 201.33 = 261.32 In this equilibrium you will notice that the full-service airline, Air New Zealand sets a much higher average airfare, NZ$261.32, than the low-cost carrier Jetstar, NZ$124.64, which is to be expected. Let us suppose now that the marginal cost of both airlines increases by NZ$10. In this case the output solutions become:

Monopoly and Oligopoly Airline Competition

Q‫= ୕୎כ‬ Q‫୒כ‬୞ =

ିଶସ଴ ିସଷ଴ ିଶ ቚ ି଴.ହ ିଶ ቚ ି଴.ହ ିଶ ቚ ି଴.ହ

ቚ

ି଴.ହ ቚ ସ଼଴ିଶଵହ ଶ଺ହ ିଶ ି଴.ହ = ସି଴.ଶହ = ଷ.଻ହ = 70.7 ቚ ିଶ ିଶସ଴ ቚ ଼଺଴ିଵଶ଴ ଻ସ଴ ିସଷ଴ ି଴.ହ = ସି଴.ଶହ = ଷ.଻ହ = 197.3 ቚ ିଶ

237

(9.71) (9.72)

Substituting (9.71) and (9.72) into (9.62) and (9.63) yields the equilibrium airfares with the higher marginal cost: PJQ = 300 – 70.7 - 0.5(197.3) = 130.65 PNZ = 500 – 0.5(70.7) – 197.3 = 267.35 In this case Jetstar’s and Air New Zealand’s average airfares both increase by NZ$6. This is 60% of the cost increase indicating that both airlines in equilibrium will only pass part of the increase in costs through into higher airfares. They will not pass all the cost increases through into higher airfares because they will be concerned about the reaction of demand to the higher airfares.

9.6 n-Airline Models 9.6.1 Theory and Analytics When there are 3 airlines that compete along a route this is called a triopoly and when there are four airlines the market structure is called a quadropoly. There can obviously be even more airlines. Examples of routes that are triopolies, include SAS, Norwegian and KLM on Amsterdam to Copenhagen, and GOL, LATAM and Azul on Sao Paulo to Rio de Janeiro. An example of a pentopoly is Qantas, Jetstar, LATAM, Air New Zealand and Virgin Australia on Sydney to Auckland, and an example of a hexopoly is Singapore Airlines, Silk Air, Jetstar, Air Asia, Garuda, and Scoot on Singapore to Bali. In this section we build a model in which there are n airlines competing along a route, where the value for n is 3 or more. We assume the airlines produce an identical good once again. The pricing function for the market in this case is: P = P(Q1 + Q2 + …….. + Qn)

(9.73)

We assume once again that the pricing function in (9.73) is a smooth and continuous function. Equation (9.73) says that the average airfare that all n airlines charge on the route depends on the market passenger numbers. We assume that the variable cost per passenger of our airlines are c1, c2, ….., cn and the fixed costs are K1, K2, ……, Kn. We use (9.73) and the cost information to derive the profit functions for each airline as follows: Ɏଵ = P(Qଵ + Qଶ + ‫ … ڮ‬+ Q୬ )Qଵ െ cଵ Qଵ െ Kଵ Ɏଶ = P(Qଵ + Qଶ + ‫ … ڮ‬+ Q୬ )Qଶ െ cଶ Qଶ െ K ଶ ……………………..……………………..……………………..

(9.74)

Ɏ୬ = P(Qଵ + Qଶ + ‫ … ڮ‬+ Q୬ )Q୬ െ c୬ Q୬ െ K ୬ Each airline chooses output to maximise profit. This involves differentiating the profit functions in (9.74) with respect to the relevant output and setting the result equal to zero. This yields the following set of n first order conditions (one for each airline): dP dɎଵ = P(Qଵ + Qଶ + ‫ … ڮ‬+ Q୬ ) + Q െ cଵ = 0 dQଵ ଵ dQଵ dP dɎଶ = P(Qଵ + Qଶ + ‫ … ڮ‬+ Q୬ ) + Q െ cଶ = 0 dQଶ ଶ dQଶ ……………………..……………………..……………………..

(9.75)

dP dɎ୬ = P(Qଵ + Qଶ + ‫ … ڮ‬+ Q୬ ) + Q െ c୬ = 0 dQ୬ ୬ dQ୬ As we have done with the homogeneous duopoly model, since all firms produce an identical product and charge the same price, we can solve (9.75) by adding the first order conditions as follows:

Chapter 9

238

n × P(Qଵ + Qଶ + ‫ … ڮ‬+ Q୬ ) +

ୢ୔ ୢ୕

(Qଵ + Qଶ + ‫ … ڮ‬+ Q୬ ) െ cଵ െ cଶ െ ‫ … ڮ‬. . െc୬ = 0

(9.76)

If we transfer the cost terms in (9.76) to the right-hand side, and simplify the marginal revenue on the left-hand side, then we obtain: ୢ୔ (୕భ ା୕మ ା‫…ڮ‬ା୕౤ )

P(Qଵ + Qଶ + ‫ … ڮ‬+ Q୬ ) ቂn +

ୢ୕భ ୔(୕భ ା୕మ ା‫…ڮ‬ା୕౤ )

ቃ = cଵ + cଶ + ‫ … ڮ‬. +c୬

(9.77)

If we substitute the market price elasticity of demand into the square parentheses on the left-hand side of (9.77), then we obtain: ଵ

P(Qଵ + Qଶ + ‫ … ڮ‬+ Q୬ ) ቂn + ቃ = cଵ + cଶ + ‫ … ڮ‬. +c୬ க

(9.78)

Forming the average marginal cost for the industry and further simplifying the left-hand side marginal revenue expression yields: P(Qଵ + Qଶ + ‫ … ڮ‬+ Q୬ ) ቂ

୬கାଵ

ୡభ ାୡమ ା‫…ڮ‬.ାୡ౤

க

୬

ቃ=ቀ

ቁn

(9.79)

This condition is a market marginal revenue equals market marginal cost condition. This in turn can be simplified for the equilibrium average airfare charged by the n airlines, which is: P‫= כ‬

୬கୡത

(9.80)

୬கାଵ

The equilibrium average airfare in this case depends not only on average industry marginal costs and the elasticity of demand, but also on the number of competitors. We can show that as the number of competitors increase the average airfare is lower. This is done by differentiating (9.80) with respect to n. The easiest way to do this is to first take the natural logarithm of the average airfare at (9.80): Loge P* = loge n + loge H + loge cത – loge (1 + nɂ)

(9.81)

Differentiating (9.81) with respect to n yields: ௗ௉‫ כ‬ଵ ௗ௡ ௉‫כ‬

ଵ

ఌ

௡

ଵା௡ఌ

= െ

=

ଵା௡ఌି௡ఌ ௡(ଵା௡ఌ)

=

ଵ

(9.82)

௡(ଵା௡ఌ)

We know that (1 + ݊ߝ) must be less than zero for aggregate marginal revenue to be positive and so (9.82) must be less than zero, which in turn means that the market average airfare falls as the number of competitors increases.

9.6.2 Numerical Example An airline pentopoly (n=5) competes on a particular route, such as Sydney to Auckland, where each airline offers an identical service. The average marginal cost of the airlines is $100 and the inverse demand function for the route is: P = 320 – 0.00005 u (Q1 + Q2 + Q3 + Q4 + Q5)

(9.83)

The profit function of each airline on the route is: S1 = P u Q1 – c1 u Q1 – K1 S2 = P u Q2 – c2 u Q2 – K2 S3 = P u Q3 – c3 u Q3 – K3

(9.84)

S4 = P u Q4 – c4 u Q4 – K4 S5 = P u Q5 – c5 u Q5 – K5 The first order condition for profit maximisation for each of the 5 airlines in the pentopoly is found by differentiating each of the profit functions in (9.84) with respect to Qi and setting the result equal to zero as follows:

Monopoly and Oligopoly Airline Competition ୢ஠భ ୢ୕భ ୢ஠మ ୢ୕మ ୢ஠య ୢ୕య ୢ஠ర ୢ୕ర ୢ஠ఱ ୢ୕ఱ

239

= 320 – 0.00005u(Q1 + Q2 + Q3 + Q4 + Q5) – 0.00005 u Q1 – c1 = 0 = 320 – 0.00005u (Q1 + Q2 + Q3 + Q4 + Q5) – 0.00005 u Q2 – c2 = 0 = 320 – 0.00005u (Q1 + Q2 + Q3 + Q4 + Q5) – 0.00005 u Q3 – c3 = 0

(9.85)

= 320 – 0.00005u (Q1 + Q2 + Q3 + Q4 + Q5) – 0.00005 u Q4 – c4 = 0 = 320 – 0.00005u(Q1 + Q2 + Q3 + Q4 + Q5) – 0.00005 u Q5 – c5 = 0

If we sum the conditions at (9.85) we obtain the following ‘market’ first order condition: 5 u 320 – 5 u 0.00005 u Q – 0.00005 u Q – (c1 + c2 + c3 + c4 + c5) = 0

(9.86)

This can be solved for market output Q as follows: Q‫= כ‬

ଵ,଺଴଴ିହ×ୡ ଺×଴.଴଴଴଴ହ

=

ଵ,଺଴଴ିହ×ଵ଴଴ ଺×଴.଴଴଴଴ହ

= 3,666,667

(9.87)

Substituting the solution for market output (9.87) into the pricing function (9.83) yields: P = 320 – 0.00005× 3,666,667 = $137

(9.88)

Let us determine the market price elasticity of demand at the market clearing price and output. The price elasticity of demand is found by finding the inverse of the coefficient on market output in (9.83) and multiplying this by the equilibrium price and dividing by the equilibrium market number of passengers as follows: H* =

ௗொ ௉‫כ‬ ௗ௉ ொ‫כ‬

=

ିଵ ଴.଴଴଴଴ହ

×

ଵଷ଺.଺଻ ଷ,଺଺଺,଺଺଻

= െ0.745

(9.89)

We can use the information at (9.89) to find the equilibrium market price in a different way using the general formula (9.80). We know from (9.80) that the equilibrium condition for the market price is: P* =

ହ ×க‫ כ‬u ୡ

(9.90)

ଵାହக‫כ‬

Substituting the market price elasticity of demand at (9.89) and the average marginal cost ܿ = $100 into (9.90) yields: P* =

ିହ ×଴.଻ସହu ଵ଴଴ ଵିହ(଴.଻ସହ)

= $137

which is equal to the market price determined above at (9.88).

Quiz 9-3 Oligopolistic Airline Competition 1. (a) (b) (c) (d)

A duopoly in an airline market consists of: A single airline. Two airlines. Three airlines. Four airlines.

2. (a) (b) (c) (d)

Which of the following are the key features of a Cournot airline duopoly? There are two airlines that are identical in every respect. There are two airlines that produce an identical air transport service. There are two airlines that produce differentiated air transport services. There are three airlines that produce an identical air transport service.

Chapter 9

240

3.

In a market consisting of two airlines that supply an identical airline service, airline A has a variable cost per passenger of c1 = 100 and airline B has a variable cost per passenger of c2 = 120. The market price elasticity of demand is -0.75. What is the profit maximising market clearing price in this model? (a) P ‫= כ‬ (b) P ‫= כ‬ (c) P = ‫כ‬

(d) P ‫= כ‬

భబబశభమబ ቁ మ

ିଶ×଴.଻ହ×ቀ

ିଶ×଴.଻ହାଵ ିଶ×଴.଻ହ×(ଵ଴଴ାଵଶ଴) ିଶ×଴.଻ହାଵ భబబశభమబ ିଶ×ଵ.ହ×ቀ ቁ మ

ିଶ×ଵ.ହାଵ భబబశభమబ ି଴.଻ହ×ቀ ቁ మ

ି଴.଻ହାଵ

On a particular airline route there are two airlines that operate identical services. The market inverse demand curve for the route is P = 1,500 – 0.00005 u (Q1 + Q2), where P is the market clearing average airfare that is charged by both airlines, Q1 is the passenger supply of airline 1 in 1 year and Q2 is the passenger supply of airline 2 in 1 year. The passenger varying cost per passenger of airline 1 is c1 = 150 and for airline 2 it is c2 = 200. The fixed cost of both airlines 1 and 2 per year are K1 = K2 = 1,000,000. Use this information to answer the following 9 questions. 4.

5.

What is the market clearing average airfare when industry output is 1,000,000 passengers per year? (a) P = 1,500 - 5 u 1,000,000 (b) P = 1,500 – 0.00005 u 1,000,000 (c) P = 1,500 – 0.00005 u (1,000,000 + 1,000,000) (d) P = 1,500 – 0.00005 u (1,000 + 1,000) What is the market price elasticity of demand at an average airfare of P = 1,000? ିଵ

(a) ቀ ቁ × ቆ భ,ఱబబబ ଴.଴଴଴଴ହ

ଵ,଴଴଴

భ బ.బబబబఱ బ.బబబబఱ×భ,బబబ

ି

(b) (െ0.00005) × ቆ భ,ఱబబబ ିଵ

(c) ቀ ቁ×ቆ ଴.଴଴଴଴ହ ିଵ

ቇ

ଵ,଴଴଴

భ బ.బబబబఱ బ.బబబబఱ×భ,బబబ భ,ఱబబ భ ି బ.బబబబఱ బ.బబబబఱ×భ,బబబ

ି

ଵ,଴଴଴

ቇ

ቇ

ଵ,଴଴଴

(d) ቀ ቁ × ቀଵ,ହ଴଴ି଴.଴଴଴଴ହ×ଵ,଴଴଴ቁ ଴.଴଴଴଴ହ 6.

What is the revenue function of airline 2? (a) R2 = 1,500 – 0.00005 u (Q1 + Q2) u Q2 (b) R2 = 1,500 – 0.00005 u (Q1 + Q2) u (Q1 + Q2) (c) R2 = 1,500 u Q2 – 0.00005 u (Q1 + Q2) u Q2 (d) R2 = 1,500 u (Q1 + Q2) – 0.00005 u (Q1 + Q2) u (Q1 + Q2)

7.

What is the cost function of airline 1? (a) C1 = 150 + 1,000,000 (b) C1 = 150 u Q1 + 1,000,000 (c) C1 = 150 + 1,000,000 u Q1 (d) C1 = 150 u Q1 + 1,000,000/Q1

8.

What is the profit function of airline 2? (a) S2 = 1,500 – 0.00005 u (Q1 + Q2) u Q2 - 150 u Q1 - 1,000,000 (b) S2 = 1,500 u Q2 – 0.00005 u (Q1 + Q2) u Q2 - 200 u Q1 - 1,000,000 (c) S2 = 1,500 u Q1 – 0.00005 u (Q1 + Q2) u Q2 - 200 u Q1 - 1,000,000 (d) S2 = 1,500 u (Q1 + Q2) – 0.00005 u (Q1 + Q2) u (Q1 + Q2) + 200 u Q1 + 1,000,000

9.

What is the first order condition for maximising profit of airline 2? ୢ஠మ (a) = 1,500 െ 0.00005 × Qଵ െ 2 × 0.00005 × Qଶ െ 200 = 0 (b) (c) (d)

ୢ୕మ ୢ஠మ ୢ୕మ ୢ஠మ ୢ୕మ ୢ஠మ ୢ୕మ

= 1,500 െ 0.00005 × Qଵ െ 2 × 0.00005 × Qଶ = 0 = 1,500 െ 2 × 0.00005 × Qଶ െ 200 = 0 = 1,500 െ 0.00005 × Qଵ െ 2 × 0.00005 × Qଶ െ 200 × Qଶ = 0

Monopoly and Oligopoly Airline Competition

241

10. What is the profit maximising level of airline 2 passenger supply as a function of airline 1 supply? ଵ,ହ଴଴ିଶ଴଴ (a) Q‫כ‬ଶ = (b) Q‫כ‬ଶ = (c) Q‫כ‬ଶ = (d) Q‫כ‬ଶ =

ଶ×଴.଴଴଴଴ହ ଵ,ହ଴଴ି଴.଴଴଴଴ହ×୕భ

ଶ×଴.଴଴଴଴ହ ଵ,ହ଴଴ି଴.଴଴଴଴ହ×୕భ ିଶ଴଴ ଶ×଴.଴଴଴଴ହ ଵ,ହ଴଴ି଴.଴଴଴଴ହ×୕భ ିଶ଴଴ ଴.଴଴଴଴ହ

11. What is the profit maximising market output? ଶ×(ଷ,଴଴଴ିଷହଽ) (a) Q‫= כ‬ (b) Q‫= כ‬ (c) Q‫= כ‬ (d) Q‫= כ‬ 12. (a) (b) (c) (d)

ଷ ଶ×଴.଴଴଴଴ହ

ଷ,଴଴଴ିଷହ଴ ଵ,ହ଴଴ିଷହ଴ ଶ×଴.଴଴଴଴ହ ଷ,଴଴଴ିଷହ଴ ଶ×଴.଴଴଴଴ହ

What is the profit maximising market clearing price? 1,675 1,500 2,000 1,200

Two full-service airlines, A and B, compete against each other on a route. They produce a differentiated product. The average airfare functions of the two airlines are PA = 600 – 0.005QA – 0.0025QB and PB = 500 – 0.003QA – 0.006QB where QA and QB are the levels of passenger output of airlines A and B for a year. The variable cost per passenger of airlines A and B are respectively cA = 200 and cB = 220. The fixed cost is K for both airlines. Use this information to answer the following 8 questions. 13. What is the profit function of airline A? (a) Ɏ୅ = 500Q୆ െ 0.003Q୅ Q୆ െ 0.006Qଶ୆ െ 220Q୆ െ K (b) Ɏ୅ = 600Q୆ െ 0.005Q୅ െ 0.0025Q୅ Q୆ െ 200Q୅ െ K (c) Ɏ୅ = 600Q୅ െ 0.005Qଶ୅ െ 0.0025Q୅ Q୆ െ 200Q୅ െ K (d) Ɏ୅ = 600 െ 0.005Qଶ୅ െ 0.0025Q୅ െ 200Q୅ െ K 14. What is the profit function of airline B? (a) Ɏ୆ = 500Q୆ െ 0.003Q୅ Q୆ െ 0.006Qଶ୆ െ 220Q୆ െ K (b) Ɏ୆ = 600Q୆ െ 0.005Q୅ െ 0.0025Q୅ Q୆ െ 200Q୅ െ K (c) Ɏ୆ = 600Q୅ െ 0.005Qଶ୅ െ 0.0025Q୅ Q୆ െ 200Q୅ െ K (d) Ɏ୆ = 500Q୆ െ 0.003Qଶ୆ െ 0.0025Q୆ െ 200Q୆ െ K 15. What is the first order condition for profit maximisation for airline A? ୢ஠ఽ = 600 െ 2 × 0.005Q୅ െ 0.0025Q୆ െ 200 = 0 (a) (b) (c) (d)

ୢ୕ఽ ୢ஠ఽ ୢ୕ఽ ୢ஠ఽ ୢ୕ఽ ୢ஠ఽ ୢ୕ఽ

= 600 െ 2 × 0.005Q୅ െ 0.0025Q୅ െ 200 = 600 െ 0.005Q୅ െ 0.0025Q୆ െ 200 = 0 = 600 െ 2 × 0.005Q୅ െ 0.0025Q୆ െ 220 = 0

16. What is the first order condition for profit maximisation for airline B? ୢ஠ా (a) = 500 െ 2 × 0.003Q୆ െ 0.0025Q୅ െ 220 = 0 (b) (c) (d)

ୢ୕ా ୢ஠ా ୢ୕ా ୢ஠ా ୢ୕ా ୢ஠ా ୢ୕ా

= 500 െ 2 × 0.005Q୅ െ 0.0025Q୆ െ 220 = 0 = 600 െ 0.005Q୅ െ 0.0025Q୆ െ 200 = 0 = 500 െ 0.003Q୅ െ 2 × 0.006Q୆ െ 220 = 0

17. What is the 2x2 coefficient matrix in the matrix system of the first order conditions? െ0.005 െ0.0025 (a) ቂ ቃ െ0.003 െ0.006 െ2 × 0.005 െ2 × 0.0025 (b) ቂ ቃ െ2 × 0.003 െ2 × 0.006

Chapter 9

242

െ2 × 0.005 െ0.003 െ2 × 0.005 (d) ቂ 0

(c) ቂ

െ0.0025 ቃ െ2 × 0.006 0 ቃ െ2 × 0.006

18. What is the 2x1 vector of exogenous variables in the matrix system of the first order conditions? 600 (a) ቂ ቃ 200 600 (b) ቂ ቃ 500 600 െ 200 (c) ቂ ቃ 500 െ 220 െ200 (d) ቂ ቃ െ220 19. What is the solution in matrix terms for the output of firm A? (a) Q‫כ‬୅ =

ିଶ×଴.଴଴ହ ଺଴଴ିଶ଴଴ ቚ ቚ ି଴.଴଴ଷ ହ଴଴ିଶଶ଴ ିଶ×଴.଴଴ହ ି଴.଴଴ଶହ ቚ ቚ ି଴.଴଴ଷ ିଶ×଴.଴଴଺

(b) Q‫כ‬୅ =

ିଶ×଴.଴଴ହ ି଴.଴଴ଶହ ቚ ቚ ି଴.଴଴ଷ ିଶ×଴.଴଴଺ ଺଴଴ିଶ଴଴ ି଴.଴଴ଶହ ቚ ቚ ହ଴଴ିଶଶ଴ ିଶ×଴.଴଴଺

(c) Q‫כ‬୅ =

଺଴଴ିଶ଴଴ ି଴.଴଴ଶହ ቂ ቃ ହ଴଴ିଶଶ଴ ିଶ×଴.଴଴଺ ିଶ×଴.଴଴ହ ି଴.଴଴ଶହ ቃ ቂ ି଴.଴଴ଷ ିଶ×଴.଴଴଺

(d) Q‫כ‬୅ =

଺଴଴ିଶ଴଴ ି଴.଴଴ଶହ ቚ ቚ ହ଴଴ିଶଶ଴ ିଶ×଴.଴଴଺ ିଶ×଴.଴଴ହ ି଴.଴଴ଶହ ቚ ቚ ି଴.଴଴ଷ ିଶ×଴.଴଴଺

20. What is the solution in matrix terms for the output of firm B? (a) Q‫כ‬୆ =

ିଶ×଴.଴଴ହ ଺଴଴ିଶ଴଴ ቚ ି଴.଴଴ଷ ହ଴଴ିଶଶ଴ ିଶ×଴.଴଴ହ ି଴.଴଴ଶହ ቚ ቚ ି଴.଴଴ଷ ିଶ×଴.଴଴଺

(b) Q‫כ‬୆ =

଺଴଴ିଶ଴଴ ି଴.଴଴ଶହ ቚ ቚ ହ଴଴ିଶଶ଴ ିଶ×଴.଴଴଺ ଺଴଴ିଶ଴଴ ି଴.଴଴ଶହ ቚ ቚ ହ଴଴ିଶଶ଴ ିଶ×଴.଴଴଺

(c) Q‫כ‬୆ =

ିଶ×଴.଴଴ହ ି଴.଴଴ଶହ ቃ ି଴.଴଴ଷ ିଶ×଴.଴଴଺ ିଶ×଴.଴଴ହ ି଴.଴଴ଶହ ቃ ቂ ି଴.଴଴ଷ ିଶ×଴.଴଴଺

(d) Q‫כ‬୆ =

଺଴଴ିଶ଴଴ ି଴.଴଴ଶହ ቚ ቚ ହ଴଴ିଶଶ଴ ିଶ×଴.଴଴଺ ିଶ×଴.଴଴ହ ି଴.଴଴ଶହ ቚ ቚ ି଴.଴଴ଷ ିଶ×଴.଴଴଺

ቚ

ቂ

9.7 Measuring Competition Competition can be measured by determining the intensity of competition. The best measure of competitive intensity in my opinion is described by an index called the Herfindahl-Hirschman Index or the HHI. The HHI was developed by two economists, Orris Herfindahl and Albert Hirschman between 1945 and 1950 (Taylor 2020). It is used extensively in competition law to determine if the merger of two entities results in a level of market concentration in an industry that is likely to cause a substantial lessening of competition, which in turn results in higher prices and lower consumer welfare. The HHI is calculated by using information about the market shares of each airline on a route in the following way: HHI = (Share Airline 1 u100)2 + (Share Airline 2 u100)2 + ……. + (Share Airline N u100)2

(9.91)

The index (9.91) says that competition on a route can be measured by adding up the market shares squared of each of the airlines in a market. Note in the calculation in (9.91) that the market share should not be left in decimal form – it must be multiplied by 100. When there is only one airline operating in the market, which means the market is a monopoly, then the HHI is equal to 10,000 = 1002. In this case there is a single calculation on the right-hand side of (9.91) which is the number 100 squared. This is the highest level of the HHI and is the highest level of market concentration. When there are many airlines operating in the market, or the market is highly competitive, then the HHI

Monopoly and Oligopoly Airline Competition

243

is close to zero (but not equal to zero). A low value for the HHI therefore indicates that the market is not concentrated, or competition is high, and a high value indicates that the market is highly concentrated, or competition is low. Let us now consider an example of how the HHI is used to calculate concentration in less extreme cases. Let us suppose that a route is serviced by a duopoly with one airline holding a market share of 60% and the other airline holding a share of 40%. The HHI in this case using (9.91) is: HHI = 602 + 402 = 5,200 Let us now suppose that our route consists of 5 airlines operating services, with one dominant airline possessing an 80% market share and the other remaining 4 airlines with 5% each. The HHI in this case using (9.91) is computed as follows: HHI = 802 + 52 + 52 + 52 + 52 = 6,500 Let us finally assume that there are three airlines in the market each with one-third of the market share. The HHI in this case is calculated to be: HHI = 3 u ቀ

ଵ଴଴ ଶ

ቁ = 3,333

ଷ

In this example we can see that the route in which just 3 airlines operate has about half the levels of concentration according to the HHI as the route that has 5 operating airlines. The HHI generates this outcome because it takes into consideration the market share of each operating airline in determining the level of concentration of competition on airline routes as discussed in section 9.1. The U.S. Department of Justice considers an airline route with a HHI of less than 1,500 to be a competitive route, and unlikely to have anti-competitive effects. An airline route with a HHI that is between 1,500 and 2,500 is considered moderately competitive and is likely to raise concerns about anti-competitive behaviour. The U.S. Department of Justice is likely to conduct further analysis on routes with HHI’s at these levels to determine if there is any further evidence of actual or potential anti-competitive behaviour. A route in which the HHI is 2,500 or greater is believed to be a highly concentrated route and more likely to give rise to anti-competitive behaviour (United States Department of Justice 2019). One of the difficulties in applying the HHI to aviation routes is associated with the problem of market definition that was raised in section 9.1.2. To compute the HHI it is necessary to have a clear definition of the route, because this will determine the number of airlines that operate on the route. Once we know the number of airlines that operate on the route, we can then use their market shares to determine the HHI. There are two significant problems in obtaining a clear definition of the market along routes. The first involves an answer to the question: Should we include indirect services in our definition of the market? The second involves an answer to the question: Should we include other modes of transport in our definition of the market? These are very difficult questions to answer. If the answer to the first question is that indirect services should be included, this could conceivably double or perhaps triple the number of airlines that are operating on the relevant route, which could change the assessment of the route using the HHI from one that is highly concentrated to one that is highly competitive. If the answer to the second question is that mode of transport should be taken into consideration, and most passengers can drive their own vehicle between the relevant city pairs, then it would seem very difficult to argue that the route is susceptible to anti-competitive behaviour. As indicated in section 9.1.2, whether the market definition should include indirect services and mode of transport competition depends heavily on the passenger demographics of the route. Specifically, it will depend on whether the passengers on the route place a high value on travel time or not and therefore a high value on the opportunity cost associated with air travel. If they do place a high value on travel time, this limits the amount of substitution that is likely to occur from taking direct flights into taking indirect flights, and from flying by air to flying by land or sea. As there is limited substitution between the types of travel, this may persuade competition regulators to define the market as simply direct air services between the relevant cities. Conversely, if the passenger demographic is such that there is likely to be considerable substitution from direct to indirect travel and from air travel to other modes of transport, then competition authorities may decide to define the market more broadly to include indirect travel and other transport modes. The HHI is used as an indicator of market concentration because it is related to the aggregate margins that are earned in a market. To see this, consider the n-player Cournot model of airline competition once again presented in section 9.6. In this model, we know that the first order condition for profit maximisation of the representative airline i is given by the following expression: P + qiu

ୢ୔

ୢ୕

– ci = 0

(9.92)

which is a simplified version of equation (9.75) given in section 9.6.1. We can re-write (9.92) in terms of the margin of the ith airline: (P – ci) = - qiu

ୢ୔

ୢ୕

(9.93)

Chapter 9

244

If we divide both sides of (9.93) by P we obtain: ୔ିୡ౟ ୔

= - qiu

ୢ୔ ଵ

(9.94)

ୢ୕ ୔

The left-hand side of (9.94) is the margin of the ith airline. If we then multiply the top and bottom of the right-hand side of (9.94) by Q this yields: ୔ିୡ౟ ୔

୯

= - ౟u

ୢ୔ ொ

୕ ୢ୕ ୔

{-

௦೔

(9.95)

ఌ

Equation (9.95) says that the profit margin of the ith airline in the market is equal to the division of the market share of the ith airline, si, by the absolute value of the market price elasticity of demand, H. If we multiply the top and bottom of the right-hand side of (9.95) by 100, then square the left and right-hand sides of this expression, and then sum this squared term over every airline in the market, we obtain: σ୬୧ୀଵ ቀ

୔ିୡ౟ ଶ ୔

ቁ =

మ σ౤ ౟సభ(ଵ଴଴×௦೔ )

ଵ଴,଴଴଴୬கమ

=

ୌୌ୍ ଵ଴,଴଴଴×୬கమ

(9.96)

Equation (9.96) tells us that the aggregate margins squared for the route that is being analysed is proportional to the HHI, with the proportion determined by the number of airlines in the market and the price elasticity of demand at the market level. As the HHI is proportional to the aggregate margins that are earned along a route then changes in the HHI will also tell us about changes in margins, or the ability to increase the price above marginal cost, which is the essence of anti-competitive behaviour.

9.8 Understanding why Fares Differ Across Routes and Cabins 9.8.1 Why do Fares Differ Across Routes? We can use our model of the average airfare for n airline players on a route from section 9.6 to understand why the average airfare may be different across different routes. Let us suppose that we have two routes A and B. The n-airline Cournot model from section 9.6 at equation (9.80) tells us that the average airfares on the two routes A and B are predicted to be: ୬ க ୡത

ఽ P୅ = ୬ ఽக ఽାଵ ఽ ఽ

୬ க ୡത

ా P୆ = ୬ ాக ాାଵ ా ా

The model that we have built indicates that the average airfares will be different across the two routes A and B if at least one of the number of competitors, the price elasticity of demand and the average marginal cost differ across the two routes. If route A is a higher cost route than B, or ܿҧ஺ > ܿҧ஻ , then the average airfare will be higher for A then for B, other things being equal. A route may have higher unit costs than another route for a variety of reasons, but the most important is likely to be the route distance. If route A involves a greater distance and longer travel time than route B, then route A marginal costs will be higher than that of route B. For example, the city pair Oslo (OSL) to Stockholm (ARN), which is a distance of 386km, will have higher variable passenger costs per passenger than the city pair OSL to LHR, which is 1,207km. This is because there are a range of airline operating costs in the medium run that are higher on routes that are longer, such as fuel costs and the wages that are paid to the flight crew (both cabin crew and pilots). Different routes involve flying into different airports, and different airports charge different airport and terminal navigation charges. Costs therefore might be higher on route A compared to route B because the airports that constitute A route may charge higher airport and terminal navigation charges than the airports that constitute B route. For example, if we continue our OSL-LHR versus OSL-ARN example from before, the airport charges at ARN at the time of writing for a flight involving a 180 seat Boeing 737-800 aircraft with an 80% seat factor is approximately £3.3 per passenger while the airport charge for LHR at the time of writing is around £16 per passenger (Swedavia Airports 2021, 5; Heathrow Terms and Conditions 2021, 31). It follows that higher airport charges on the route OSL-LHR will driver higher marginal passenger costs on this route compared to OSL-ARN. Different routes will also involve the use of different sets of air space. Different countries will apply different charges for the use of their air space, which may in turn cause route A to have different marginal passenger costs than route B. Other costs that might be different across routes include the cost of providing airport staff and customer service, ground handling and catering costs, and fuel costs may be different because different weather conditions may result in more fuel burn and higher fuel costs for one route compared to another. Another important driver of the difference in fares across routes is the difference in the price elasticity of demand. The price elasticity of demand varies across routes, in the main part, because the demand demographic or the demand mix varies across routes. Another way of saying this is that the mix of leisure and business-purpose traffic varies across different routes. The greater is the leisure mix on a route, the more elastic is demand to the average airfare on that route,

Monopoly and Oligopoly Airline Competition

245

and the lower is the average airfare. For example, there is more likely to be a higher percentage of leisure travellers on the route London City Airport to Ibiza compared to the route London Heathrow to Frankfurt. There is more likely to be a higher percentage of business-purpose travellers on Sydney to Melbourne compared to Sydney to Hamilton Island. On this basis alone, and after adjusting for the distance travelled, we would expect lower airfares on London City to Ibiza compared to London Heathrow to Frankfurt, and Sydney to Hamilton Island compared to Sydney to Melbourne. We would also expect lower airfares on the route Sydney to Bali (Denpasar) compared to the route Sydney to Singapore or Sydney to Hong Kong. For example, according to the National Visitor Survey in Australia in calendar 2019, there were 66,000 overnight trips using air transport by Sydney residents to the Whitsundays, with 57,000 of those or 86.4% for holiday or leisure purposes. The Whitsundays is a region of North Queensland, Australia that includes one of the natural wonderers of the world, the Great Barrier Reef and several other memorable reefs and islands. It is a classic leisure, sun and cocktail getaway destination for many Australians and overseas tourists, with a travel distance by air of 1,513km. Conversely, over 2019 there were 154,000 Brisbane, Queensland residents who travelled to Adelaide, South Australia by air. These are the capital cities of their respective states, with relatively large populations and a significant amount of business activity and trade between them. Just 16.2% of the travel on the Brisbane to Adelaide city pair is for holiday or leisure purposes, while 50% is for business purposes.51 The Brisbane to Adelaide great circle distance of 1,621km is close to the Sydney to Whitsundays distance of 1,513km. Given that Sydney to the Whitsundays has a far greater percentage of passengers who are travelling for leisure purposes than Brisbane to Adelaide we would expect that the elasticity of air travel demand to the average airfare will be more elastic for Sydney to Whitsundays than for Brisbane to Adelaide. For example, Sydney to Whitsundays may have an elasticity of air travel demand to the average airfare of HA = -2 while that for Brisbane to Adelaide may be HB = -1. If there are nA = nB = 2 airlines operating on both routes and it costs both airlines ܿҧ஺ = ܿҧ஻ = A$200 per passenger to operate services on the route, then the average airfares expected on both routes using our pricing formulae (9.80) above are: Pୗ୷ୢ୬ୣ୷ି୛୦୧୲ୱ୳୬ୢୟ୷ୱ =

୬ఽ கఽ ୡതఽ ୬ఽ கఽ ାଵ

=

ଶ(ିଶ)(ଶ଴଴) ଶ(ିଶ)ାଵ

= A$267

P୆୰୧ୱୠୟ୬ୣି୅ୢୣ୪ୟ୧ୢୣ =

୬ా கా ୡതా ୬ా கా ାଵ

ଶ(ିଶ)(ଶ଴଴)

=

ଶ(ିଵ)ାଵ

= A$800

The airlines will charge significantly more on Brisbane-Adelaide than on Sydney-Whitsundays because demand is much more elastic on Sydney-Whitsundays than on Brisbane-Adelaide. Lastly, the other important parameter in determining why airfares will differ across routes is the extent of competition. If there are more airlines operating services on route A compared to route B, or nA > nB, then we would expect to see lower airfares on route A, other things being equal. For example, if the costs on both routes is $200 and the elasticity of air travel demand on both routes is -1.5, but route A has nA = 2 competitors and route B has nB = 4 competitors then the difference in airfares on the two routes using our pricing formulae becomes: ୬ఽ கఽ ୡതఽ

P୅ =

୬ఽ கఽ ାଵ

=

ଶ(ିଵ.ହ)(ଶ଴଴) ଶ(ିଵ.ହ)ାଵ

= $300

P஻ =

୬ా கా ୡതా ୬ా கా ାଵ

ସ(ିଵ.ହ)(ଶ଴଴)

=

ସ(ିଵ)ାଵ

= $240

The greater competition on route B generates a $60 lower airfare on route B compared to route A.

9.8.2 Understanding why Fares Differ Across Cabins First class average airfares are higher than business class average airfares, which are higher than premium economy average airfares which are higher than economy cabin average airfares. In this section we use the pricing formula (9.80) to understand why this is the case. Using (9.80), we can write this formula on a cabin basis for a given route as follows: ୬ க ୡത

P୊‫ = כ‬୬ ూக ూାଵూ = First class average airfare ూ ూ

‫כ‬ P୔୉ =

୬ౌు கౌు ୡതౌు ୬ౌు கౌు ାଵ

= Premium economy average airfare

୬ க ୡത

ా P୆‫ = כ‬୬ ాக ాାଵ = Business class average airfare ా ా

୬ க ୡത

P୉‫ = כ‬୬ ుக ుାଵు = Economy class average airfare ు ు

The average airfare is different across the cabins because the three key elements of the formula - costs, elasticity, and competition - are different across the cabins. Marginal passenger cost is higher in the premium cabins, competition is lower in the premium cabins, and premium cabin demand is less elastic to the average airfare. Let us first understand why the variable cost per passenger associated with servicing the premium cabins is higher. The first important reason why premium cabin cost is higher is because there are more cabin crew attending to premium cabins on a per passenger basis and thus there are greater cabin crew costs allocated to premium cabins. For example, on a flight between Sydney and Melbourne on a 180 seat Boeing 737-800 aircraft there are 14 seats in business class and 166 seats in economy. Two cabin crew attend to the needs of the 14 seats in business class, which is 7 seats per cabin crew member, while there are just 3 cabin crew attending to the needs of the 166 passengers in economy class, with around 55 seats per cabin crew member in economy. This means that 14 seats in business class will pay for the 51

Both the Brisbane to Adelaide and Sydney to the Whitsundays air passenger data by purpose were made available by the Tourism Research Australia on-line database to which this author has a subscription.

Chapter 9

246

hours of two cabin crew members while 166 seats in economy will pay for the hours of three cabin crew, generating far greater crew costs per seat in business than in economy. The second reason why premium cabin costs per passenger are higher is because premium cabin passengers are offered more food and beverages, higher quality food and beverages, and greater food and beverage variety. This leads to higher air meal costs per passenger in the premium cabins. Premium cabin passengers also have higher checked-in and carry-on baggage allowances. This adds more to the weight that is loaded onto the aircraft, which in turn leads to more fuel burn, and it increases the amount of time required by ground handling crews to handle premium passenger baggage. Premium passengers are therefore allocated more of the airline’s fuel and ground handling costs. Premium cabin seats are also allocated more costs because they take-up more surface area inside the aircraft. For example, Table 9.1 below presents the seat dimensions by cabin of the seats that are sold on the Singapore Airlines A380. Cabin First Business Premium Economy Economy

Seat Pitch (inches) 81 55 38 32

Seat Width (inches) 35 30 19.5 19

Surface Area (Squared inches) 2,835 1,650 741 608

Source: Seat Guru 2021.

Table 9-1: Seat Dimensions on the Singapore Airlines Airbus A380 Aircraft The seat pitch in the second column of Table 9-1 is the distance between a particular point on a seat and the same point on the seat in front of it or behind it. The pitch therefore effectively measures the length of the seat. We can see that the surface area taken up by the first-class seat on a Singapore Airlines Airbus A380 is more than four times the surface area of an economy class seat, just below four times the surface area of a premium economy seat and almost twice the surface area of a business class seat. Costs that are independent of the number of passengers and the cabins in which they are seated, for example the wages of the pilots, maintenance expenses and aircraft capital costs (such as depreciation and operating lease costs), are allocated across cabins and passengers. One method of allocating these fixed costs across the seats of the aircraft involves ascertaining the total surface area of the aircraft taken up by these seats. On a per seat basis, and using this method of allocating costs, first and business class seats are allocated more costs than premium economy and economy class seats, leading to higher costs per seat. To demonstrate this allocated cost point, let us consider our Singapore Airlines A380 aircraft once again with 12 first class seats, 60 business class seats, 38 premium economy seats and 333 economy seats (representing 443 seats in total). The A380 is used by Singapore Airlines on a flight from Singapore to Los Angeles, which has a great circle distance of 14,113km. The costs that are fixed, unrelated to the number of passengers amounts to S$400,000 per flight. These costs are assumed to be allocated across the cabins of the A380 according to the percentage of the total seats surface area that each cabin contributes. The total seat surface area of the aircraft that is devoted to each class of travel using the surface areas presented in Table 9-1 above, are given below in Table 9-2 along with the S$400,000 cost that is allocated to each cabin and each cabin seat. Cabin

Seats

Total Surface Area (square inches)

% Surface Area

Cost Allocated to the Cabins

Allocated Cost per Seat

(1) First Business Premium Economy Economy Total

(2) 12 60 38

(3) 34,020 99,000 28,158

(4) 9.4% 27.2% 7.7%

(5) S$37,421 S$108,898 S$30,973

(6) = (5) y (2) S$3,118 S$1,815 S$815

333 443

202,464 363,642

55.7% 100%

S$222,707 S$400,000

S$669 S$903

Source: Seat Guru 2021.

Table 9-2: Allocated Cost per Seat by Cabin Singapore Airlines A380 Flight Singapore to Los Angeles Illustration We can see in Table 9-2 that the first-class cabin is allocated 9.4% of the S$400,000 fixed cost even though this cabin only has 12/443 = 2.7% of the seats on the aircraft. The allocated cost per seat in the first-class cabin is therefore very high at S$3,118. This is followed by business class with an allocation of S$1,815 per seat, then premium economy with S$815 and economy with the lowest cost allocation of S$669. Using the cost per seat estimates from Table 9-2 as our estimates of ܿҧி , ܿҧ஻ , ܿҧ௉ா and ܿҧா , assuming that nF = nB = nPE = nE = 3 and HF = HB = HPE = HE = -0.75 then the estimated average airfares for each cabin are:

Monopoly and Oligopoly Airline Competition P୊‫= כ‬

ଷ(ି଴.଻ହ)(ଷ,ଵଵ଼) ଷ(ି଴.଻ହ)ାଵ

‫כ‬ P୔୉ =

ଷ(ି଴.଻ହ)(଺଺ଽ) ଷ(ି଴.଻ହ)ାଵ

= S$5,613

P୆‫= כ‬

ଷ(ି଴.଻ହ)(ଵ,଼ଵହ)

= S$1,467

P୉‫= כ‬

ଷ(ି଴.଻ହ)(ଽ଴ଷ)

ଷ(ି଴.଻ହ)ାଵ

ଷ(ି଴.଻ହ)ାଵ

247

= S$3,267

= S$1,204

We can see that differences in the costs allocated to each cabin generates significant variation in the average airfares across the cabins, assuming that the number of competitors and the airfare elasticity of air travel demand are equal across the cabins. The premium cabins are more likely to have a higher mix of business-purpose passengers, and high income and net worth passengers. These passengers have a higher willingness and capacity to pay, which in turn means that their demand for air travel is usually less sensitive to airfares. First class demand is less elastic than business, business is less elastic than premium economy and premium economy is less elastic than economy. When demand is less elastic, this enables the airline to charge a higher mark-up over cost, resulting in higher airfares. To see this in an example, let us refer once again to the Singapore to Los Angeles illustration. Rather than assuming the air travel elasticity of demand is equal across the cabins, we now assume that they vary across the cabins in the following way HF = -0.6, HB = -0.65, HPE = -0.75 and HE = -1. Assuming the same allocated cost figures obtained from Table 9-2 and assuming that the number of carriers is once again equal to 3 across all cabins, then the average airfares for the different cabins using (9.80) are: P୊‫= כ‬

ଷ(ି଴.଺)(ଷ,ଵଵ଼) ଷ(ି଴.଺)ାଵ

‫כ‬ P୔୉ =

ଷ(ି଴.଻ହ)(଺଺ଽ) ଷ(ି଴.଻ହ)ାଵ

= S$7,016 = S$1,467

P୆‫= כ‬

ଷ(ି଴.଺ହ)(ଵ,଼ଵହ)

P୉‫= כ‬

ଷ(ିଵ)(ଽ଴ଷ)

ଷ(ି଴.଺ହ)ାଵ

ଷ(ିଵ)ାଵ

= S$3,726

= S$1,003

These assumptions generate a much greater difference between the premium cabin airfares and the premium and economy cabin fares. On a particular route, airlines will fly aircraft with different configurations. As a result, there could be different levels of competition in the classes of travel even within the same route. For example, on London Heathrow to John F. Kennedy in New York at the time of writing, the operating airlines include Virgin Atlantic, which flies the A350, the A330 and the A340, Delta Air Lines which flies the A330, British Airways which flies the B747 and the B777, and American Airlines which flies the B777. This means that the route is operated by 5 different aircraft types. Virgin Atlantic’s A350, A330 and A340 fleet have a three-class configuration, with Upper, Premium and Economy classes. The British Airways B747 has four classes of travel, First, Club World, World Traveller Plus and World Traveller with varying combinations of premium and economy seats. The airline’s B777 is fitted with both 4 and 3 class configurations. The Delta Air Lines A330 aircraft has a three-class configuration consisting of Delta One, Delta Comfort and Economy. And the American Airlines B777 fleet has five classes, including First, Business, Premium Economy, Economy extra and Economy. On any day, there will be four operators on the route in a quadropoly, but there may not be four competitors offering each class of travel. This means that we should expect different classes of travel facing different levels of competitive intensity because different airlines operate different configurations. As different classes of travel face different levels of competition, this in turn affects the average airfares in one class of travel compared to the others. Continuing our Singapore to Los Angeles example once again, instead of assuming that there are 3 competitors in each class of travel, we assume that first, business, and premium economy class has just two competitors or nA = nB = nPE = 2, while economy has three competitors. In this case the average airfares by class become: P୊‫= כ‬

ଶ(ି଴.଺)(ଷ,ଵଵ଼) ଶ(ି଴.଺)ାଵ

‫כ‬ P୔୉ =

ଶ(ି଴.଻ହ)(଺଺ଽ) ଶ(ି଴.଻ହ)ାଵ

= S$18,708

P୆‫= כ‬

ଶ(ି଴.଺ହ)(ଵ,଼ଵହ)

= S$4,075

P୉‫= כ‬

ଷ(ିଵ)(ଽ଴ଷ)

ଶ(ି଴.଺ହ)ାଵ

ଷ(ିଵ)ାଵ

= S$7,865

= S$1,003

We can see in this case that the reduction in competition has significantly lifted the first, business and premium economy average airfares relative to economy.

Quiz 9-4 Measuring Competition, Why Fares Differ Across Routes and Classes 1. (a) (b) (c) (d)

A triopoly consisting of United Airlines, Delta Air Lines and American Airlines is such that United has 30% market share, Delta has 45% market share and American has 25% market share. What is the HerfindahlHirschman index in this market? HHI = 0.3 + 0.45+0.25 HHI = 30 + 45+25 HHI = (30 + 45 + 25)u100 HHI = 302 + 452+252

Chapter 9

248

2. (a) (b) (c) (d) 3. (a) (b) (c) (d) 4. (a) (b) (c) (d)

Which of the following would be valid reasons as to why route A will have a lower average airfare than route B? There are fewer airlines competing on route A compared to route B. Route A has a lower airfare elasticity of demand than route B. Route A has lower airport charges than route B. There are more low-cost carriers flying on route A than on route B. Which of the following would be valid reasons as to why route A will have a lower average airfare than route B? The mix of low-cost carriers is lower on route A compared to route B. The demand demographic is such that there is a higher proportion of leisure travellers on route A than on route B. Route A is longer than route B. Route A serves better passenger meals than route B. Which of the following would be valid reasons as to why first-class fares are greater than business class fares on a particular route? There are more cabin crew attending to each passenger in first class than business class. First class passengers are less sensitive to fares then business class passengers. Other airlines that compete with the airline in question on the route in question do not offer first class as an option. All the above.

CHAPTER 10 AIRLINE RELATIONSHIPS AND BUSINESS MODELS

There are number of different types of relationships that airlines can form. They range from as simple as an airline selling a ticket to a passenger who flies on a route that is operated by another airline, to something as complex as one airline buying or merging with another airline. In between these two extremes, airlines will share resources, they will form joint ventures, they will share revenue and they will form alliances. There are generally three reasons why airlines form relationships. The first is that they reduce the competition they face. As we saw in chapter 9, when there is less competition on airline routes this generally results in lower passenger numbers and higher average airfares on those routes, which in turn improves an airline’s profitability. If a reduction in competition can improve an airline’s yields and revenue by just 1% this can in turn result in a significant improvement in profitability. The second reason why airlines form relationships is that they allow the airline to expand the network of city pairs that can be offered to passengers, which in turn improves the demand for their product and thus their profitability. In the case of some relationships, such as interline and codeshare relationships which we will discuss in sections 10.2 and 10.3 respectively, an airline can expand the network of routes that it offers to passengers without even operating services on those routes. Being able to offer passengers a wide network of city pairs is important for demand and thus airline profitability because passengers like the simplicity and the convenience of one-stop-shopping for air travel services. The third reason why airlines form relationships is to reduce costs. Airlines can reduce costs by allowing the frequent flyer members of other airlines to use their lounge facilities, which saves the other airline from financing and resourcing a lounge of its own. Airline relationships can also be as simple as an airline using the ground handling and check-in resources of another airline in the case of low frequency or seasonal routes. This saves the airline paying for ground handling and check-in crews that sit around an airport with nothing to do until their planes arrive or take-off. Airlines that are a part of a joint venture, a revenue sharing arrangement, or an alliance may also save on the costs of marketing a particular route, if all the airlines that are party to the relationship commit to sharing the costs of that marketing. By reducing costs this in turn improves the profitability of airlines. Taking costs out of the business is particularly important for low-cost carriers since they operate on routes in which most passengers fly for leisure purposes. Low-cost carriers can pass-on these lower costs, at least in part, to passengers which in turn stimulates them to plan more air travel on the airline. One key risk that airlines face when they form relationships is that they draw the attention of competition regulators. This is referred to as regulatory risk. Competition regulators such as the Federal Trade Commission in the U.S.A., the Competition and Consumer Commission in Australia, the Fair Trade Commission in Japan, and the European Commission will scrutinise relationships between airlines to determine if such relationships are likely to result in a substantial lessening of competition. If a substantial lessening of competition is likely to occur, this usually manifests itself in higher airfares and margins and lower passenger welfare. Most airlines will anticipate a regulator taking an interest in the relationship it hopes to form with another airline and will apply to have the relationship endorsed by the competition regulator. If this doesn’t happen, the airlines forming a relationship may face fines and court costs if the regulator deems the relationship to have had an adverse effect on passenger welfare. This chapter will also investigate the difference in the business models of full-service airlines and low-cost carriers, two very important flying segments in world aviation. After reading this chapter you will gain a deeper understanding of why their costs are different, the extent to which they have different product offerings and the implications this may have for the markets in which they fly. I hope you enjoy learning something about airline relationships in this chapter.

10.1 Spectrum of Airline Relationships There are many potential relationships that airlines can form. The full spectrum of such relationships ranked in terms of the interest that regulatory authorities are likely to take in them, and the extent of any cooperation, is described in Figure 10-1 below.

Chapter 10

250

Limited Cooperation on Specific Routes

Interlining

Expanded cooperation to develop a joint network FF and Lounge Access

Code Sharing

Direct coordination including prices, routes, and schedules

Equity Alliance Revenue Joint Sharing Venture Investment

Merger-like integration

Merger Acquisition

Fig. 10-1: The Spectrum of Airline Cooperation Figure 10-1 indicates that there are nine key sets of relationships between airlines. In order of the extent of cooperation, they include (1) interline relationships, (2) frequent flyer program and lounge access, (3) codeshare relationships, (4) alliances, (5) revenue sharing relationships, (6) joint ventures and partnerships, (7) equity investments, (8) mergers, and (9) acquisitions. In terms of the spectrum of relationships presented in Figure 10-1, the relationships to the left of the spectrum are the weakest in terms of the potential adverse impact on competition and the consumer. The relationships to the right of the Figure 10-1 spectrum are those for which the impact on competition and the consumer may be relatively significant and will draw the attention of competition authorities. The relationship with the least amount of cooperation between airlines is the interline relationship. As discussed in section 10.2 below, interline relationships are formed to allow airlines to broaden their network offering to passengers without operating flights on that broader network. The next type of relationship with relatively low amounts of cooperation includes frequent flyer (FF) and lounge access. This involves airline A allowing the members of airline B’s frequent flyer program to earn points on A flights and to burn or redeem points on flights operated by airline A. Airline B will usually reciprocate, allowing members of airline A’s frequent flyer program to earn and burn points on B operated flights. Airline A may also allow the passengers of airline B to use airline A lounges at airports where airline B does not have a lounge, with reciprocal arrangements for airline A passengers using airline B lounges where A does not have a lounge. A deeper form of relationship involves the formation of codeshare arrangements as discussed at length in section 10.3 below. In a codeshare arrangement, airline A can sell seats onto airline B operated flights, using the flight code of airline A. This is a deeper form of relationship than we see in the case of interline arrangements, because it usually involves airlines cooperating on the scheduling of flights on certain routes along with other forms of cooperation. Airlines can also become part of alliances. As discussed in section 10.4 below, there are several major alliances in global aviation including Oneworld, Star Alliance and SkyTeam. Airlines that become a member of a major alliance form a deeper level of cooperation to save costs, improve revenue and broaden their network reach. Another type of relationship between airlines which involves a deeper level of cooperation than alliances and codeshare relationships is a revenue share relationship. This is a deeper form of relationship because airlines will coordinate on setting prices and the way they schedule capacity. The extent of cooperation on pricing is such that airlines are indifferent between selling tickets on their own operated flights and selling tickets on the flights of the airline with which it has a revenue sharing relationship. As there is a significant amount of cooperation on prices in the case of revenue sharing relationships, competition regulators take a keen interest in them and will usually only allow them to proceed if they are in competition against a highly dominant airline for the route or routes in question. This is discussed in more detail in section 10.5. A joint venture relationship involves an even deeper relationship than revenue sharing, with much stronger coordination on selling tickets, scheduling capacity, and improving the product offering of both carriers. This will be discussed in more detail in section 10.6 where we will feature several different joint venture relationships that have been formed in recent years. An airline can invest in an aviation market in two ways – it can buy aircraft and operate that aircraft on the routes in question, if permitted by law, or it can buy shares in airlines that already operate airline services in the relevant market. This latter option is referred to as an equity investment by one airline in another and is a topic that we will examine in some detail in section 10.7. The last and strongest form of relationship is that involving mergers and acquisitions. In the case of a merger, airlines join forces to become a single, larger entity often retaining their own names, such as Air France and KLM. In the case of acquisitions, one airline buys another airline that is usually in financial trouble. The acquiring airline will benefit because it has either taken out a competitor on its key routes, which generates a yield advantage, or it allows the airline to access a range of routes that it did not have access to previously, which broadens the acquiring airline’s network. Mergers and acquisition will be covered in more detail in section 10.8 below. The following sections will discuss each of these relationship types, how they might be of benefit to airlines and why competition authorities may or may not be concerned with them.

Airline Relationships and Business Models

251

10.2 Interlining52 10.2.1 What is an Interline Agreement? The term interlining is used to represent a commercial agreement between airlines that enables one airline to sell a ticket to a passenger who flies on multiple sectors and multiple airlines. The airline that sells the ticket is usually referred to as the marketing or issuing airline. The airline that carries the passenger is called the operating airline. Interline agreements will allow the operating airline to check-in and handle the baggage of a passenger that is sold a ticket by the issuing airline. To explain this further, consider the following example. A resident of Bogota Colombia visits the Avianca Airlines website and books a flight from Bogota’s El Dorado Airport (BOG) to Charles de Gaulle Airport in Paris (CDG). The flight is to depart BOG at 20:20 on Wednesday, December 11, 2019, and arrive the next day at CDG at 16:50. Avianca does not fly directly to Paris and so the Avianca website offers a 1 stop flight to Paris via Josep Tarradellas BarcelonaEl Prat Airport (BCN). The flight from BOG to BCN is an Avianca Airlines operated flight AV18 on a Boeing B787-8 aircraft while the leg from BCN to CDG is an Air France operated flight AF1649 on an Airbus A320-100/200 aircraft – refer to Figure 10-2 below for a summary map of the journey.

AF1649 AVIA18

Source: www.gcmap.com

Fig. 10-2: Map of a Trip from Bogota to Paris Via Barcelona To enable this flight to take place, which includes Avianca selling a ticket onto a flight that is operated by Air France with an Air France flight number, Avianca and Air France will need to have an interline agreement in place, which they do. The interline partners of Avianca Airlines include all airlines in the Star Alliance, which we will discuss in section 10.4 below. It also includes airlines that are outside of the alliance, including those presented in Table 10-1 below.53 Aeroflot Alaska Airlines Cathay Pacific Delta KLM Silver Airways Olympic Airlines

Aerolíneas Argentinas Alitalia JetBlue China Southern Hawaiian Airlines Korean Airlines APG Airlines

Heli Air Monaco

TAME Ecuador

Aeroméxico

Air France

Air Europa

British Airways China Airlines Cayman Airways Philippine Airlines Dragon Air Cubana Airlines

GOL Iberia Emirates Etihad El Al Qatar

Qantas Luxair Hahn Airlines Japan Airlines Surinam Airways Transportes Aeromar

Malaysian

Table 10-1: Interline Partners of Avianca Airlines 52 As well as reading this section, you may also wish to read the excellent article on interline arrangements by IATA, IATA Interline 2019. 53 Avianca Holdings Interline Agreements 2019.

Chapter 10

252

Interline agreements allow air passengers to travel across the networks of many airlines in one journey using a single reservation. The interline agreement also allows each airline to accept the other’s tickets and covers baggage transfers and liability if the baggage of the passenger has been lost. For example, our passenger who books a ticket on Avianca Airlines once again, has a journey from Bogota to Paris and return which is booked on a single ticket that is emailed to the passenger by Avianca Airline’s reservation systems. To complete this journey, the passenger travels across the networks of two airlines using a single ticket. The single ticket is accepted by Air France on the flight from Barcelona to Paris and return even though the booking is made through Avianca Airlines. Air France knows that the passenger will be arriving in Barcelona from Bogota. In fact, the passenger is likely to receive a boarding pass in Bogota for the flight on Air France between BCN and CDG. Air France will also know that the passenger will have baggage that will need to be checked-in to the Air France flight from Barcelona to Paris, and if the baggage goes missing on the Air France leg, then Air France will be responsible for its recovery or compensation.

10.2.2 Interline Prorating 10.2.2.1 What is Interline Prorating? When a ticket is issued for an interline itinerary, one of the airlines in that itinerary will be selected by the ticketing agent as the issuing airline. The issuing airline collects the entire fare from the customer. This fare must be split between the issuing airline and the operating airline. The process of splitting up the fare so that the operating carrier receives compensation for carrying the passenger and the issuing airline receives compensation for making the sale is called interline prorating. In our example from section 10.2.1, the issuing airline is Avianca Airlines, and the operating airlines are both Avianca Airlines (on the legs between Bogota and Barcelona) and Air France (on the legs between Barcelona and Paris). Avianca Airlines collects the money from the passenger and will forward some of that money to Air France. The interline arrangement between Air France and Avianca Airlines will determine how much of the money Air France will receive for the single payment that the passenger makes to Avianca Airlines.54 Let us now consider another example. We examine the case of a passenger who flies from Port Macquarie, which is a small coastal city that is a 4-hour drive north of Sydney in Australia, to Las Vegas with a ticket issued by Qantas. Qantas (QF) uses its own aircraft to fly the passenger from Port Macquarie to Sydney and then from Sydney to Los Angeles, while American Airlines (AA) uses its aircraft to fly the passenger from Los Angeles to Las Vegas. This journey is described in Figure 10.3 below.

AA

QF

QF

Source: www.gcmap.com

Fig. 10-3: Map of a Trip from Port Macquarie to Las Vegas Via Los Angeles Qantas in this case is the issuing carrier, with the Qantas economy fare in the saver fare class quoted as $1,526 for the entire journey, including the leg from Los Angeles to Las Vegas.55 Qantas is also an operating carrier because it operates the two legs Port Macquarie to Sydney and Sydney to Los Angles. American Airlines operates the beyond leg of Los Angeles to Las Vegas. American Airlines is therefore an operating carrier but not an issuing carrier while Qantas is both an issuing carrier and an operating carrier. Qantas collects the fare of $1,526 from the passenger and will need to split this fare so that it receives a part of it, while the other part is forwarded to American Airlines for the segment of

54

Further examples of interline agreements are presented by Schlappig 2017. The booking date for this fare was April 5, 2015, and the travel date was July 15, 2015. The fare information was obtained from the Qantas booking website, www.qantas.com. 55

Airline Relationships and Business Models

253

the journey that it operates. The components of the fare received by Qantas and American Airlines will be determined as part of the interline agreement between the airlines. There are four ways that the money the issuing airline receives from the passenger is split between the issuing airline and the operating airline under an interline prorate agreement – fare straight prorating, mileage straight prorating, weighted mileage prorating and special prorate agreements. Each of these methods will now be discussed in the following subsections. 10.2.2.2 Fare Straight Prorating The most simplistic approach to splitting fares is referred to as straight prorate. This method splits the fares based on market fares or distance. In the case of fare straight prorating, the fare is split according to the market fares that are paid for each leg of the interline journey. If we assume that there are just two legs in the interline journey, legs 1 and 2, the market fares that are paid for these two journeys are P1 and P2, and we assume that the issuing airline operates leg 1 and the operating-only carrier operates leg 2 then the split of the fare that is paid by the issuing airline to the operating carrier will be: Fare Straight Prorate Share =

୔మ

(10.1)

୔భ ା୔మ

If PFare is the fare that is paid to the issuing carrier, then the share of the fare that is paid by the issuing carrier to the operating carrier is: Fare Straight Prorate =

୔మ ×୔ూ౗౨౛

(10.2)

୔భ ା୔మ

Of course, we can have more complicated settings as well. Let us suppose that there are three legs in the interline journey, and the third leg is operated by an operating-only carrier and the first two legs are operated by the issuing carrier. The fare straight prorate share in this case is: Fare Straight Prorate Share =

୔య

(10.3)

୔భ ା୔మ ା୔య

There may also be multiple operating-only carriers. For example, if there are four legs in the journey and the first two legs are completed by the issuing carrier and legs 3 and 4 are flown by operating-only carriers A and B respectively then the fare straight prorate shares for A and B will be: Fare Straight Prorate Share A = Fare Straight Prorate Share B =

୔య

(10.4)

୔భ ା୔మ ା୔య ା୔ర ୔ర

(10.5)

୔భ ା୔మ ା୔య ା୔ర

If both legs 3 and 4 were operated by the same carrier, then the straight prorate share for the operating carrier would be the sum of (10.4) and (10.5). The market fares, Pi, in each of the scenarios presented above are determined by finding the average fare that a passenger would pay if the passenger were to fly on each of the legs in the journey as a stand-alone service. These market fares are likely to be tabulated in interline agreements between the airlines. Let us show how these formulae work in terms of our Port Macquarie to Los Vegas example from section 10.2.2.1. In this example, Qantas is the issuing carrier, as well as an operating carrier over legs 1 and 2 while American Airlines is the operating-only carrier over leg 3, which is Los Angeles to Las Vegas. As indicated in the interline agreement between Qantas and American Airlines, the market fares are $135 for Port Macquarie to Sydney, $1,001 for Sydney to Los Angeles and $201 for Los Angeles to Las Vegas. The share of the total fare that is paid to American Airlines using the fare straight prorate method described by equation (10.3) is: Fare Straight Prorate % for American Airlines =

ଶ଴ଵ ଶ଴ଵାଵଷହାଵ,଴଴ଵ

= 15%

The invoice that American Airlines sends to Qantas is therefore the Los Angeles to Las Vegas share of market fares of 15% multiplied by the total fare that is paid which is $1,526 as follows: Fare for American Airlines = 0.15 × 1,526 = $229

Chapter 10

254

10.2.2.3 Mileage Straight Prorating In the case of mileage straight prorating, the fare allocated to a sector that is flown by an operating-only carrier is determined based on the distance that is flown by the operating-only carrier relative to the total distance that is flown by the passenger. If we define DOA as the distance that is flown by the operating-only carrier for a journey, which could involve just one leg of flying or multiple legs, and D is the total distance that is flown then the share of the total fare that is allocated to the operating only carrier is: Distance Straight Prorate =

ୈోఽ ×୔

(10.6)

ୈ

Let us now apply formula (10.6) to our Port Macquarie to Los Vegas illustration. In this illustration, the share of the $1,526 paid by the passenger for the full journey that is allocated to American Airlines is determined by dividing the Los Angeles to Las Vegas distance by the total flying distance from Port Macquarie to Las Vegas. The respective distances for each leg of the journey are Port Macquarie to Sydney is 320km, Sydney to Los Angeles is 12,051km and Los Angeles to Las Vegas is 380km. Using formula (10.6) the fare that is allocated to American Airlines for the Los Angeles to Las Vegas leg is: Distance Straight Prorate American Airlines =

ଷ଼଴×ଵ,ହଶ଺ ଷଶ଴ାଵଶ,଴ହଵାଷ଼଴

= $45

(10.7)

The Los Angeles to Las Vegas distance is 3% of the total distance between Port Macquarie and Las Vegas so that the fare allocated to this leg of the journey is just $45 as indicated by (10.7), which is an allocation that is unlikely to be acceptable to American Airlines. 10.2.2.4 Weighted Mileage Straight Prorating Rather than apply the fare prorate in this simple fashion using distance, it is more likely that the fare is prorated using the weighted distance. In this case, relatively short sectors are provided more weight than relatively long sectors. The weighted distances are often referred to as prorate factor kilometres. The justification for providing more weight to relatively short sectors is that cost per available seat kilometre is higher in the case of short sectors compared to long sectors (see Chapter 6, section 6.12 for a discussion of the relationship between non-fuel cost per ASK and the passenger average sector length). This means that operating-only carriers that carry passengers over relatively short sectors will be faced with much higher cost per available seat kilometre than operating carriers that are flying passengers over long sectors. These relatively high-cost sectors should be compensated with a higher prorate. The prorate amount that an operating-only carrier will receive in the case of weighted mileage straight prorating can be described by the following formula: Weighted Distance Prorate =

୵×ୈోఽ ×୔ಷೌೝ೐

(10.8)

ୈ

where w in (10.8) is a number that is greater than 1 that is proportional to the CASK of the short sectors of the journey relatively to the CASK of the long sectors of the journey. In the case of the Port Macquarie to Las Vegas illustration, if it is believed that Los Angeles to Las Vegas cost per available seat kilometre is around 5 times more than the cost per available seat kilometre of the Sydney to Los Angles leg, then the weighted component in (10.8) is w=5. The prorate amount that is paid to American Airlines using the weighted distance prorate is therefore: Weighted Distance Prorate =

ହ×ଷ଼଴×ଵ,ହଶ଺ ଷଶ଴ାଵଶ,଴ହଵାଷ଼଴

= $227

(10.9)

The weighted distance represents 14.9% of the total distance travelled for the journey, generating a prorate for the U.S. leg of $227, which is around the same as that for the fare prorate method. 10.2.2.5 Special Prorate Agreements Some airlines also adopt their own bilateral prorate agreements referred to as special prorate agreements. These agreements simply specify how revenue for a journey involving carriage of a passenger on flights operated by both the issuing and the operating carrier should be divided between the parties. There are many different methodologies for determining the special prorate. The most common is the fixed rate method, which is simply a fixed dollar amount that one carrier pays to another when they carry another airline’s passenger on the route in question. In the case of our Port Macquarie to Las Vegas example once again, the interline agreement between Qantas and American Airlines would specify under a special prorate agreement the exact amount that American Airlines would be paid if it carried a passenger between Los Angeles and Las Vegas on behalf of Qantas. This prorate amount would be factored into the fare that Qantas would charge the passenger for the compete journey from Port Macquarie to Las Vegas.

Airline Relationships and Business Models

255

10.2.3 Advantages and Disadvantages of Interline Agreements The main advantage of interline agreements from an airline’s perspective is that it generates more passengers and revenue for both the issuing and the operating airlines. An issuing airline can sell seats onto a city-pair that it doesn’t operate, and it can boost the number of passengers on services that it does operate. An operating airline can also fly passengers that it would not normally carry because of the existence of an interline agreement. To provide an example of an interline agreement boosting demand, consider a resident of Hong Kong who wishes to fly from HKG to Warsaw Chopin Airport (WAW), Poland. The resident would rather take a direct flight from Hong Kong to Poland but neither the national carrier of Hong Kong, Cathay Pacific, nor the national carrier of Poland, LOT Polish Airlines (LO) fly direct services to Hong Kong because there is insufficient demand for a direct flight. Instead, the resident must choose a more complex itinerary that involves a stopover. As Cathay Pacific has an interline arrangement with LOT Polish Airlines, the passenger can fly with Cathay Pacific to Frankfurt Airport (FRA) on flight CX 289 and then fly with LOT from FRA to WAW on LO 382. This flight can be booked on the Cathay Pacific website and the passenger will receive boarding passes for each flight on check-in at HKG. An alternative to this flight involves the use of a hub carrier like Turkish Airlines. Turkish Airlines can fly the passenger from HKG to Istanbul Airport (IST) on flight TK 71 and then from IST to WAW on flight TK 1265. If Cathay Pacific was not able to offer the full journey from HKG to WAW via FRA, instead only offering the journey from HKG to FRA with the passenger completing the remainder of the journey on another ticket, then Cathay Pacific may have missed out on both the direct flight HKG to FRA, as well as a component of the money earned from the passenger being flown between FRA and WAW. This is because the passenger may choose to fly with Turkish Airways instead. The interline arrangement with LOT has therefore provided a boost to the number of passengers and the revenue earned by Cathay Pacific. Similarly, if LOT did not have an interline agreement with Cathay Pacific, it may have missed out on the passenger demand and revenue generated by flying the passenger between FRA and WAW. Both the issuing airline and the operating airline have benefited. Another key advantage of an interline arrangement is that the fare paid on an interline journey is usually less than the sum of the market fares paid over the individual sectors of the journey. To illustrate, let us consider the Hong Kong to Warsaw example once again. In the case of that journey, Cathay Pacific offers the passenger a single fare for the whole trip from HKG to WAW via FRA, which is US$1,000. If the passenger were instead to buy a ticket from HKG to FRA on Cathay Pacific booking through the Cathay Pacific reservation system and website and then a separate ticket from FRA to WAW on LOT booking through the LOT reservation system and website, which is a booking that does not engage the interline agreement between the airlines, the passenger would have paid US$900 on the HKG to FRA leg and US$400 on the FRA to WAW leg, therefore paying an additional US$300. The reason the interline fare is often lower than the sum of the market fares of the individual sectors flown is because the behind and beyond legs of interline journeys are often priced closer to the marginal cost of the journey rather than priced on the full cost of a seat. The observation that an interline fare for the whole journey is less than the sum of the market fares for the component sectors is particularly true for journeys that involve relatively small regional city airports, where the fare for a trip between a major city hub and a regional airport is typically relatively high because of high costs per available seat kilometre. In our example that involves travel from Port Macquarie to Sydney to Los Angeles to Las Vegas (see Figure 10-3 above), if the passenger were to book a separate ticket on the regional airport to hub sector Port Macquarie to Sydney and a single ticket for the remainder of the journey, the total price of the journey is likely to have been greater than the price paid under an interline arrangement that involves the full journey because the market fares for this regional airport to hub sector are typically very high. Air travellers also benefit from interline agreements from a convenience standpoint. Passengers prefer a one stop shop when it comes to buying airline tickets. They would prefer to buy a ticket for their journey in one place and be issued with a single ticket that covers their entire journey. This is as opposed to buying a ticket separately for each sector that is flown. Returning to our journey presented in Figure 10-3 once again which involves travel between Port Macquarie and Las Vegas, it is conceivable that the passenger could buy three different tickets for this journey, for example, a ticket on Rex Express for the journey from Port Macquarie to Sydney, a ticket on Qantas from Sydney to Los Angeles, and a ticket on Southwest Airlines for the journey from Los Angles to Las Vegas. Having to deal with three different tickets for three different journeys is certainly less convenient than a single ticket purchased on Qantas or Virgin Australia that accounts for travel on all three sectors. Interline agreements also allow passengers to enjoy the benefits of not being concerned about the transfer of their baggage between flights, because interline agreements usually include this transfer. Without such an agreement the passenger would need to collect their baggage at the end of a flight and then check-in that baggage for the next flight, which can be a daunting task for passengers that are not familiar with the destination airport and have limited transfer time between flights. Returning to our Port Macquarie to Las Vegas example once again, without an interline arrangement and travelling on 3 separate tickets the passenger would have to pick up their baggage from the domestic terminal 2 in Sydney, carry that baggage for around a kilometre to the international terminal 1 in Sydney, check-in that baggage at terminal 1 and then rush through quarantine and security inspections at terminal 1 to board the flight to Los Angles. The passenger will arrive at Tom Bradley international terminal in Los Angeles, proceed through passport control and collect baggage at that terminal and transport it to terminal 1 at Los Angles which is a distance exceeding one kilometre, check-in that baggage once again and race to board the flight at the relevant terminal 1 gate. This

256

Chapter 10

experience is likely to be significantly more stressful than that involving a passenger flying on a single ticket under an interline arrangement. Another advantage of an interline agreement is the airline that issues the ticket enjoys a cash flow benefit. This is because the ticket revenue for both the issuing and operating airlines is collected by the issuing carrier. The issuing airline can deposit the ticket revenue into an interest-bearing account until it is time to pay the invoice that has been supplied to it by the operating carrier. In a relatively high interest rate environment, earning interest on revenue received in advance for several weeks can generate a solid stream of earnings for the issuing airline. A key disadvantage of an interline agreement is that it typically results in passengers paying higher airfares than other types of agreements between airlines. While an interline fare may be lower than the sum of the market fares for each sector that is travelled, it is still typically higher than the fare under a codeshare agreement, or a closer form of cooperation such as a joint venture or revenue-sharing agreement. This is largely because the operating carrier will only agree to an interline fare for the sector it operates if it maximises profit for that sector. The operating carrier will not wish to sacrifice earnings by leaving seats aside for interline passengers when they could have sold those seats at a much higher price to passengers not travelling on interline tickets. The operating carrier therefore sets the interline rate by only considering the economics and profitability of that single sector. As a result, they will typically set an interline fare that is lower than the market fare but not as low as the fare they would charge if they were carrying the passenger over several sectors. Returning to our Port Macquarie to Las Vegas example again, the interline fare that American Airlines charges Qantas for the sector Los Angeles to Las Vegas is lower than the market fare for that sector, but it is higher than the implied fare that American would charge if American were to carry the passenger for the entire journey from Port Macquarie to Las Vegas. Over this longer journey, American Airlines would set the Los Angeles to Las Vegas fare component at a level that considers the possibility of losing the passenger to other carriers for the other parts of the journey. Another disadvantage of interline agreements is the quality of the interline product that is being offered. The passenger may book a ticket on an issuing carrier that has an excellent product, but on one or more legs of the journey they may fly on an airline that has a product offering that is lower in quality. This adversely impacts the flying experience of the passenger. It is also the case that the interline arrangement that is in place may not involve baggage transfer. The passenger may be required to check-in and pass-through airport security multiple times in this case. Passengers that need to transfer between flights involving the issuing carrier and the operating-only carrier, may also need to walk a significant distance between the arrival gate and the departure gate. This is as opposed to deeper relationships between airlines, such as codeshare and joint venture relationships, where airlines are in a better position and have stronger motivation to coordinate arrival and departure gates.

10.3 Codeshare 10.3.1 What is a Codeshare Agreement? A codeshare agreement, which we usually simply refer to as a codeshare, is an aviation business arrangement in which one airline, A sells tickets using its own flight number coding onto a flight that is operated by another airline B. This differs from an interline arrangement, in which airline A sells tickets onto a flight operated by airline B using the flight code of airline B. To explain this in more detail let us consider an example. A passenger wishes to fly from Sydney to London Heathrow. The passenger books a flight on the Qantas website, www.qantas.com.au, which means that Qantas is the issuing airline because it issues the ticket. The passenger books flight QF8415 from Sydney to Dubai leaving Sydney at 06:00 and arriving at Dubai at 13:20 and then flies on QF8003 leaving Dubai at 14:30 and arriving at London Heathrow at 18:20. These are flights that bear the Qantas code (QF) but they are actually flown by Emirates (EK). In this case, Qantas has issued a ticket using the QF code on flights operated by Emirates. The flights are therefore Qantas codeshare flights. To see the difference between these codeshare flights and non-codeshare flights our passenger who wishes to fly from Sydney to London Heathrow does so by using an alternative set of flights. The passenger books a ticket on the Qantas website once again, but this time books on the QF1 service from Sydney to London via Singapore. This service is operated by Qantas both between Sydney and Singapore and Singapore to London, with no change in the flight number. In this case Qantas is both the issuing and operating carrier and this is not a codeshare flight. Consider another example of codeshare flights. A German resident wishes to travel from Frankfurt in Germany to Seattle in the U.S.A. The passenger clicks on the Lufthansa website and books flight LH492 from Frankfurt (FRA) to Vancouver (YVR) and flight LH6564 from YVR to Seattle (SEA). The LH492 flight FRA-YVR is issued and operated by Lufthansa, while the LH6565 flight FRA-YVR is issued by Lufthansa but operated by Air Canada Express Jazz. The flight FRA-YVR is a Lufthansa codeshare flight because the ticket for the flight is issued by Lufthansa, the flight is operated by Air Canada Express Jazz and the code used for the flight is that of Lufthansa (LH). The term code refers to the identifier used in the flight schedule. Flight numbers typically include a two-digit airline code, such as QF for Qantas, EK for Emirates and LH for Lufthansa, and a set of numbers that relate to the flight details. In the case of codeshare arrangements, the number component of the flight number for the issuing airline usually consist of four digits. For example, in the case of our Qantas flight between Sydney and London via Dubai that is marketed by Qantas and operated by Emirates, the flight numbers are QF8415 for the Sydney to Dubai leg (which includes the four digits 8415) and QF8005 for the Dubai to London leg (which includes the four digits 8005). In the case of the flight

Airline Relationships and Business Models

257

from Vancouver to Seattle the flight number is LH6565 (which includes the four digits 6565 indicating a codeshare flight).

10.3.2 Types of Codeshare Agreements In this section we consider three popular forms of codeshare arrangements - block-space codeshare, free-flow codeshare, and capped free-flow codeshare. The block-space codeshare is an arrangement in which the issuing airline purchases a fixed number of seats from the operating carrier on a particular route. The issuing airline pays the operating airline a fixed price for those seats, and the seats are kept away or blocked from the operating carrier's seat inventory. The issuing airline uses its own yield management systems to optimise the revenue it generates from the sale of the codeshare seats to passengers. There are generally two types of block codeshare arrangements. The first is a hard block codeshare agreement and the second is a soft block codeshare agreement. Under a hard block codeshare arrangement, if the issuing airline cannot sell all the seats that it has been allocated under the codeshare agreement, they go unsold. The unsold seats cannot be returned to the operating airline, and the issuing airline remains liable to pay the operating airline the fixed price for the seats. When deciding on the number of seats that it wishes to be allocated under the codeshare arrangement, it is therefore very important that the issuing airline understands the demand for the codeshare services by passengers. Overestimation of this demand under a hard block codeshare can be costly for the issuing airline. In the case of a soft block codeshare agreement, the issuing carrier has access to a fixed number of seats once again, however if those seats remain unsold, they are returned to the operating carrier at an agreed number of days prior to departure, enabling the operating carrier to sell those seats. Under a soft block codeshare, the risk that the fixed number of seats that are allocated to the issuing carrier remain unsold falls back on the operating carrier. The fact that the operating carrier retains the risk of the seats remaining unsold indicates that the price the issuing carrier pays for the seats will be higher under the soft block codeshare arrangement than under the hard block codeshare arrangement. Under the free-flow codeshare arrangement, the inventory and reservation systems of the issuing and operating airlines communicate in real-time. The issuing and operating airlines sell seats on a first-come, first-served basis, up to the seat capacity of the operating airline. The scheduled number of seats that are flown is determined by the operating carrier, and the operating carrier bears the costs associated with operating the flight. The operating carrier also has control of the seat inventory and yield management of codeshare flights, which means that it can close seats at the cheap fares and open seats at more expensive fares at its own discretion. The issuing airline must accept these decisions by the operating carrier. In the case of a capped free flow codeshare arrangement, the issuing airline is subject to a cap on the number of seats to which it has access.

10.3.3 Motivations for Codeshare Agreements Airlines form codeshare agreements because they believe that they will enhance the air travel product that they offer for sale to passengers. By enhancing their product, this results in stronger demand and the ability to set prices at higher levels, both of which result in higher levels of revenue and profitability. Table 10-2 below presents examples of the type of product improvements that codeshare agreements can enable. As indicated in Table 10-2, codeshare agreements can improve the product offering of an airline in many ways. The first and most important is by increasing the network offering of the airline. This essentially means that an airline can offer passengers a wider network of destination offerings. The second is that it can reduce the transit time of passengers that are flying on complex itineraries and journeys by coordinating schedules. Thirdly, codeshare partners can offer passengers wider access to airport lounges by allowing the passengers of partner airlines access to their lounges. Codeshare airlines can also improve their frequent flyer program offering by allowing their own airline frequent flyer program members to earn and burn points by flying on the flights operated by codeshare partners. Fourthly, codeshare partners can combine marketing resources to promote the combined demand for their flights. Finally, airlines in codeshare agreements can share ground handling, engineering, airport check-in and customer service resources, thereby saving costs which can be passed on to passengers in the form of lower airfares. Let us explain the contents of Table 10-2 further by using an illustrative example. Qantas Airlines flies passengers from Australia’s east coast (Sydney, Melbourne, and Brisbane) into Los Angeles on flights QF11, QF93 and QF15 respectively – refer to Figure 10-4 below. The Sydney and Melbourne flights arrive in Los Angeles at around 7.00am while the Melbourne flight lands in Los Angeles at around 7pm. Qantas has codeshare agreements with American Airlines and Alaska Airlines. Both airlines have very wide networks in the U.S.A. domestic market, with American flying to 95 domestic destinations and Alaska to 110 domestic destinations. These wide domestic networks in the U.S.A allow Qantas to market flights to a wide variety of destinations in the U.S.A. Three population destinations in the U.S.A. include Denver, Colorado, Orlando, Florida and New York City, New York. A passenger from Brisbane who wishes to fly to Denver can do so by flying on the Brisbane to Los Angeles QF15 service which is operated by Qantas and then the QF4494 codeshare service which is operated by Skywest As American Eagle for American Airlines. The codeshare service to Denver departs at 10am in the morning, giving the passenger at least 3 hours to enjoy the American Airways Admirals Club lounge at Los Angeles terminal 4.

Chapter 10

258

Product-Improving Feature Coordination of routes

Coordination of schedules

Reciprocal Lounge Use Coordination of Frequent Flyer Program

Joint Marketing Resource Sharing

Explanation Airline A1 which is domiciled in country A has passengers that wish to travel to cities in country B that are not serviced by A1. Airline A1 finds a codeshare partner in country B, B1 with a wide network of domestic services within country B that can take A1’s passengers to their desired destination in B. Reciprocally, airline A1 provides a wide range of domestic route options for the passengers brought into country A by codeshare partner B1 who need to on-travel to other domestic destinations within country A. Airline B1 sets the departure time for its domestic flights so that it aligns with the arrival time of the international flights of its codeshare partner A1. The departure times of B1’s flights are set with a view to minimising, as far as possible, the transit time of A1’s passengers who need to on-travel to other destinations in B. Reciprocally, airline A1 sets the departure time for its domestic flights to align with the arrival time of B1 flights in country A. Airline B1 allows passengers of codeshare partner A1 to use B1’s airport lounges when they are in transit in country B airports. Reciprocally, airline A1 allows the passengers of codeshare partner B1 to use A1’s lounges when they in transit in country A. The members of airline A1’s frequent flyer program can earn frequent flyer points by travelling on flights operated by codeshare partner B1 when they are travelling on domestic services within country B. The members of A1’s frequent flyer program can exchange their points to buy seats on flights operated by airline B1 when flying on domestic services within country B. Reciprocally, airline B1 frequent flyer program members can earn and burn points on domestic services operated by airline A1. Airline A1 promotes flights on codeshare partner B1 and B1 promotes flights on A1 on the same routes. When airline A1 lands in country B the ground handling needs of airline A1 are handled by airline B1’s ground handling resources. If any of airline A1’s aircraft require maintenance and repairs while located in country B this can also be handled by airline B1 engineers and maintenance facilities. Airline B1 also checks-in airline A1 passengers for its flights on A1 that depart country B as well as provide gate services. Airline B1 uses the ground handling, engineers and check-in and gate staff of airline A1 when B1 flies into country A.

Table 10-2: Codeshare Partners Product Enhancement

Source: www.gcmap.com

Fig. 10-4: Map of a Trip from Australia’s East Cost to Destinations Beyond LAX in the U.S.A. A passenger from Sydney wishes to fly to New York City. The passenger flies on Qantas flight QF11 to Los Angeles arriving at 7:05am and then codeshare flight QF3664 that is operated by Alaska Airlines which departs LAX at 9:20am. Alaska Airlines flights departing from LAX depart from terminal 6, which is a short underground airside walk from the

Airline Relationships and Business Models

259

Tom Bradley International terminal. Alaska Airlines ground handling crews located at LAX take the passengers baggage and deliver it to the Alaska Airlines flight. The passenger is a member of the Qantas frequent flyer program and earns points by flying on the Alaska Airlines flight. A passenger from Melbourne wishes to fly to Orlando, Florida in the U.S.A. The passenger boards Qantas flight QF93 at 9:25am and arrives in Los Angeles at 6:45pm. The passenger waits at LAX until 10pm which is the next flight to Orlando. This is operated by American Airlines with the Qantas flight code QF3223. This flight departs from terminal 4, which is a very short walk from the Tom Bradley International terminal. The Qantas flight QF93 lands in LAX with some minor mechanical issues which are fixed by the American Airlines engineers that are located at LAX.

10.3.4 Codeshare Agreement Route Types There are three types of routes that are covered by codeshare agreements - parallel operations on trunk routes, unilateral operations on trunk routes, and connections to trunk routes. In the case of parallel operations on trunk routes, also called on-line codeshares, the two codeshare partners both operate on the same sector. For example, both Qantas and American Airlines operate on Sydney to Los Angeles and both codeshare on each other on these flights. Passengers can buy a ticket on the Qantas website to fly on codeshare flight QF4111 which leaves SYD at 9:00am and lands at LAX at 5:45am, which is operated by American Airlines, or they can buy a ticket from the same website to fly on QF11 which leaves SYD at 10:20am and lands at LAX at 7:05am and is operated by Qantas Airways. Singapore Airlines and its codeshare partner Lufthansa both operate services on the Singapore to Frankfurt route. On the Singapore Airlines booking website, a passenger can book a flight on the codeshare service SQ2008 which is operated by Lufthansa leaving Singapore at 23:40 and arriving in Frankfurt at 06:20, while they can also book a flight from the same website on SQ326 which leaves Singapore at 13:55 and arrives in Frankfurt at 20:40 and is operated by Singapore Airlines. In the case of unilateral operation on a trunk route, also called network extension codeshares, only one of the airlines in the codeshare operate services on the trunk route. For example, Emirates Airlines is the only airline operating services between Dubai and Sydney at the time of writing. A codeshare relationship between Qantas and Emirates allows Qantas to sell tickets onto Emirates services between Sydney and Dubai. For example, the QF8415 codeshare service from the Qantas booking website leaves Sydney at 06:00 and arrives in Dubai at 14:10. The most common types of routes covered by codeshare arrangements are beyond and before connections to trunk routes, also called connecting codeshares. In this case carrier A issues a ticket for a multi-leg journey in which A operates at least one of those legs, usually involving a trunk leg, while the codeshare partner B operates the remaining legs and usually involves a before or beyond leg. For example, a passenger from Singapore wishes to travel to Brussels in Belgium. Singapore Airlines will carry the passenger from Singapore to Frankfurt which is the trunk route, and its codeshare partner Lufthansa will carry the passenger from Frankfurt to Brussels (which is the beyond route). As another example, a passenger from Helsinki wishes to travel to Guangzhou in China. The passenger books a ticket on Finnair to travel from Helsinki to Guangzhou via Beijing. Finnair operates the first leg of the flight on the trunk route Helsinki to Beijing, while the beyond leg is operated by Finnair’s codeshare partner Air China.

Quiz 10-1 Interline and Codeshare Relationships 1. Which of the following is NOT typical of an interline arrangement between airlines? (a) It is a commercial agreement that enables the operating carrier to handle passengers on behalf of the issuing carrier. (b) The issuing carrier has its own four-digit code on the flight that is operated by the operating carrier. (c) The issuing and the operating carrier have procedures for settlement of revenue owed to the carrying airline. (d) Passengers pay the carrying airline who then transfers money to the issuing airline. (e) Interline agreements cover baggage transfer. (f) Interline agreements cover transport to and from the airport. (g) Usually in the case of interline agreements the passenger must collect baggage and re-check-in baggage for second and other legs of flights. (h) A passenger travelling on multiple legs with multiple operating carriers travels on a single reservation if all operating carriers have an interline agreement. (i) Under an interline agreement, passengers are charged multiple prices for a journey, one price for each leg flown. 2. On a flight between Auckland and Denver a New Zealand resident books the flight at the Air New Zealand website in economy for a total price of $1,753. The flight consists of two legs – Auckland to Los Angeles operated by Air New Zealand on flight NZ10 and then Los Angeles to Denver operated by United on flight UA383. Air New Zealand and United Airlines have an interline agreement for domestic U.S. services. Use this information to answer the following questions. (a) On the leg Auckland to Los Angeles, which airline is the issuing carrier and which airline is the operating carrier? (b) On the leg Los Angeles to Denver, which airline is the issuing carrier and which airline is the operating carrier? (c) Which airline, United or Air New Zealand, collects the money from the passenger?

Chapter 10

260

(d) Will the passenger need to pick-up her checked-in baggage in Los Angeles and check-in her baggage once again in Los Angeles for the flight to Denver? (e) Which airline has responsibility for the safe arrival of the baggage in Denver? (f) The fare of $2,693 is prorated based on the straight prorate method. The straight prorate method that is used is the fare prorate method. The market fare for the journey Auckland to Los Angeles is $1,650 and the market fare for the journey Los Angeles to Denver is $240. What is the prorate percentage for the Los Angeles to Denver leg? What will be paid to United Airlines by Air New Zealand for the carriage of the passenger between LAX and DEN? (g) Now assume that the airlines straight prorate based on distance. The distance between Auckland and Los Angeles is 10,467km and the distance between Los Angeles and Denver is 1,387km. Determine the invoice that United would send to Air New Zealand if the prorate is based on distance. 3. Tony is flying from Sydney to Orlando one way in business class with a price quote of $5,193 for a date of travel January 16, 2019. The booking is made on the Fiji Airways website, which has indicated that the baggage will be checked-in all the way to Orlando. The itinerary for the flight is as follows: 1:45 PM Sydney Airport (SYD) on Fiji Airways Flight FJ 910 Airbus A330 (313 Seats) Travel time: 3h 55m (3,170 km) 6:40 PM Nadi International Airport (NAN) 27h 35m layover Nadi Overnight Layover 10:15 PM+1 Nadi International Airport (NAN) on Fiji Airways Flight FJ 870 Airbus A330 (313 Seats) Travel time: 10h 30m (8,783 km) Overnight 12:45 PM+1 San Francisco International Airport (SFO)

18h 50m layover San Francisco Overnight Layover

7:35 AM+2 San Francisco International Airport (SFO) Alaska Airlines AS 90 Boeing 737 (159 Seats) Travel time: 5h 10m (3,936 km) 3:45 PM+2 Orlando International Airport (MCO) Use this information to answer the following questions. (a) Which carrier is the issuing or marketing carrier for this flight and which airline is the operating carrier? (b) Is the leg from SFO to MCO likely to involve an interline arrangement between Fiji Airways and Alaska Airlines or a codeshare? Explain your answer. (c) Which airline is likely to receive payment from the passenger for the flight? (d) What are the available seat kilometres per flight for Fiji Airways for the leg SYD-NAN and for the leg NANSFO? (e) What are the available seat kilometres per flight for Alaska Airlines for the leg SFO-MCO? (f) The amount that is paid to Alaska Airlines for carrying the passenger on the SFO-MCO leg is determined using distance prorating. Determine the amount that Alaska Airlines would receive for flying the passenger on this leg. (g) If the passenger were to fly on three separate tickets from SYD to NAN, NAN to SFO and SFC to MCO the market fares that would be paid are $1,046, $3,216, and $927 respectively. What would Alaska Airlines be paid on the SFO to MCO leg if fare prorating were used rather than mileage prorating? (h) What is the basis for using weighted distance prorating? If SFO to MCO received a distance weighting of 3, what would be the new weighted distance prorate received by Alaskan Airlines? 4. Which of the following is true about a codeshare relationship between two carriers? (a) One airline, call it A, sells tickets on a flight operated by another carrier, call it B, with the tickets sold by A using the flight number of B.

Airline Relationships and Business Models

261

(b) One airline, call it A, sells tickets on a flight that is operated by A, with the tickets sold by A using the flight number of another airline B. (c) One airline, call it A, sells tickets on a flight that is operated by A, with the tickets sold by A using the flight number of A. (d) One airline, call it A, sells tickets on a flight that is operated by B, with the tickets sold by A using the flight number of A. 5. What is the major difference between an interline agreement and a codeshare agreement? (a) Under an interline agreement the passenger must collect bags from baggage collection and check-in those bags again for the next leg of the journey, which does not occur under a codeshare. (b) Under an interline agreement the issuing carrier issues tickets under the flight number of the operating carrier, whereas under a codeshare agreement the issuing carrier sells tickets under its own code. (c) Under an interline agreement the issuing carrier is paid by the operating carrier whereas the reverse is true under a codeshare agreement. (d) Under an interline agreement the passenger does not receive a food and beverage service in the air whereas under a codeshare agreement the passenger receives light refreshments. 6. What is the difference between a block space codeshare, a free flow codeshare and a capped free codeshare? Which airline bears the risks of unsold seats 7. Explain the difference between on-line codeshares, network extension codeshares and connecting codeshares.

10.4 Alliances There are three major alliances in international air travel - Star Alliance, SkyTeam, and Oneworld. The biggest alliance in terms of members is the Star Alliance. The Star Alliance was founded in 1997 by a combination of United Airlines, Scandinavian Airlines, Thai Airways, Air Canada, and Lufthansa. The alliance currently has 26 members at the time of writing (December 2020), with those members presented in Table 10-3 below. The airline members of Star Alliance fly Airline Member Aegean Airlines Air Canada Air China Air India Air New Zealand All Nippon Airways Asiana Airlines Austrian Airlines Avianca Brussels Airlines Copa Airlines Croatia Airlines EgyptAir Ethiopian Airlines Eva Air LOT Polish Airlines Lufthansa Scandinavian Airlines Shenzhen Airlines Singapore Airlines South African Airways Swiss Air Lines TAP Airlines Thai Airways Turkish Airlines United Airlines Source: Star Alliance 2021 Table 10-3: Star Alliance Members 2020

Airline Code A3 AC CA AI NZ NH OS OS AV SN CM OU MS ET BR LO LH SK ZH SQ SA LX TP TG TK UA

Year Joined 2010 1997 2007 2014 1999 1999 2003 2000 2012 2009 2012 2004 2008 2011 2013 2003 1997 1997 2012 2000 2006 2006 2005 1997 2008 1997

Chapter 10

262

to more than 1,300 destinations worldwide covering 98% of the world’s countries. The total fleet of its members number 5,033 aircraft, which are used to operate more than 19,000 daily flight departures and carry more than 762 million passengers per year. The next biggest alliance in terms of membership is SkyTeam, with 19 members at the time of writing as indicated in Table 10-4 below. Skyteam is the youngest of the three alliances, formed in 2000 by a combination of Delta Air Lines, Aeroméxico, Air France and Korean Air. SkyTeam members fly to 1,036 destinations across 170 countries. The alliance carries 676 million passengers annually on more than 15,445 daily flights. SkyTeam also has a cargo division called SkyTeam Cargo, which consists of 12 member airlines. Airline Member Aeroflot Aerolineas Argentinas Aeroméxico Air Europa Air France Alitalia China Airlines China Eastern Airlines China Southern Airlines Czech Airlines Delta Air Lines Garuda Indonesia Kenya Airways KLM Korean Air Middle East Airlines Saudia TAROM Vietnam Airlines Xiamen Airlines

Airline Code SU AR AM UX AF AZ CI MU CZ OK DL GA KQ KL KE ME SV RQ VN MF

Year Joined 2006 2012 2000 2007 2000 2000 2011 2011 2007 2001 2000 2014 2004 2004 2000 2014 2012 2010 2011 2012

Source: SkyTeam Alliance 2021

Table 10-4: SkyTeam Members 2020 The third biggest airline alliance in terms of membership is Oneworld with 14 members at the time of writing as indicated in Table 10-5 below. Oneworld was formed in 1999 by a combination of American Airlines, Qantas, British Airways and Cathay Pacific. The alliance has a combined fleet of 3,296 aircraft which is used to fly to 170 countries and service 1,000 airports worldwide. The alliance carries around 500m passengers per year with more than 13,000 daily departures. In addition to the large airlines that are a part of an alliance, there are also many relatively large airlines in global aviation that are not part of an alliance. A list of some of these non-member airlines are presented in Table 10-6 below and includes major full-service carriers such as Emirates Airlines, which had a pre-COVID-19 fleet of 270 aircraft as at the end of March 2019 and carried just under 59m passengers over the 12 months to March 2019, as well as Etihad Airlines, with a fleet of 107 aircraft as at December 2019, and carried 17.5m passengers over calendar 2019. It also includes the world’s biggest low-cost carriers in Southwest Airlines (737 fleet units), Ryanair (479 fleet units) and easyJet (217 fleet units). The most significant benefits of being part of an alliance are that they provide a much deeper network of connectivity, greater service frequency and more varied departure times for passengers travelling across the world and particularly across continents. This in turn adds to passenger convenience and welfare. Passengers often travel internationally on complex itineraries that include multiple stops and multiple operating airlines. If a passenger books with an airline that is a part of a major alliance, each of the operating carriers on the itinerary are usually from the same major alliance. For example, a passenger from Los Angeles who is a member of the Delta Air Lines frequent flyer program, SkyMiles, wishes to book on Delta’s website for a trip to Ho Chi Minh City, Vietnam. Delta Air Line’s reservation system in this case will book the passenger on a Korean Air (KE) flight from Los Angeles to Seoul (Incheon International Airport or ICN) and then book the passenger on a Vietnam Airlines (VN) flight from Seoul to Ho Chi Minh City (SGN). Both Korean Airlines and Vietnam Airlines are a part of the SkyTeam Alliance. By keeping the booking within the alliance, the passenger can book the flight using a single transaction, without the need to visit multiple airline websites. It also increases the chance that the passenger minimises any layover time at airports because the alliance airlines have coordinated schedules (arrival and departure times). The chance that baggage will go missing is also lower as is the chance that there is a lengthy and confusing walk between gates and terminals in the case of connecting flights. If the

Airline Relationships and Business Models

Airline Member Alaska Airlines American Airlines British Airways Cathay Pacific Finnair Iberia Airlines Japan Airlines LATAM Airlines Malaysian Airlines Qantas Airways Qatar Airlines Royal Air Moroc Royal Jordanian Airlines S7 Airlines Sri Lankan Airlines Fiji Airways

Airline Code AS AA BA CX AY IB JL LA MH QF QR AT RJ S7 UL FJ

263

Year Joined 2021 1999 1999 1999 1999 1999 2007 2000 2013 1999 2013 2020 2007 2010 2014 2018

Source: Oneworld Alliance 2021

Table 10-5: Oneworld Members 2020 Air Mauritius (MK) Hainan Airlines (HU) Ryanair (FR) Fiji Airways (FJ) EasyJet (U2)

Air Asia (AK) Jet Airways (9W) Virgin Atlantic (VS) Etihad Airways (EY) Gulf Air (GF)

Cebu Pacific (5J) Lion Air (JT) GOL (G3) Volaris (Y4) Norwegian Air Shuttle (DY)

Emirates (EK) Royal Brunei Airlines (BI) JetBlue (B6) Hawaiian Airlines (HA) Southwest Airlines (NK)

Table 10-6: Airlines Not Part of an Alliance passenger is a premium passenger the alliance will also increase the chance that the passenger can use the airport lounges of alliance partners and allow the passenger to earn and burn frequent flyer points. A key benefit that alliance members enjoy is potentially lower costs. This may come about because airlines within an alliance share resources at the airport. For example, alliance airlines may share the same lounge at airports, and they may use the same check-in, ground handling and catering staff. They may also use the same hangers and equipment for maintenance. These are the same potential cost savings that are available in the case of a codeshare arrangements, but the potential savings are at a much deeper level in the case of alliances because the extent to which resources can be shared covers a much wider network. It is also the case that an alliance can reduce costs by bulk purchasing of goods such as in-flight service items and jet fuel. Alliance airlines also conduct reciprocal safety audits and assist members in dealing with unexpected incidents. Alliances regularly conduct drills and prepare themselves so that they can initiate immediate action to handle emergencies. These actions include setting up customer support, information gathering and linguistic support. One of the potential costs of an alliance is the impact that it may have on non-alliance members, particularly lowcost carriers. Alliances may make it difficult for non-alliance members to compete on routes, which will reduce the variety of competition that passengers can access, including the low-cost fares offered by low-cost carriers. It is also the case that different airlines within an alliance offer different products. This means, for example, that a passenger who books a flight on the Delta Air Lines website may travel on a city pair that is operated by another SkyTeam member (such as Korean Air and Vietnam Airlines in our earlier example). The quality of the product that the passenger experiences on flights operated by these other SkyTeam members may be different from the product quality on Delta Air Lines flights. This may include a different safety standard, a different quality of food and beverage offering, and a different standard of inflight entertainment and seat comfort.

10.5 Revenue Sharing Relationships In the case of a revenue sharing relationship between two airlines, the revenue that the airlines receive from selling tickets is pooled together. The pooled revenue is then allocated to the two airlines according to a formula that determines the revenue share to each of the airlines in the arrangement. The airlines have incentives to add to the revenue pool because they are indifferent between selling tickets on their own aircraft and selling tickets on the aircraft of its revenuesharing partner. The split of the revenue pool for each airline is not typically a 50/50 split because the airlines commit different levels of available seat kilometres to the route. It is also the case that at the start of the revenue-sharing relationship the airlines have different levels of yield (the going-in yield), which means the premium that one carrier has on yield over the partner

Chapter 10

264

going into the revenue-sharing relationship will need to be preserved. The going-in yield of each airline is often referred to as the base yield. The split of the revenue pool allocated to each airline will therefore depend on a combination of the ASKs they are committing to the route as well as the yield premium or discount that they bring onto the route. To elaborate on these points, consider an example that has as its basis the revenue sharing relationship between Virgin Australia and Air New Zealand on its trans-Tasman route flying (which ended in 2016). The trans-Tasman route is the route between Australia and New Zealand. It is referred to as the trans-Tasman route because the sea between Australia and New Zealand is the Tasman Sea. Let us assume that before the revenue sharing relationship between Air New Zealand (NZ) and Virgin Australia (VA) the available seat kilometres of the two airlines were ASKNZ and ASKVA and the revenue per ASK of the two airlines were PNZ and PVA, which are the base yields of the airlines. The pooled revenue of the two airlines that is generated by the revenue sharing relationship is based on total pooled ASKs of ASKP and pooled revenue per ASK of PP. The pooled revenue in the revenue-sharing relationship is the pooled yield multiplied by the pooled ASKs as follows: R = PP u ASKP

(10.10)

If we assume that Air New Zealand’s and Virgin Australia’s ASKs before the revenue sharing relationship were the same as after, this implies the pooled ASKs: ASKP = ASKNZ + ASKVA

(10.11)

The share of pooled revenue that Virgin Australia receives is based on the following ratio: Virgin Revenue Share =

୔౒ఽ ×୅ୗ୏౒ఽ ୔౒ఽ ×୅ୗ୏౒ఽ ା୔ొౖ ×୅ୗ୏ొౖ

(10.12)

We can see that the Virgin share of pooled revenue at (10.12) is equal to Virgin’s pre-revenue sharing RASK multiplied by its post-revenue sharing ASKs, divided by the revenue that both airlines combined would have earned if they sold their ASKs at the pre-revenue sharing RASKs of the two airlines. Similarly, the share of pooled revenue of Air New Zealand is: Air NZ Revenue Share =

୔ొౖ ×୅ୗ୏ొౖ ୔౒ఽ ×୅ୗ୏౒ఽ ା୔ొౖ ×୅ୗ୏ొౖ

(10.13)

The revenue that Virgin Australia receives in the revenue-sharing relationship is equal to the Virgin Australia revenue share at (10.12) multiplied by the pooled revenue at (10.10) and the revenue that Air New Zealand receives is the revenue share of Air New Zealand at (10.13) multiplied by (10.10). Revenue sharing relationships are not common and will require competition authorities to clear them. They are difficult to maintain for long periods of time as the partners in the shared arrangement attempt to maximise the share they gain, which creates instability. Another problem with the revenue-sharing model is that it creates disincentives for unilateral investment in product on the revenue-sharing route. Unilateral investment by one partner airline in product, brand, and the business model, is likely to skew the benefits of the alliance to the non-investing airline if the base RASK of the investing airline is not increased. This follows from the fact that some of the demand, RASK and revenue benefits that are generated from the unilateral investment are shared with the non-investing airline partner. The benefits of the revenue sharing arrangement due to network effects are also likely to be skewed more in favour of one of the partner airlines. The dominant network benefits include access to a wider variety of beyond and before legs, a reduction in transit time, and a greater variety of departure times. For example, in the case of our Virgin Australia and Air New Zealand example, consider a passenger from New Zealand who books on the Air New Zealand website and wishes to travel to a destination within Australia. Because Virgin Australia has a wide domestic network, Air New Zealand can offer the passenger a wide variety of destinations in Australia to which the New Zealand resident can travel with relatively low layover times. Conversely, if a passenger from Australia wishes to travel to New Zealand and books on the Virgin Australian website, this passenger has access to a smaller domestic network offering from Air New Zealand compared to the domestic offering of Virgin Australia. The relative sizes of these network effects are not likely to be reflected in base RASK estimates, in the main part because it is exceptionally difficult to translate these network effects benefits into an adjustment in base RASK.

10.6 Joint Ventures and Partnerships 10.6.1 An Introduction to Airline Joint Ventures Airline joint ventures typically involve a high degree of cooperation between a relatively small number of airlines on specific routes or route groups. The geographic scope and degree of cooperation of different joint ventures varies widely. Joint ventures generally require a far deeper level of cooperation and coordination between the airlines than interline, codeshare, revenue sharing arrangements and alliances. Not only will airlines that have formed a joint venture have

Airline Relationships and Business Models

265

codeshare arrangements in place, but they will also coordinate on prices, connecting services, frequent flyer programs, airport services and lounges. Joint Venture cooperation is designed to maximise the revenue and profitability of the joint venture. The revenues generated by the airlines of the joint venture are generally pooled together, along the same lines as the revenue pooling discussed in the revenue sharing arrangements presented in section 10.5, and then shared between the joint venture members according to an agreed methodology. The deeper cooperation and coordination required by joint ventures generally requires authorisation from local competition authorities. This authorisation is usually granted under the proviso that the joint venture can demonstrate that it will generate significant benefits for passengers including improved flight schedules and discounted airfares. In addition, the joint venture airlines will often be asked by regulatory authorities to make available take-off and landing slots at relevant airports to support the potential entry of new airlines onto the joint venture route. This is to reduce the risk that there is a substantial lessening of competition on joint venture routes. A key benefit to passengers of joint venture partnerships is the ability to offer a wider variety of service frequencies compared to a world without the joint venture. Joint ventures can do this by avoiding wingtip flying. Wingtip flying occurs when two airlines flying on the same city pair schedule services at the same time, or just minutes apart. This is problematic when there are passengers who would prefer to fly at a later or earlier time than the scheduled flights but are forced into flying at the same time because the carriers that fly on the route have decided to fly services at the same time of the day. Joint venture partners can schedule departing flights that are hours apart, thus providing passengers with a greater variety of departure time options. The first joint venture between airlines involved the participants KLM Royal Dutch Airlines and Northwest Airlines back in 1992. There were other relatively small joint ventures shortly after this, such as British Airways and US Airways in 1996, Delta and Swiss Air in 1997, and United Airlines and Lufthansa in 1997, but large trans-Atlantic joint ventures took place much later than this as indicated in Table 10-7 below. The following sub-sections will describe the key details of some of the larger joint ventures that have taken place over the past two decades. Joint Venture Airlines British Airways, Finnair, Iberia, American Airlines, Aer Lingus Delta Air Lines and Virgin Atlantic Qantas and Emirates United Airlines, Swiss Air, Lufthansa, Brussels Airlines, Austrian Airlines, Air Canada KLM, Air France, Alitalia, Delta Air Lines, Virgin Atlantic Lufthansa and All Nippon Airways Lufthansa and Singapore Airlines Lufthansa and Air China Delta Air Lines and LATAM

Date Created 2010

Major Alliance Oneworld

Destinations 433 (Europe and U.S.A.)

2012 2013 2013

N.A. Oneworld Star Alliance

245 (Europe and U.S.A.) 105 (Australia and Europe) 570 (Europe, U.S.A., and Canada)

2015

SkyTeam

2013 2017 2017 2020

Star Alliance Star Alliance

306 (Europe, U.S.A., and Canada) 190 (Europe and Japan) Europe and Southeast Asia Europe and China U.S.A. and Latin America

SkyTeam

Source: SIAPARTNERS 2018

Table 10-7: Examples of Airline Joint Venture

10.6.2 Initial Trans-Atlantic Joint Ventures The Oneworld trans-Atlantic joint venture which began in 2010 involves British Airways, Iberia, Finnair, American Airlines and recently Aer Lingus, all of which are a part of the Oneworld alliance. The joint venture involves collaboration on air transport services across the North Atlantic Ocean for flights between Europe and North America, hence the reference to trans-Atlantic in the name of the joint venture. As indicated in the British Airways media release, British Airways Joint Venture 2021, the joint venture is expected to deliver a much smoother set of connections for passengers, with coordinated schedules allowing more connected onward flights through the major hubs London Heathrow, Chicago O’Hare, Dallas/Fort Worth, Los Angeles, Madrid Barajas, Miami, and New York JFK Airports. At London Heathrow Airport, the three major carriers in the joint venture British Airways, American Airlines and Iberia all operate services from terminal 5, increasing the level of convenience for passengers who are using Heathrow Airport as a transit point for onward flying. All five carriers in the joint venture can take bookings for travel on any of the joint venture airlines allowing passengers to connect with 160 cities in Europe and over 240 cities in the U.S. Passengers that are members of the frequent flyer programs of the joint venture airlines can earn and burn points, request upgrades, and access the premium lounges of all joint venture airlines. The trans-Atlantic joint venture between Delta Air Lines and Virgin Atlantic began in 2012. It initially involved 9 daily, nonstop flights between London and New York and 37 peak-day flights between North America and the U.K. It also included 15 nonstop destinations in the U.S. and 230 one-stop cities across North America. The agreement included

Chapter 10

266

all traffic between the United States, Canada, and Mexico to London, but explicitly excluded Virgin Atlantic traffic destined for the Caribbean. The Delta and Virgin Atlantic agreement permits Delta One or Upper-Class passengers to use the Virgin Atlantic Upper-Class Wing on arrival at Heathrow Airport, with queue-free check-in and security clearance. Virgin Atlantic’s Clubhouse is available to Delta One or Upper-Class passengers and Diamond or Platinum Medallion members when travelling on same day travel on Delta or Virgin Atlantic operated trans-Atlantic flights to or from the U.K. On arrival in London Delta One passengers have access to Virgin Atlantic’s Revivals Lounge. In the U.S., there are 50 Delta Sky Clubs for Delta One and Virgin Atlantic Upper-Class passengers to access, including flagship clubs in Atlanta, New York JFK, and Seattle (Delta Virgin Atlantic JV 2021). Both Delta and Virgin Atlantic operate from terminal 3 at London Heathrow allowing more convenient connectivity between Delta and Virgin Atlantic flights.

10.6.3 Qantas and Emirates Joint Ventures The Qantas and Emirates joint venture was announced as part of a Qantas media release on March 31, 2013 (Qantas Emirates Partnership 2013). When it was first announced by Qantas, the major selling point was that it provided Australians with one-stop access to 65 destinations in the Middle East, North Africa, the UK, and Europe, with 30 of those 65 destinations in Europe alone. Qantas is a carrier with a relatively high proportion of business-purpose traffic which places a high value on minimising total travel time. The joint venture will support this passenger mix as it is estimated to reduce the total travel time from Australia to the top ten destinations in Europe by more than 2 hours. The joint venture will add 98 services per week between Australia and Dubai, many of which will involve both Qantas and Emirates A380s. It will also offer frequent flyer members of both airlines the chance to earn and burn points and provides access to lounges both in Dubai and Australia. Tables 10-8 and 10-9 below provide more detail associated with the way that Emirates and Qantas coordinate on pricing and product as a part of the joint venture.

QF1 QF487 then QF9 QF1 then QF8011 QF8413 then QF9 QF8413 then QF8015

Departure Time 16:00

Arrival Time 06:55

Duration

Aircraft

Seats

Saver Fare

23h 55m

A380-800

$1,364

20:00

13:50

26h 50m

$1,367

$1,967

16:00

07:20

24h 20m

No Seats

$1,955

21:10

13:50

25h 40m

B737-800 then A380-800 A380-800 and then B777-300ER A380-800

$1,474

$1,955

21:10

12:30

24h 20m

14F, 64B, 35PE, 371E 14F, 64B, 35PE, 371E 14F, 64B, 35PE, 371E 14F, 64B, 35PE, 371E (QF only) NA

Flex Fare $1,955

No Seats

$1,955

A380-800 and then B777-300ER

Source: www.qantas.com. Departure date Wednesday July 15, 2015. Booking Date Sunday April 5, 2015

Table 10-8: Qantas Marketed Flights from Sydney to London via Dubai – Economy Cabin Both tables present information about 5 flights from Sydney to London that were available on the Qantas and Emirates websites on April 5, 2015, for a travel date of July 15, 2015. Both tables include the departure times, the arrival times, the duration of the flight, the aircraft type, seat count (where available) and two sets of economy fare class offers – Saver and Flex for Qantas and Economy Flex and Economy Flex Plus in the case of Emirates.56 Table 10-8 includes 3 Qantas booking offers in which there is a codeshare with Emirates, including shared codeshare where Qantas operates one leg and Emirates the other, as well as full codeshare where Emirates operates both legs. Table 10-9 includes two booking offers from Emirates that involve full codeshare and two booking offers that involve codeshare services that are shared with Qantas.

56

Saver fares are generally inflexible fares and Flex fares are flexible fares. Fares are flexible when passengers are permitted to change flight parameters such as the time of flight and the departure date without penalty.

Airline Relationships and Business Models

EK413 then EK029 EK5001 then EK007 EK5001 then EK5101 EK5001 then EK5109 EK413 then EK5109

267

Departure Time

Arrival Time

Duration

Aircraft

Seats

Economy Flex Fare

21:10

14:20

26h 10mins

A380-800

NA

$1,363

Economy Flex Plus Fare $1,954

16:00

07:00+1

24h 0mins

A380-800

4F, 64B, 35PE, 371E (QF)

$1363

$1,9542

16:00

06:55+1

23hrs 55mins

A380-800

14F, 64B, 35PE, 371E

$1,363

$1,954

16:00

00:35+1

30hrs 50mins

A380-800

14F, 64B, 35PE, 371E

$1,363

$1,954

21:10

13:50+1

25hrs 40mins

A380-800

14F, 64B, 35PE, 371E (QF)

$1,473

$1,954

Source: www.emirates.com. Departure date Wednesday July 15, 2015. Booking Date Sunday April 5, 2015

Table 10-9: Emirates Marketed Flights from Sydney to London via Dubai – Economy Cabin A key point to note about the Qantas and Emirates partnership as indicated in both Tables 10-8 and 10-9 is that both airlines operate an equivalent aircraft type for most services, the A380-800, particularly for the long sector between Sydney and Dubai. For some flights between Dubai and London the Boeing 777-300ER is used. By using an equivalent aircraft type, this contributes to the joint venture being able to offer an in-flight product that is as similar as possible across the flying brands, although the decor and the layout of the cabins will be different. Tables 10-8 and 10-9 also indicate how the airline partners attempt to match fare classes as much as possible, with the Qantas Saver class matched against the Emirates Economy Flex class and the Qantas Flex class matched against the Economy Flex Plus class of Emirates. There will also be matches against the fare buckets that are used within these broad fare classes. If we compare the fare offers in the two tables, we note that the Qantas Saver fare is almost identical to the Emirates Economy Flex fare and the Qantas Flex fare is almost identical to the Emirates Economy Flex fare. A key part to any partnership or joint venture will be the extent to which frequent flyers can earn and redeem points on partner services. In the case of the Qantas and Emirates partnership, Qantas FF members can earn and use points when flying on the joint Qantas and Emirates network. Qantas Frequent Flyers as well as travellers in premium cabins have access to Emirates and Qantas Lounges. This includes access to Emirates' First Class and Business Class Lounges in Concourse A in Dubai Terminal 3 and Qantas' First Lounges in Sydney and Melbourne, as well as Qantas and Emirates Lounges in other parts of Australia, Europe, the Middle East, and North Africa. Premium frequent flyers with Qantas can enjoy priority check-in, boarding and baggage delivery when flying on Emirates and vice-versa. Baggage allowance is also a critical issue for passengers. In the case of Qantas marketed passengers, the baggage allowance is the same on Qantas and Emirates (except for sporting goods and infant allowances). Baggage allowances are aligned with Emirates' ticketed checked baggage allowances for all flights between Australia, Europe, the Middle East, North Africa, and Asia. This includes additional Frequent Flyer and Qantas Club allowances. To align with Emirates’ practise, Qantas had to move from a piece and weight system to a weight system, for all international flights excluding those to and from the Americas. Qantas and Emirates recently extended their joint venture, with the joint venture permitted to operate out to 2028 (Qantas and Emirates Extend Joint Venture 2021). This announcement indicates that the joint venture had increased its reach to 100 destinations, including 38 codeshare destinations in the U.K. and Europe, 55 codeshare destinations in Australia and New Zealand, 13 codeshare destinations in Africa and the Middle East, and 2 codeshare destinations in Asia. Since 2013, more than 13 million passengers have travelled on the joint venture, generating more than 87 billion kilometres of travel.

10.6.4 SkyTeam and Star Alliance Trans-Atlantic Joint Ventures The Star Alliance trans-Atlantic joint venture, which involves United Airlines, Air Canada, and the Lufthansa Group of airlines, is one of the largest trans-Atlantic joint ventures, consistent with the fact that it joins together one of the largest airline groups in Europe and one of the biggest airlines in North America and the world. The joint venture serves around 10,000 daily flights and covers 570 different destinations across North and Central America, Europe, the Middle East, Russia, India, and Africa. Like the other joint ventures, it offers to passengers a larger inventory of seats to choose from, more flights to more destinations, a greater selection of non-stop and direct routes and expanded lounge access. For the SkyTeam trans-Atlantic joint venture, which is an extension of the previous joint venture involving Virgin Atlantic and Delta, this covers more than 200 destinations in North America, 6 in the United Kingdom and 100 in

Chapter 10

268

Europe. The joint venture is serviced by 9 daily flights between London and New York, and 39 daily flights between the UK and North America. It also includes more than 300 flights a day between Europe and the U.S. The joint venture therefore represents a significant share of the daily services between Europe and North America and is largely driven by the membership of SkyTeam alliance except for Virgin Atlantic which is not an alliance member.

10.6.5 Asian Joint Ventures For the Singapore Airlines and Lufthansa joint venture, which gained approval from the Competition Commission in Singapore in 2017, this covers flights by the two carriers between Singapore, Australia, Germany, Switzerland, Austria, and Belgium. Singapore Airlines and Lufthansa subsidiaries Swiss Air and Austrian Airlines are a part of the joint venture. It involves revenue sharing across all flights between Singapore and Dusseldorf, Frankfurt, Munich, and Zurich operated by Singapore Airlines, Lufthansa Mainline and Swiss Air. Both Singapore Airlines and Lufthansa operate many of these routes using the Airbus A350, which allows the airlines to synchronise two very good airline products on Europe-Southeast Asia routes (Bright 2017). A joint venture between All Nippon Airways and the Lufthansa Group was set up in June 2013. The joint venture has allowed All Nippon to offer fares to 190 cities in Europe which compares to just 120 cities before the joint venture. All Nippon was able to offer lower fares because of the joint venture by simplifying fare structures. This was achieved by setting up travel zones for European services, with cities charged the same fare if they were in the same zone regardless of the need for transferring between flights. This has contributed to stimulating market demand and improving the market share of the joint venture. Under the joint venture, as opposed to regular codeshare agreements, both All Nippon and Lufthansa have no reason to prioritise sales on their own operating capacity. Both carriers also collaborate on marketing under the shared goal of maximising revenue and profit of the joint venture. Prior to the joint venture, All Nippon had difficulties selling seats into foreign points of sale. With the access that Lufthansa has to European markets, and one of the biggest markets in Europe in Germany, this has provided All Nippon with the opportunity to access new demand through the sales networks and distribution channels of the Lufthansa Group, particularly sales to corporate customers. All Nippon has sent pricing analysts to Lufthansa in Frankfurt, and Lufthansa has sent its pricing analysts to ANA Revenue Management in Tokyo, to form a Joint Pricing Team. The role of the Joint Venture Pricing Team is to maximise the revenue of the joint venture. The Joint Venture Pricing team has been able to offer passengers of both airlines fares that are consistent across the airlines. The joint venture teams share revenue forecasts and marketing data to ensure that the harmonised fares are at levels that maximise revenue and profit for the joint venture. On December 1, 2020, All Nippon and Lufthansa announced a joint venture in the air cargo business. The joint venture benefits customers by providing a larger and faster network with more direct flights, more destinations, and more frequencies. The key target markets in the joint venture will be Frankfurt and Dusseldorf on the Germany side and Tokyo and Nagoya on the Japan side. The added benefit of the joint venture for ANA is that it has access to Lufthansa’s dedicated freighter aircraft which enhances the ability of the airline to send high volumes of freight and cargo between Japan and Germany.

10.6.6 Latin America Joint Venture The trans-America joint venture between Delta Air Lines and LATAM offers a wide range of benefits to passengers. Codeshare agreements between Delta Air Lines and LATAM means that Delta passengers have a wider network of travel in Latin America and LATAM passengers have a wider network of travel in the U.S.A. and beyond. LATAM Pass members and Delta SkyMiles customers will be able to earn and burn points on Delta and LATAM services respectively. Both airlines will share terminals at T4 in New York’s JFK Airport and T3 at São Paulo’s Guarulhos Airport, which will result in more stress-free connections for passengers of both airlines. LATAM Pass members will also be able to access 35 Delta lounges in the U.S. if they qualify, while Delta passengers will have access to 5 lounges in Latin America (Bailey 2021).

Quiz 10-2 Alliances, Revenue Sharing and Joint Ventures 1.

Name the three major global airline alliances in order from biggest to smallest.

2. (a) (b) (c) (d)

Which of the following airlines is NOT a part of the star alliance? Singapore Airlines. All Nippon Airways. Copa Airlines. Delta Air Lines.

3. Which of the following airlines is NOT a part of the SkyTeam alliance? (a) KLM. (b) Garuda Indonesia.

Airline Relationships and Business Models

269

(c) Qantas. (d) Korean Air. 4. (a) (b) (c) (d)

Which of the following airlines is NOT a part of the Oneworld alliance? Japan Airlines. United Airlines. Qantas. Royal Jordanian.

5.

What are the major benefits to an airline of being part of an alliance? Are there any restrictions on airlines due to being part of an alliance?

6.

In the case of a revenue sharing relationship between airlines A and B, the pooled revenue is equal to $300m. The ASKs of the two airlines in the revenue sharing relationship are ASKA = 2,000m and ASKB = 1,000m. The base RASKs of the two airlines are RASKA = 0.09 and RASKB = 0.12. Use this information to determine the share of revenue that is allocated to airlines A and B in the revenue share relationship. Provide a reason why this revenue share relationship between the two airlines may not be fair.

7.

What factors are likely to be present in a joint venture relationship between airlines that are not in a codeshare or interline relationship?

8.

What is the likely impact of the major trans-Atlantic joint ventures on airfares and airline product offering? How would you attempt to quantify these impacts?

10.7 Airline Equity Investments Airline equity investments involve an airline buying a share of another airline. In the case of airlines that are listed on stock exchanges, the equity investment involves an airline buying shares in a listed airline company. In the case of airlines that are privately owned, it involves an airline buying an equity stake in another airline. Most equity investments involve the airline buying shares in a listed airline company. Airline equity investments, like any equity investment, allows an airline to make money from aviation by receiving a dividend income stream from the airline in which it has purchased shares, as well as a capital gain from the sale of the airline shares that it has bought. The capital gain is generated when the airline share price rises above the price at which the shares were bought, and the dividend income stream is earned when the airline shares some of its profit with shareholders as dividends. An airline will typically invest in another airline when it wishes to gain exposure to an aviation market in which it cannot directly operate, or in which it chooses not to operate. For example, Singapore Airlines owned a 20% stake in Virgin Australia before Virgin Australia went into voluntary administration and was bought by the company Bain Capital towards the latter half of 2020. Singapore Airlines acquired a stake in Virgin Australia because it wanted to gain access to the highly profitable Australian domestic market, knowing that it would be more profitable to do this than launching its own domestic operations in Australia, which it is permitted to do by Australian law. Qatar Airways has a 9.99% stake in Cathay Pacific as at June 30, 2021, according to Cathay Pacific’s 6 months to June 30, 2021 interim report while Air China has a stake of around 29% in Cathay Pacific as at the same point in time. Both airlines seek deeper access to the profit that is earned in the Hong Kong air passenger and freight markets (Cathay Pacific Interim Report 2021, 43). Qatar currently flies services into Hong Kong as does Air China, however both airlines have limited access to the Hong Kong and Southern China markets. Qatar is likely to have limited rights to fly aircraft into and out of Hong Kong outside of flights between Qatar and Hong Kong, although it may have rights to use Hong Kong as a technical stop for services between Qatar and other parts of Asia. Air China’s hub is further to the north-east of mainland China at Beijing Capital International Airport (PEK). To gain access to the profits generated from traffic and operations that are in Southern Greater China, including Hong Kong, the airline has decided to invest in an equity stake in Cathay Pacific, rather than invest heavily in operations to the south of the country where airlines such as China Southern possess a dominant share of the market. Qatar Airways also has a large stake in the International Airlines Group, with the airline owning 25.1% of the shares of the company as at May 10, 2020. This investment has allowed Qatar to access the profits that are being made in European aviation markets, with the portfolio of International Airlines Group airlines suggesting that Qatar has strong access to the profits being earned in the U.K., Irish and Spanish aviation markets, which are some of the biggest in Europe. Prior to their bankruptcy, Etihad Airlines bought major stakes in Alitalia, Air Berlin, and Virgin Australia. Alitalia went into bankruptcy in 2017 and was eventually taken over by the Italian Government in 2020 after several interested buyers, including Delta Air Lines and easyJet, lost interest in purchasing the carrier. Etihad invested in Alitalia because of the very strong domestic and international markets for aviation in Italy, and the potential for high returns in those markets. Air Berlin was a major airline in the German market, in fact it was Germany’s second biggest airline behind Lufthansa, until it faced significant financial difficulty and became insolvent in the later part of calendar 2017. Etihad

270

Chapter 10

saw the significant potential in the German domestic and international aviation markets and hence saw Air Berlin as a way of gaining access to the profits that were available in that market. As a result, it bought a small stake in the German carrier in 2011 before lifting that stake to 29% by the start of 2012, making it the biggest shareholder in the airline. In hindsight Air Berlin was a poor investment for Etihad, but one can understand why Etihad chose to pour money into the German carrier. Like Singapore Airlines, Etihad also saw the potential value in investing in Virgin Australia prior to it going into voluntary administration, buying a 20% stake in the company in 2012 and 2013, to give it access to the profits being generated in the Australian domestic market. The portfolio of airline equity investments acquired by Etihad, which also included Air Seychelles and Air Serbia, in addition to the airlines indicated above, was a very different approach to gaining access to the profits being generated in overseas aviation markets compared to other airlines in the Middle East. The Emirates approach, for example, was to grow airline capacity as quickly as possible over as wide a range of markets as possible. The problem that Etihad saw with this approach was that it was becoming increasingly difficult to obtain the rights to send additional seats into these markets, particularly given that various governments around the world were receiving complaints about Emirates dumping capacity in some of these markets and destroying the yields of the local national carriers. By buying an equity stake in national airlines this avoided the problem of needing to constantly lobby foreign governments for the rights to bring additional capacity into their countries. Delta Air Lines owns a 20% stake and Qatar Airlines owns a 10% stake in LATAM airlines as at March 31, 2021. The stake provides both the U.S. carrier Delta and the Middle East carrier Qatar access to the profits that are expected to be earned in the Latin America aviation market, with LATAM airlines having a dominant presence in that market. LATAM has subsidiaries in Brazil, Chile, Peru, Colombia, Paraguay, and Ecuador.

10.8 Mergers and Acquisitions 10.8.1 Mergers An airline merger involves two airlines of roughly equivalent size combining to form a single airline entity. Two airlines merge to expand their network size, to increase market share, to take out a competitor and to reduce costs. Airline mergers are either congeneric or market extension mergers or combination of the two. In the case of a congeneric merger, also known as a product extension merger, two airlines compete on the same routes and often operate services at similar times of the day. They usually operate with similar products and business models, such as two full-service carriers or two low-cost carriers. Two airlines that undertake a congeneric merger will usually either withdrawal one of the airlines from overlapping routes or keep both airlines on overlapping routes but re-schedule services so that they are not flying at the same time of the day and thus not competing against each other for the same time-of-day traffic. For most congeneric mergers there is a reduction in market capacity, with the intention of boosting average airfares and increasing margins and profitability. In the case of a market extension merger, the two merging airlines tend not to compete on the same routes. As a result, they do not compete against one another for the same group of passengers. The intention of the merger is to widen the network of the merged entity. By widening its network, this not only brings into the airline the customers of both airlines, but it may also bring in customers that are new to both airlines because the airline is able to offer services that connect the networks of the merging airlines. A market extension merger is less likely to result in a reduction in capacity, although the merged entity will conduct a review of routes with a view to removing those routes that are not contributing to fixed costs. Airline mergers may also involve a combination of congeneric and market extension features. In this case the two airlines to the merger will have some routes on which they both compete, and other routes in which they don’t. Most mergers involving airlines are with other airlines, which is referred to as a horizontal merger. Airlines may also merge with other companies in the aviation supply chain, however, which is referred to as a vertical merger. This will occur, for example, when an airline merges with a ground handling company, an aircraft manufacturer, an oil refinery, or an airport, however, this rarely occurs. It is more often the case that an airline will acquire these parts to the aviation supply chain rather than merge with them. It is also possible that an airline merges with a component of the tourism supply chain, such as a company that offers accommodation. This can provide benefits to airlines because air travel and accommodation are complementary goods, as discussed at length in chapter 3. Airlines that merge can achieve unit cost reductions, which can occur in many ways. The first and likely most important source of cost reduction involves a reduction in overhead, the most significant of which is the non-operational labour of the merged airlines. The merged airline will not require two CEOs and CFOs and two of each of the critical departments that are a part of airline management, such as treasury, yield management, marketing, security, human resources, and strategy. The airline will need to merge the two sets of departments together and in that process, they will shed labour. The merged airline will also only need a single head office, with a consolidation of the airline’s owned property and property rentals generating overhead cost savings. The merged airline will also be considerably bigger, which means that it has stronger purchasing power in its negotiations with aircraft manufacturers, airports, aircraft leasing companies and oil refineries. This stronger purchasing power can translate into lower prices paid to suppliers, which results in unit cost reduction. The stockholders from both airlines generally retain an interest in the new merged airline. The new merged airline may have a completely different name or brand, it may have a named that is a combination of the two airlines, or it may

Airline Relationships and Business Models

271

adopt the name of the biggest of the merged airlines. The merged entity may continue to fly the same flying brands, as is the case with Air France and KLM, or the merged entity may combine the flying brands into a single flying entity, as is the case with the merger of LAN and TAM into LATAM Airlines. Some of the biggest airline mergers that have occurred over recent years and the name of the merged entity are presented in Table 10-10 below. Year of Merger 2004

Pre-Merger Airlines Air France and KLM

Post-Merger Entity Air France/KLM

Merger Type

Comment

Combined congeneric and market expansion horizontal merger Mostly congeneric horizontal merger

Company still operates Air France and KLM flying brands

2005

American West and US Airways

US Airways

2007

United Airlines and Continental Airlines

United Airlines

Mostly congeneric horizontal merger

2008

Northwest Airlines and Delta Air Lines

Delta Air Lines

Mostly congeneric horizontal merger

2010

British Airways and Iberia Airlines

International Airlines Group

2012

LAN Airways and TAM Airways

LATAM

2013

American Airlines and US Airways

American Airlines Group

Combined congeneric and market expansion horizontal merger Mostly market expansion horizontal merger Mostly congeneric horizontal merger

Company no longer in operation under the name US Airways but merged with American Airlines Name changed to the dominant carrier. Became the biggest carrier in the world by passenger traffic Name changed to the dominant carrier. Became the biggest carrier in the world by passenger traffic Both carriers continue to operate under their own brand names Merged entity renamed under a combined brand name Name changed to the dominant carrier

Table 10-10: Biggest Airline Mergers in Recent Years There are advantages and disadvantages of mergers for the travelling public, for airlines and for airports. Mergers often result in capacity being withdrawn from a route, particularly congeneric horizontal mergers. This is particularly true for routes in which the two merged airlines operate on the same route at similar times of the day. The withdrawal of capacity is usually achieved by reducing the frequency of services. If the airports associated with the routes in which services are withdrawn are slot-constrained airports, the merger may free-up valuable slots for other airlines that wish to operate more services to the slot-constrained airports. By reducing some of the pressure at slot-constrained airports, the withdrawal of services may also lead to lower levels of congestion at the airports. Lower levels of congestion will lead to better on-time performance, a reduction in the number of services that are cancelled and fewer incidences of miss-handled baggage. It will also lead to a reduction in the number of aircraft that must circle an airport, therefore burning extra fuel, because they are waiting for a landing slot. Another potential benefit of a merger between airlines that leads to a reduction in the frequency of services, is an improvement in asset utilisation which leads to a reduction in the unit cost of the merged entity. This may come about because the merged entity is able to carry the same or similar number of passengers for fewer aircraft movements. By reducing the number of aircraft movements this reduces airline operating costs. If these lower operating costs are spread over the same number of passengers, this in turn leads to lower airline cost per passenger. The frequencies that are withdrawn from routes may also be used more profitably somewhere else in the merged airlines network, thus leading to higher merged airline profits. Another potential benefit of a merger is a reduction in wingtip flying. If the two airlines to the merger compete on the same route prior to the merger, they will often schedule their departures at the same time, or very similar times prior to the merger. This is because both airlines wish to schedule their departures at the most popular departure times of the day. This may be during the morning and afternoon peaks such as 8am or 4pm on weekdays to capture the commuter market, or on Friday afternoons and Monday mornings to capture the weekend getaway market. Airlines also schedule departure times for long haul flights so that that the arrival time is convenient for passengers, or the arrival time occurs outside airport curfews. As an example of pre-merger airlines setting the same departure times on routes, consider the London (LHR, LGW) to Madrid (MAD) market. In the pre-merger world, this city pair was operated by both British Airways (BA) and Iberia Airlines (IB). Both airlines operated services at 7.20am and 11.30am daily prior to the merger. To avoid wingtip flying by the airlines that are a part of the merger, the airline will schedule one of the merged airlines to fly at an earlier or later time. This can be of benefit to passengers who are more flexible with their departure times because it provides them with greater departure time choice. Alternatively, the merged airlines may simply withdrawal frequencies from the market to avoid the airlines flying at the same time. This is precisely what happened

Chapter 10

272

in the case of the International Airlines Group merger between British Airways and Iberia on the London to Madrid route. After the merger, the International Airlines Group reduced capacity by 50% by operating only the BA service at 7.20am and only operating the Iberia service at 11.30am. Not only was there a reduction in capacity, but there was also a reduction in competition for flights at these different times. For those passengers that prefer to fly at 7.20am there was now only one carrier to choose from at that time, and for passengers that prefer to fly at 11.30am there is now also just one carrier available where before the merger there were two. The most important rationale, however for an airline merger is to reduce capacity and increases fares. It is most likely to reduce capacity on those routes in which the two pre-merger carriers operate overlapping services, or services that depart at similar times of the day on the same route. The effect of lower levels of capacity in the post-merger world is to unambiguously raise average airfares. One of the ways we can demonstrate this is by analysing the impact on average airfares if the number of competitors is reduced on a route. We saw in chapter 9, equation (9.80) that the equilibrium average airfare in the case of an n-player market depends on the number of players in the market, the average marginal cost, and the price elasticity of demand as repeated below for your convenience: P‫= כ‬

୬கୡത

(10.14)

୬கାଵ

If we move from an n-player market to an n-1 player market the average airfare rises to: P ‫= ככ‬

(୬ିଵ)கୡത

(10.15)

(୬ିଵ)கାଵ

After some manipulation, the increase in the average airfare due to the merger, which is (10.15) minus (10.14), becomes: P ‫ ככ‬െ P ‫= כ‬

ିகୡത [(୬ିଵ)கାଵ](୬கାଵ)

>0

(10.16)

Equation (10.16) is greater than zero because nH + 1 and (n-1)H + 1 are both less than zero according to the first order conditions for profit maximisation of the n-player Cournot model. The percentage increase in the average airfare is (10.16) divided by (10.14) which gives: ୔‫כ୔ି ככ‬ ୔‫כ‬

=

ିଵ [(୬ିଵ)கାଵ]௡

>0

(10.17)

The percentage increase in the average airfare therefore depends on the number of players in the pre-merger market, and the price elasticity of demand. The more concentrated is a market before the merger, and the less price elastic is air travel demand, the higher will be the percentage increase in the average airfare. For example, if before the merger a route was a five-player market, or n = 5, the airfare elasticity of air travel demand is H = -1.25 and the average industry marginal cost is ܿҧ = 100 then the average airfare after the merger is expected to increase by: ୔‫כ୔ି ככ‬ ୔‫כ‬

=

ଵ.ଶହ (ଵିସ×ଵ.ଶହ)(ଵିହ×ଵ.ଶହ)ହ

= 1.19%

(10.17)

If the market was a four-player market before the merger, the increase in the average airfare after the merger becomes: ୔‫כ୔ି ככ‬ ୔‫כ‬

=

ଵ.ଶହ×ଵ଴଴ [ଵିଷ×ଵ.ଶହ](ଵିସ×ଵ.ଶହ)ସ

= 2.84%

(10.18)

Finally, if the market was a three-player market prior to the merger, then the increase in the average airfare after the merger is: ୔‫כ୔ି ככ‬ ୔‫כ‬

=

ଵ.ଶହ×ଵ଴଴ [ଵିଶ×ଵ.ଶହ](ଵିଷ×ଵ.ଶହ)ଷ

= 10.10%

(10.19)

The extent to which the average airfare increases therefore accelerates as the number of pre-merger airlines reduces from 5, to 4 to 3 to 2 airlines. In other words, the more concentrated is the airline market pre-merger the greater is the extent of the average airfare increase that is expected. If we also assume that the price elasticity of air travel demand is less elastic at H = -1.1 then the percentage increase in the average airfare at (10.19) becomes: ୔‫כ୔ି ככ‬ ୔‫כ‬

=

ଵ.ଵ×ଵ଴଴ [ଵିଶ×ଵ.ଵ](ଵିଷ×ଵ.ଵ)ଷ

= 13.29%

(10.19)

Airline Relationships and Business Models

273

The extent to which average airfares increase in response to a merger which reduces the number of competitors from 3 to 2 is higher with airfares increasing by 13.29% at an elasticity of H = -1.1, compared to an increase in airfares of 10.1% at an elasticity of H = -1.25.

10.8.2 Acquisitions Airline acquisitions involve an airline obtaining a majority stake in the acquired airline. Most airline acquisitions involve large airlines buying small airlines that are facing financial difficulty. The biggest such airline acquisitions over the past decade includes the Air Canada acquisition of Canadian Airlines in 2000. In this case, Canadian Airline’s systems and employees fully integrated into Air Canada by October 2000, shortly after which Air Canada significantly cut the number of airline employees (Loh 2020). Another significant acquisition includes the Cathay Pacific acquisition of Dragon Air, which took place in 2006. This acquisition allowed Cathay Pacific greater access to the mainland China market and eventually resulted in a 37% reduction in the number of Dragonair staff (Garger 2006). We also saw Lufthansa acquire Swiss International Airlines in 2005 costing the airline €310m, with the airline stating that the acquisition would allow it to expand its international network and boost its competitive position in Europe (dw.com 2005). Lufthansa also acquired Austrian Airlines in 2008 for €220m, with Lufthansa indicating that it hoped to turn the financially troubled airline around and to focus on Eastern Europe and the Middle East markets (dw.com 2009). Lufthansa also acquired Eurowings and Germanwings in 2005 (European Commission 2005), with the European Commission stating that it was satisfied that the acquisition would not substantially lessen competition, particularly in the low-cost carrier segment of the market. In the Americas, Southwest Airlines acquired Air Tran Airways in 2010, which gave Southwest critical gates at LaGuardia and Atlanta Hartfield Jackson Airports and prepared the airline for expansion into international markets (Southwest News Release 2011). American Airlines acquired Transworld Airlines in 2001 just months before the September 11, 2001, terror attacks in a deal valued at around US$500m (Wall Street Journal 2001). Alaska Airlines bought Virgin America in 2016, establishing Alaska as the West Coast’s dominant carrier with bases from Seattle to San Diego (Alaska Airlines News Release 2016). Finally, Virgin Australia acquired Tiger Airways Australia in 2011 for the bargain price of just one Australian dollar, which allowed it deeper access to the price-sensitive, low-cost segment of the Australian domestic aviation market (The Guardian 2014). Like mergers, the main rationale for an acquisition is to take out a financially weak competitor, or alternatively to enable the airline almost immediate access to a much wider route network. By taking out a competitor this lifts fares and margins, potentially converting unprofitable routes into sustainably profitable routes.

Quiz 10-3 Airline Equity Investments, Mergers and Acquisitions 1. (a) (b) (c) (d)

Which of the following represents an airline equity investment? Airline A allows airline B to sell seats on the services it operates. Airline A buys shares in the listed airline B. Airline A lends money to airline B. Airline A buys corporate bonds issued by airline B.

2. (a) (b) (c) (d)

What return would an airline receive from an airline equity investment? An increase in demand for its product. Codeshare and interline revenue. Dividend payments and capital gain on airline shares bought and sold. A share of the revenue of the airline.

3.

Why would an airline prefer an equity investment over an investment of operating aircraft in a market?

4.

Name two major airline mergers and two major airline acquisitions over the past two decades.

5.

Assume that the price elasticity of demand on a route is -1.5 and the average marginal passenger cost on the route is $150. Estimate the change in the average airfare when a merger or an acquisition causes a change in the number of competitors on a route from n=5 to n=4, from n=4 to n=3, from n=3 to n=2 and from n=2 to n=1.

10.9 Low-Cost Versus Full-service Carrier Business Models 10.9.1 List of Low-Cost Carriers by Region Tables 10-11 through to 10-13 provide a list of the low-cost airlines currently operating in the Asia Pacific, Europe, the Middle East, Africa, and the Americas aviation markets at the time of writing.

Chapter 10

274

Region

Low-Cost Carrier (Country) Hong Kong Express Airlines (Hong Kong) Air Busan (South Korea) Jeju Air (South Korea) Jin Air (South Korea) T’way Airlines (South Korea) Spring Airlines (China) China United Airlines (China) Air Do (Japan) Jetstar Japan (Japan) Peach (Japan) Skymark Airlines (Japan) Solaseed Air (Japan) Starflyer (Japan) Tigerair (Taiwan)

Flight Code UO BX 7C LJ TW 9C KN HD GK MM BC 6J 7G IT

Fleet Units in Service as at October 2021 3 A320 8 A320, 15 A321 26 B737 19 B737 23 B737 96 A320, 8 A321 52 B737 8 B737, 4 B767 17 A320 31 A320 27 B737 14 B737 11 A320 8 A320

North-East Asia

South-East Asia

Air Asia Malaysia Air Asia Thailand Air Asia Indonesia Jetstar Asia (Singapore) Scoot (Singapore) Nok Air (Thailand) Thai Lion Air (Thailand) Citilink (Indonesia) Cebu Pacific (Philippines) Golden Myanmar Airlines Air Asia X (Malaysia) Thai AirAsia X

AK FD QZ 3K TR DD SL QG 5J Y5 D7 XJ

22 A320, 2 A321 16 A320 5 A320 5 A320 12 A320, 4 A321, 14 B787 8 B737, 2 DHC-8 1 A330, 6 B737 6 ATR, 37 A320, 1 B737 11 ATR, 14 A320, 9 A321, 5 A330 3 ATR 4 A330 4 A330

South Asia

AirAsia India Air India Express (India) IndiGo (India) SpiceJet (India) Jetstar Airways (Australia)

I5 IX 6E SG JQ

26 A320 B737 23 31 ATR, 164 A320, 45 A321 35 B737, 25 DHC-8 39 A320, 2 A321, 4 B787

Oceania Source: Planespotters 2021.

Table 10-11: Low-Cost Carriers in Asia and Oceania Some of these low-cost carriers are the biggest airlines and names in global aviation which you are likely to recognise immediately, such as Ryanair (Table 10-12), easyJet (Table 10-12), Southwest (Table 10-13) and Air Asia (Table 1011). Between them, these airlines have around 1,350 fleet units. Many of the carriers in Tables 10-11 through to Table 10-13, however, you will not recognise. In fact, by the time you read this chapter in this book some of these more obscure low-cost carriers may not exist, such is the rate at which these carriers enter and exit aviation markets all around the world. Some of the lesser-known, smaller low-cost carriers include Air Busan (Table 10-11) which is South Korea’s third biggest low-cost carrier with a fleet of 23 A320/21 aircraft (Air Busan About Us 2021), and Blue Air (Table 1012) which is another small, lesser-known low-cost carrier from Romania, with a fleet of 11 Boeing 737-700/800 aircraft (Blue Air About Us 2021) just to name 2. Tables 10-11 to 10-13 indicate that there are many low-cost carriers in the global aviation community. While these carriers have many things in common, such as the strive to be the lowest cost carrier on routes and unbundling services, there are also product features of each low-cost airline that will generate some degree of product differentiation. As indicated in section 10.9.2 below, this generates a spectrum of low-cost carrier types. We will elaborate on this spectrum in the analysis below.

10.9.2 Low-Cost Airline Spectrum and Key Characteristics There are a wide variety of low-cost carriers operating in the global aviation market. Some airlines have many of the typical features of a low-cost carrier, while others will only possess a handful of such features. This means that there is a wide spectrum of low-cost carriers based on the number of characteristics that the carrier possesses. Figure 10-5 below provides an illustration of such a spectrum for the major (not all) low-cost carriers.

Airline Relationships and Business Models

Country Czech Republic France Hungary Ireland Italy Netherlands Norway Romania Russia Spain

Turkey Switzerland United Kingdom Kuwait Saudi Arabia UAE Yemen Egypt Kenya Morocco

Low-Cost Carrier Smartwings HOP! Transavia.com France Wizz Air Ryanair Group Blue Panorama Transavia.com Corendon Dutch Norwegian Air Shuttle Blue Air Pobeda Iberia Express Volotea Vueling LEVEL AnadoluJet Corendon Airlines Pegasus Airlines easyJet Switzerland EasyJet Jet2.com Jazeera Airways Flynas Air Arabia Flydubai Felix Airways Air Arabia Egypt Jambojet Air Arabia Moroc

Flight Code QS A5 TO W6 FR BV HV CD DY 0B DP I2 V7 VY IB TK XC PC DS U2 LS J9 XY G9 FZ FO E5 JM 3O

Fleet Units in Service as at October 2021 27 B737 14 CRJ, 30 ERJ 46 B737 61 A320, 56 A321, 1 A330 458 B737 3 B737 39 B737 3 B737 49 B737 11 B737 42 B737 10 A320, 8 A321 19 A319, 14 A320 5 A319, 85 A320, 13 A321 5 A330 49 B737 16 B737 53 A320, 7 A321, 17 B737 5 A319, 22 A320 33 A319, 90 A320, 10 A321 1 A321, 73 B737, 4 B757 14 A320 31 A320 29 A320, 4 A321 52 B737 1 B737, 3 CRJ 4 A320 6 DHC-8 9 A320

Source: Planespotters 2021.

Table 10-12: Low-Cost Carriers in Europe, the Middle East and Africa Country Canada Mexico United States

Brazil Chile Colombia

Low-Cost Carrier (Country) Air Canada Rouge WestJet VivaAerobus Volaris Allegiant Air Frontier Airlines JetBlue Southwest Airlines Spirit Airlines Sun country Airlines Azul Brazilian Airlines Gol Transportes Aeroes Avianca Sky Airline EasyFly Viva Air Colombia

Flight Code RV WS VB Y4 G4 F9 B6 WN NK SY AD G3 AV H2 VE VH

Source: Planespotters 2021.

Table 10-13: Low-Cost Carriers in the Americas

275

Fleet Units in Service as at October 2021 2 A319, 1 A320, 7 A321 82 B737, 6 B787 38 A320, 10 A321 3 A319, 69 A320, 16 A321 29 A319, 71 A320 90 A320, 20 A321 6 A220, 116 A320, 81 A321, 60 ERJ 679 B737 28 A319, 105 A320, 30 A321 45 B737 32 ATR, 40 A320, 4 A321, 8 A330, 2 B737, 55 EMB 85 B737 11 A319, 45 A320, 5 A330, 5 B787 18 A320, 2 A321 27 ATR 19 A320

276

easyJet Pegasus

Tigerair

SpiceJet

Scoot

Indigo

Allegiant

Fig. 10-5: Illustrative Spectrum of Main Low-Cost Carriers

Southwest

Cebu

Frontier

Wizz

x Lowest unit cost and airfares in the world. x Flies mainly to secondary airports. x Common fleet of Aircraft. x Flies only economy class. x High ancillary revenue & significant fare unbundling. x Offers only intracontinent travel. x Flying outside the peak. x No frequent flyer scheme. x Seat dense configurations.

Ryanair

Volaris

Transavia

JetBlue

Vueling

Air Asia Spirit

Many low-cost carrier characteristics WestJet

Peach

GOL

/

Skymark

Norwegian

Jetstar

x Relatively high unit cost but still low compared to full-service carriers. x Flies to many primary airports. x A variety of fleet types. x Flies outside Europe. x Offers a frequent flyer program. x Flies during the peak. x Flies a relatively high percentage of business-purpose traffic.

Eurowings

Few low-cost carrier characteristics

Airline Relationships and Business Models

277

Probably the most important characteristic of a low-cost carrier is the extent to which it flies to secondary airports. Secondary airports are airports that are smaller compared to primary airports and a longer distance from major cities and population centres. These airports are typically less convenient for passengers because they must travel a relatively long distance from the airport to their ultimate holiday destination. Secondary airports are an attractive proposition for lowcost carriers because they usually offer those carriers much lower airport charges than primary airports. Airport charges represent a relatively high proportion of the cost base of low-cost carriers. If a low-cost carrier wishes to lower its operating costs, then airport charges are usually high on the radar in terms of a category of airline cost which it will target for cutting if it can. Examples of secondary airports that low-cost carriers fly to rather than key primary airports are presented in Table 10-14 below. Primary Airport England London Heathrow

Secondary Airport

Used by the Following LCCs

London Stansted

London Heathrow Manchester France Charles De Gaulle Germany Frankfurt Düsseldorf

Luton Liverpool

Ryanair, AnadoluJet, easyJet, Jet2.com, Pegasus, Ryanair Blue Air, easyJet, Ryanair, Wizz Air Blue Air, easyJet, Ryanair, Wizz Air

Beauvais-Tille

Blue Air, Ryanair,

Frankfurt-Hahn Cologne Bonn

Düsseldorf Munich

Dortmund Memmingen

Wizz Air, Ryanair Air Arabia, AnadoluJet, Blue Air, Eurowings, Pegasus, Ryanair, Wizz Air easyJet, Eurowings, Ryanair, Wizz Air, Corendon, Ryanair, Wizz Air, Pobeda

Milan Malpensa Milan Malpensa Fiumicino

Milan Bergamo (Orio al Serio) Milan Linata Ciampino Airport

Air Arabia, Blue Air, Pegasus, Pobeda, Ryanair, Wizz Air, SkyUp Blue Air, easyJet, Blue Air, Ryanair, Wizz Air

Barcelona Barcelona

Girona Costa Brava Reus

Pobeda, Ryanair, Transavia, Jet2 Jet2, Ryanair

Stockholm Skavsta Airport

Ryanair, Wizz Air

Avalon Airport Gold Coast

Air Asia X, Jetstar Jetstar, Scoot, Air Asia X

Low-Cost Carrier Terminal Kuala Lumpur

Air Asia

Don Mueang Airport

Air Asia, Scoot, Lion Air, Tigerair

Oakland Dallas Love Field William P. Hobby Chicago-Midway Norman Y. Mineta San Jose International

Southwest, Allegiant, Volaris, Spirit, Frontier Southwest Southwest, Allegiant Southwest, Allegiant, Volaris Frontier, JetBlue, Southwest, Volaris

Italy

Spain Sweden Stockholm Arlanda Australia Melbourne Tullamarine Brisbane Malaysia Kuala Lumpur Thailand Bangkok Suvarnabhumi United States San Francisco Dallas/Fort Worth Houston International O’Hare International San Francisco

Table 10-14: Secondary Airports Used by LCCs As an example of low-cost carriers flying into secondary airports rather than primary airports, Ryanair, easyJet and Wizz Air fly into Luton or Stansted Airports rather than flying into the major London airport Heathrow International. Rather than flying into the busy Frankfurt Airport airlines such as Wizz Air and Ryanair fly into Frankfurt-Hahn Airport, which is 123.9km from the city centre of Frankfurt compared to Frankfurt Airport which is 13.1km from the city centre. Finally, rather than flying into San Francisco International Airport, airlines such as Southwest, Allegiant, Volaris, Spirit and Frontier fly into Oakland International Airport. Oakland Airport is 18.4 miles from the city centre compared to San Francisco Airport which is 12.7 miles from the city centre. Another key characteristic of low-cost carriers is price unbundling. This involves setting prices for component parts of the service provided by the airline. These component services include but aren’t limited to the ticket itself, checked

278

Chapter 10

baggage, extra baggage, extra leg room, priority boarding and exiting, exit row seating, inflight entertainment, on-board food and beverages, travel insurance, hotels and other accommodation, car hire and other land transport, and payment methods (credit or debit card). This is as opposed to a full-service airline which tends to include many of these services in the airfares that it offers to the flying public. Low-cost carriers also target lower costs and thus the cheapest airfares possible. This can be seen in the slogans of the various low-cost carriers, including Jetstar, which uses the slogan ‘All day, every day, low fares’, Ryanair which uses ‘More Choice, Lower Fare, Great Care’, Southwest uses ‘Low fares. Nothing to hide. That’s TransFarency’, Wizz Air with ‘Now We Can All Fly’ and Air Asia uses ‘Now Everyone Can Fly’. Low-cost carriers also offer fewer fare options and often have unallocated seating. They also tend to fly during off-peak periods because their travelling demographic is less concerned about flying within business hours. Low-cost carriers also attempt to cut labour costs as far as possible, which means they offer predominantly kiosk or machine check-in at the airport or on-line check-in. For the same reason they also pursue the shortest possible turnaround times leading to better aircraft utilisation or fleet productivity. This in turn requires tight rules for time of check-in prior to departure to ensure that the aircraft departs and arrives on time. To reduce costs, low-cost carriers also fly a uniform fleet, which reduces costs associated with fleet variety such as simulator training costs, cabin crew training costs and spare part costs. For example, Ryanair only flies Boeing 737 aircraft and easyJet only flies Airbus A320 aircraft as indicated in Table 10-13 above. The aircraft they fly usually have a single class of travel and the seats do not recline to fit in more passengers. To reduce the weight of the aircraft, which in turn reduces fuel burn and costs, low-cost carriers will often take the seat pocket from all their seats, they will take out all window shades and remove all inflight entertainment systems. In the pursuit of lower costs, the airline will also request that their cabin crew perform multiple roles, including check-in plus gate assistants. The earlier forms of the low-cost carrier flew relatively old aircraft because this resulted in lower depreciation and lease costs. In more recent times, however, low-cost carriers have more often chosen to fly newer aircraft because this results in lower maintenance costs and improved reliability. Low-cost carriers also tend not to offer frequent flyer programs. They tend to fly relatively short sectors, although in recent times we have seen some growth in long sector flying by low-cost carriers, such as LEVEL, Scoot, Jetstar, and Air Asia X. In Tables 10-11 to 10-13 you will see in the last column of these tables that most low-cost carriers use the short-haul aircraft the A320/21 and the Boeing B737, with a limited number of airlines indicated above flying the medium to long-haul aircraft such as the Airbus A330 and Boeing B787-800. The demand demographic of the low-cost carrier is the price sensitive leisure traveller. Most tickets that are sold are inflexible tickets because these are the lowest cost tickets. To capture a large share of the price-sensitive market lowcost carriers must have low unit costs so they can afford low price offerings.

10.9.3 LCC Versus FSA Cost Categories This section answers the question: which airline cost categories are likely to be materially different between low-cost and full-service airlines? It is probably easier to answer this question by starting with the cost categories that are similar, with the remaining cost categories those in which the costs can be different, often materially so. Fuel costs are likely to be reasonably similar for both full-service and low-cost airlines. This is because they use the same type of fuel, jet kerosene, they generally obtain it from the same source and essentially pay the same price. What may drive some difference in fuel costs is the fuel productivity of the aircraft that are used by the two sets of airlines. If a low-cost carrier chooses to use an old aircraft technology, such as older Boeing 737-800 or Airbus A320 aircraft, this may add to fuel consumption per available seat kilometre compared to a full-service carrier using newer technology for the same aircraft. For the same age aircraft, aircraft depreciation is likely to be reasonably similar between full-service and low-cost airlines because they essentially buy their aircraft from the same manufacturers (largely Boeing and Airbus) and pay for that aircraft at similar prices (unless they receive a large purchase discount for the purchase of the same aircraft type). There is likely to be a difference between the aircraft depreciation costs of low-cost carriers and full-service airlines if low-cost carriers choose to use older vintage aircraft. Aircraft operating lease costs are also likely to be reasonably similar between low-cost carriers and full-service airlines mainly because they are likely to be leased from the same lessors, a few of which tend to dominate global aircraft leasing markets, such as AeroCap, GECAS and Air Lease Corporation. Finance or interest expenses are likely to be similar between low-cost carriers and full-service airlines as airlines essentially source their finance from the same banks and institutions and pay reasonably equivalent interest rates. Finance expenses may differ across airline types to the extent that full-service airlines may use equity as a source of credit more than debt, or bigger, more diverse full-service carriers enjoy a stronger credit rating which in turn lowers the rate of interest that they pay. Maintenance, repair, and overhaul (MRO) costs may be reasonably equivalent because all large commercial aircraft must have the same A, C and D checks performed. Maintenance costs may differ between low-cost carriers and fullservice airlines to the extent that low-cost carriers operate a fleet with an older average aircraft age, which require a greater level of maintenance activity and likely to break-down more often. It is also conceivable that low-cost carriers may be able to secure cheaper maintenance labour, which may drive their maintenance costs lower than that of fullservice airlines. Route navigation and overfly charges are likely to be similar between low-cost carriers and full-service

Airline Relationships and Business Models

279

airlines because both low-cost and full-service airlines tend to fly similar flight paths, including flying over the same air space and communicating with the same air traffic control services. The main cost categories outside of fuel, MRO, aircraft depreciation and operating lease costs are manpower costs and airport landing fees. Manpower costs can be significantly different between low-cost and full-service airlines. Lowcost airlines pay their pilots and cabin crew significantly lower wage rates compared to full-service airlines for flying the same city pairs. It is also the case that labour contracts negotiated by low-cost airlines will generally involve more labour hours and poorer conditions than those negotiated by full-service airlines because low-cost airlines have a greater opportunity to get around the need to negotiate with unions. As a result, productivity will be higher and unit costs lower for low-cost airlines. Low-cost airlines are also likely to be more aggressive in using pilots and cabin crew from lower wage countries in the case of international services. Low-cost airlines will also negotiate harder for lower wages and conditions for nonoperational labour. They will have fewer staff performing the same work as employees working for full-service airlines, and they will also have more cramped work environments with open plans rather than office arrangements resulting in more workers per unit of floor space. Low-cost airlines that predominantly fly into secondary airports will usually enjoy significant lower landing charges. Airports will often entice low-cost airlines into flying services to their airports by offering much lower airport charges, and by offering inducements to make large step changes to capacity, such as paying half the landing charge price if capacity into the airport is doubled, for example. Low-cost airlines are also likely to fly into airports outside of peak times. Airports will provide price incentives to low-cost carriers to operate outside of peak times because this will allow the airport more passenger throughput during those times of the day that are not subject to slot and other airport constraints. Low-cost carrier overheads are also likely to be significantly lower than that of full-service airlines. As indicated above, they cram their non-operational staff into smaller workstations in open plans, and lease property in cheaper premises thus reducing rent costs per employee. They also employ significantly fewer staff to build business cases and perform research functions thus lowering non-operational staff costs. Computer and IT costs are also usually significantly lower for low-cost carriers because they have less complicated yield management and customer reservation systems.

10.9.4 Full-Service Versus Low-Cost Unit Cost Comparisons In this section we compare the cost per available seat kilometre of full-service and low-cost airlines that operate within the same airline group. By analysing airlines within the same airline group, this will provide deeper insight into the difference in cost that is attributable to the type of carrier. Time series movements in Qantas and Jetstar EBIT level cost per ASK (inclusive of fuel costs) are presented in Figure 10-6. Qantas and Jetstar EBIT-Level Cost per ASK (c/ASK)

Qantas

Jetstar 13.80

12.02

7.46

12.09 11.00

6.97

11.28 10.38

10.20

6.78

8.28

Jun-20

6.54

7.48

Jun-19

Jun-15

Jun-14

Jun-13

Jun-12

6.52

Jun-18

7.20

11.97

Jun-17

7.04

12.06

Jun-16

15 14 13 12 11 10 9 8 7 6

Source: Airline Intelligence and Research Database 2021

Fig. 10-6: Qantas and Jetstar EBIT Level Cost per ASK The costs presented in Figure 10-6 include both domestic and international services for both Qantas and Jetstar brands. We can see in Figure 10-6 that the Qantas cost premium has varied over the timeframe from a low of 3.7 Australian cents per ASK up to 5.51 Australian cents per ASK. As a percentage of the Qantas EBIT cost per ASK, Qantas CASK has exceeded the CASK of Jetstar by between 35.9% and 41.5%. One of the factors that may be causing natural difference in the CASK of Qantas and Jetstar is the passenger average sector length, since we know from chapter 6 that an airline with a higher passenger sector length has a naturally lower CASK. In this case, however, the Qantas average sector

Chapter 10

280

length at around 2,800km is much higher than that of Jetstar at around 1,700km which would tend to make Qantas CASK lower than Jetstar’s. If we adjusted for sector length the difference between the Qantas CASK and that of Jetstar would be higher. The movement over time in the CASK of the full-service airline Singapore Airlines and the low-cost carrier Scoot Airlines, which is a fully owned subsidiary of the Singapore Airlines Group, is presented in Figure 10-7 below between 2015 and 2019. We can see in Figure 10.7 that the CASK premium that Singapore Airline holds over Scoot has varied between 3.0 Singapore cents per ASK up to 3.9 Singapore cents per ASK between 2015 and 2019. In percentage terms, Singapore airlines CASK exceeds Scoot CASK by between 36.0% and 48.1%. The Singapore Airlines average passenger sector length at around 5,000km is significantly higher than that of Scoot Airlines at around 3,000km, which would act to lower the Singapore CASK relative to Scoot. This would suggest that the estimated difference in Singapore and Scoot CASK in Figure 10-7 is likely to be conservative, like the Qantas versus Jetstar estimates. Singapore and Scoot CASK (SG c/ASK) 9

8.67

Scoot

Singapore 8.40

8.12

8.40

8

8.25

7 6 5.21 5

4.84

4.21

5.18

5.28

4 2015

2016

2017

2018

2019

Source: Airline Intelligence and Research Database 2021

Fig. 10-7: Singapore and Scoot Airlines Cost per ASK In Figures 10-8 and 10-9 below we compare the CASK of the airline Group carriers Air France/KLM and Transavia (Figure 10-8), which are a part of the Air France Group and Iberia and Vueling (Figure 10-9), which are a part of the International Airlines Group. Air France/KLM and Transavia CASK (Euro cents / ASK)

9.5

9.22

Air France/KLM 8.67

8.5

8.60

Transavia 8.79

8.92

7.5 6.5 5.5

5.00

4.68

4.72

4.73

4.95

4.5 2015

2016

2017

2018

2019

Source: Airline Intelligence and Research Database 2021

Fig. 10-8: Air France/KLM and Transavia Airlines Cost per ASK Air France/KLM’s CASK, which is full-service airline CASK, is higher than the low-cost carrier Transavia CASK by between 4.0 € cents per ASK and 4.2 € cents per ASK, which converts into a premium of 44.6% to 46.0%. The passenger average sector length of Air France/KLM is around 3,000km and that of Transavia is around 1,800km so once again the cost premium of Air France/KLM over Transavia is likely to be conservative. In the case of Iberia versus Vueling, the cost premium difference is much lower on an absolute basis, at between 1.2 € cent per ASK to 2.8 € cents per ASK, which represents a difference of between 17.5% and 35.1% over the period 2018 and 2019. Iberia’s passenger average sector length at around 2,900km is almost three times that of Vueling, which might be contributing to the relative closeness of the CASK estimates.

Airline Relationships and Business Models

Iberia and Vueling CASK (Euro c/ASK) 9.5 9 8.5 8 7.5 7 6.5 6 5.5 5

Iberia 9.00

Vueling 9.10 8.61

8.37 7.93

7.78

7.57

7.59

6.13

5.95 5.48

2008

281

2009

7.18 6.06

5.42 2010

2011

7.43

2012

7.36 6.96 6.07

5.90 2013

5.85 2014

5.89 5.84 2015

2016

5.54 2017

5.73 2018

2019

Source: Airline Intelligence and Research Database 2021

Fig. 10-9: Iberia and Vueling Airlines Cost per ASK

Quiz 10-4 FSA Versus LCC Airlines 1. Which of the following is NOT a typical characteristic of a full-service airline? (a) Unbundling services. (b) Strong frequency during the morning and afternoon peaks. (c) Significant number of flights into primary airports. (d) Relatively low airfares. (e) Affiliation with a major frequent flyer program. (f) Strong alliance partnerships. (g) Preoccupation with on-time performance. (h) Strong business-purpose passenger mix. (i) Relatively high income and wealth demographic. 2. Which of the following is NOT a typical characteristic of a low-cost airline? (a) Unbundling services. (b) Low prices. (c) Simple yield management systems. (d) Flights to major airports. (e) High levels of fleet complexity. (f) High leisure purpose mix. (g) Flying during the off-peak. (h) Free food service on-board the aircraft. 3. Which of the following is a typical cost category that is significantly different between full-service and low-cost carriers? (a) Manpower costs. (b) Fuel costs. (c) Depreciation costs. (d) Operating lease costs. (e) Airport charges. (f) Both (a) and (e). 4. Which of the following figures is the closest in terms of estimating the premium of full-service airline CASK over low-cost carrier CASK? (a) 0% (b) 1% (c) 5% to 10% (d) 20% to 50%

CHAPTER 11 AVIATION CHARGES, TAXES, AND A PRICE ON CARBON

One of the last things I did while employed by the Qantas Group as its Chief Economist was a ‘road’ trip to London, Brussels, Cologne, and Berlin with a group of three other colleagues. During that trip we met Her Majesty’s Treasury and the German Ministry of Finance to persuade them to make changes to their departure taxes – the Air Passenger Duty and the Air Travel Tax respectively. These taxes had an enormous impact on the profitability of Qantas services to London and Frankfurt. They were encouraging Australian and European passengers to fly on airlines that were taxed at lower rates, such as Emirates and Etihad, which Qantas quite rightly thought was unfair. Or in the language of regulatory economics, the taxes were not competitively neutral. It took some time, but the case we took to the UK Treasury appeared strong enough to persuade them to make changes. HM Treasury simplified its departure tax so that it became a two-distance band system where before it was a system that was based on four distance bands. We were not as successful with the German Government. The change in the UK departure tax was beneficial to Qantas, not so much directly but indirectly insofar as it treated key competitors in the same way as Qantas. Departure taxes are just one of a range of taxes that can influence the profitability of airlines. Other taxes include income or company taxes, fringe benefit taxes, payroll taxes, value added taxes, withholding taxes, carbon taxes and capital gains taxes. The introduction of these taxes, and changes to these taxes, represent a significant risk to airlines. Increases in any of these taxes will reduce the effective earnings of the airline, which will in turn disincentivise the airline to invest in new aircraft and in new routes. The aim of this chapter is to take you through the key aviation charges and taxes that affect airlines, and the way they affect airlines. Enjoy your reading and learning.

11.1 Fuel Surcharges 11.1.1 What is a Fuel Surcharge? Fuel surcharges are generally imposed by airlines on passengers as a method for recovering higher fuel costs brought about by fuel prices rising above some base level. Surcharges are usually set on a zone system. This means that fuel surcharges for city pairs that fall within the same zone will charge the same fuel surcharge. The zones are usually defined based on a particular geography. The geographies are generally chosen based on distance, with a geographic zone that involves longer travel incurring a higher fuel surcharge than a geographic zone that involves shorter travel. To illustrate, consider the international fuel surcharge zones of the Qantas Group from 2012 in Table 11-1 below. Zone Asia/Honolulu Johannesburg/Santiago United States London/Frankfurt

Fuel Surcharge One-way as at 12 April 2012 $175 $260 $340 $380

Source: Qantas Media Releases 2012

Table 11-1: Fuel Surcharge Structure of the Qantas Group as at April 2012 As can be seen in Table 11.1, the fuel surcharge is higher for zones that involve longer sectors. For example, the city pair Sydney to Hong Kong has a sector length of 7,372km. Passengers flying between these two cities on Qantas will be required to pay a fuel surcharge of $175 one-way. Sydney to Los Angeles has a sector length of 12,051km. Passengers on a Qantas flight between Sydney and Los Angeles will pay a fuel surcharge of $340 one-way, which is almost double the fuel surcharge paid for Sydney to Hong Kong. Consider another example of fuel surcharges, in this case the March 2012 fuel surcharges of Singapore Airlines as reported in Table 11-2 below. Singapore Airlines fuel surcharges vary by both distance and cabin across five surcharge zones. Relatively short sectors involving travel to and from South-East Asia such as Singapore to Indonesia, Malaysia, and Thailand, attract a fuel surcharge of US$36 in economy class, increasing to US$52 for business class and US$61

Aviation Charges, Taxes, and a Price on Carbon

Zone South-East Asia Europe and South Africa Americas Australia and New Zealand All Other Flights

First Class, Suites 61 211 318 313 150

Surcharge US$ Business Class 52 198 305 294 133

283

Economy Class 36 174 287 260 114

Source: Singapore Airlines Media Release 2012

Table 11-2: Fuel Surcharge Structure of Singapore Airlines as at March 2012 for first class. For longer distances, such as Singapore to the U.S.A. and Canada, the surcharge increases to US$287 for economy class, US$305 for business class and US$318 for first class. Singapore Airlines fuel surcharges are higher for premium travel compared to economy for at least two reasons. The first is that the premium cabin seats take up more surface area of the seated area of the aircraft, which means more of the fuel that is consumed by the aircraft in flight is allocated to premium cabin seats. As a result, premium cabin seats attract a higher fuel surcharge. The second reason is that premium cabin passengers are more likely to be able to afford a higher fuel surcharge than passengers who travel in economy class. Put in terms of air transport economics, premium cabin passenger demand is less sensitive to price increases than economy class passenger demand, allowing the airline to charge premium passengers more for fuel surcharges knowing that demand will not significantly fall in response. Let us use some simple modelling to be clearer about how we define a fuel surcharge. Let us assume that a particular airline that wishes to impose a fuel surcharge has created K fuel surcharge zones. Each fuel surcharge zone is defined based on a geographic distance, with zone 1 defined as city pairs that have a relatively short distance up to zone K which includes city pairs that have a relatively long distance. Each zone is charged a different surcharge, which we denote by ܲ௜ௌ for i = 1,……, K. The airline expects PAXi passengers in the ith fuel surcharge zone. The fuel surcharges are not split by class of travel for this simple model, which means that every class of travel pays the same fuel surcharge. These surcharge details are summarised in Table 11-3 below. Zone

Surcharge

PAX

(1)

(2)

(3)

1 2 3 . . . . K Total

Pଵୗ Pଶୗ Pଷୗ

PAX1 PAX2 PAX3 . . . . PAXK

. . . . P௞ୗ

୏

Zone Surcharge Revenue (4) = (2) u (3) Pଵୗ u PAX1 Pଶୗ u PAX2 Pଷୗ u PAX3 . . . . P௄ୗ u PAXK ୏

෍ PAX୧

෍ P୧ୗ PAX୧

୧ୀଵ

୧ୀଵ

Table 11-3: Fuel Surcharge Model Zones and Passengers The surcharge revenue that the airline expects to earn is equal to the surcharge in column (2) of Table 11-3 multiplied by the expected number of passengers to pay the surcharge in column (3) summed over each zone, which is: Surcharge Revenue = σ୏୧ୀଵ P୧ୗ PAX୧

(11.1)

The surcharge revenue is presented at the bottom of column (4) in Table 11-3. By dividing the surcharge revenue at (11.1) by the total number of expected passengers, which is the sum presented at the bottom of column (3) in Table 113, we obtain the average fuel surcharge imposed by the airline, which is: ഥୗ = Average Surcharge = P

౏ σే ౟సభ ୔౟ ୔୅ଡ଼౟

σే ౟సభ ୔୅ଡ଼౟

(11.2)

The airline wishes to set the average fuel surcharge (11.2) at a level that allows the airline to just recover its incremental fuel costs. To determine this cost-recovering average fuel surcharge, it is therefore necessary to determine the airline’s incremental fuel costs. These costs are equal to the current into-plane price of jet fuel less some base level of the into-

Chapter 11

284

plane jet fuel price, multiplied by expected jet fuel consumption times the local currency price of 1 US dollar. We can write the airline’s incremental fuel costs as follows: ୊୳ୣ୪ ୊୳ୣ୪ Incremental fuel costs = ൫Pେ୳୰୰ୣ୬୲ െ P୆ୟୱୣ ൯QFuelE

(11.3)

ி௨௘௟ . To understand what Let us now discuss each of the variables in (11.3), starting with the base fuel price variable, ܲ஻௔௦௘ the base jet fuel price should be in (11.3), it is necessary to understand why airlines began imposing fuel surcharges. Airlines began imposing fuel surcharges in and around calendar 2004. This date is important because it is this point in time at which jet fuel prices started to trend upward over time rather than cycle around a fixed mean. To see what I mean by this, consider the historical movements in jet fuel prices presented in Figure 11-1 below.

Gulf Coast Jet Fuel Prices (US$ / bbl) 140

$124

120 100

$90 $81

80

$72

60 40 $32 20

$26

$24 $22

$36 $29 $26 $17 $21 $30 $21 $24

$35

$126 $128

$123 $113 $85

$90 $64

$70

$21

$65 $53

$48

$79

$46

US$26/bbl

0 2020

2018

2016

2014

2012

2010

2008

2006

2004

2002

2000

1998

1996

1994

1992

1990

Source: Energy Information Administration Spot Prices 2021

Fig. 11-1: Calendar Annual Spot Price of Jet Fuel – 1990 to 2019 We can see in Figure 11-1 that jet fuel prices in US dollars cycled around a fixed mean of around US$26 for the 13-year period 1990 to 2003, as indicated by the dashed line in Figure 11-1. By cycling around a fixed mean, I essentially mean that when the jet fuel price rose above US$26 it reasonably promptly began to fall back towards US$26, and when it fell below US$26 it reasonably promptly began to increase back to US$26. In other words, the jet fuel price always tended to gravitate back to US$26 per barrel. After 2003, however, the jet fuel price began to increase but it no longer returned to US$26 per barrel, instead continuing to increase or trend upward over time. When the jet fuel price began to trend upward rather than cycle around a fixed mean, airlines began to impose fuel surcharges. It is this turning point between the jet fuel pricing cycling around a fixed mean and trending upward over time that defines the base period, which is around 2004. The base period price of jet fuel can therefore be considered the average price around which the jet fuel price cycled before it began to trend upward. The exact value of this average will depend on the period over which the average is computed, but it should be in the neighbourhood of US$26 per barrel. The quantity of jet fuel consumed, QJet is the jet fuel that the airline expects to consume over a given forward horizon. The length of the forward horizon is a length of time over which the airline is expected to re-consider its position once again on fuel surcharges. This forward horizon could be a week, month, quarter, half year or even a year. The length of this forward horizon will depend on the volatility of the price of jet fuel. If it is exceptionally volatile, then the airline may consider revising its fuel surcharges more frequently. The spot value of the local currency price of 1 US dollar, E in equation (11.3), can be defined in at least two different ways. The first is that it is the exchange rate which is expected over the forward horizon that is being considered. As it is difficult to predict the future exchange rate, then a reasonable assumption is that it is equal to the current value of the exchange rate. The second potential definition of E is that it is the average exchange rate value over the base period that is used to determine the base jet fuel price. In my view the preferred definition of the exchange rate is the current value, in the main part because it is this value that is likely to be the best predictor of the exchange rate over the forward horizon, and it is the forward horizon exchange rate that is relevant to converting the US dollar fuel costs that the airline wishes to recover into local currency terms. ி௨௘௟ The current value of the jet fuel price, ܲ஼௨௥௥௘௡௧ , is typically defined as the spot price of jet fuel at the point in time at which the airline is considering imposing fuel surcharges. It is this level of the jet fuel price that the airline believes will continue over the forward horizon that is being analysed. To see how (11.3) is used, consider the following illustrative example which involves the Mexican national carrier Aeroméxico. As at the end of September 2021, the airline is considering imposing a fuel surcharge on passengers because of climbing jet fuel prices. It is only considering one quarter ahead, which means it is only considering imposing a fuel surcharge for the December quarter 2021 and will reconsider its position as the end of that quarter approaches. At

Aviation Charges, Taxes, and a Price on Carbon

285

ி௨௘௟ the end of September 2021, the spot price of jet fuel is, ܲ஼௨௥௥௘௡௧ = US$95. The airline believes that in the December quarter 2021, jet fuel price will remain at this level. The airline expects to consume QFuel = 1.5m barrels of jet fuel during the quarter while it expects the Mexican Peso price of 1 US dollar to be E = 20. The airline sets the base price of jet fuel ி௨௘௟ at ܲ஻௔௦௘ = US$26 per barrel. Using (11.3) the incremental fuel costs of the airline in millions of Mexican Pesos is:

Incremental fuel costs Aeroméxico December Qtr 2021 = (95 െ 26)u1.5u20 = 2070m

(11.4)

For the average surcharge to recover incremental fuel costs we require (11.2) multiplied by the expected number of passengers to equal the incremental fuel costs at (11.3). We can write this as the following condition: ୊୳ୣ୪ ୊୳ୣ୪ ഥ ୗ PAX = ൫Pେ୳୰୰ୣ୬୲ P െ P୆ୟୱୣ ൯QFuelE

(11.5)

Equation (11.5) rearranges for the average surcharge as follows: ഥ Pୗ =

ూ౫౛ౢ ూ౫౛ౢ ୉ ቀ୔ూ౫౛ౢ ి౫౨౨౛౤౪ ି୔ా౗౩౛ ቁ୕

(11.6)

୔୅ଡ଼

Equation (11.6) describes the average fuel surcharge that an airline must charge if it is to recover its incremental fuel cost. The average surcharge at (11.6) depends on the difference between the current fuel price and the base fuel price, the quantity of jet fuel consumed, the local currency price of 1 US dollar, and the airline’s expected number of passengers. If we continue our Aeroméxico example from above, if Aeroméxico expects there to be 3.2m passengers in the December quarter 2021 then using (11.6) the fuel surcharge that is expected to recover fuel costs will be: (ଽହିଶ଺)(ଵ.ହ)(ଶ଴) ୗ ഥୈୣୡିଶଵ P = = 647 Mexican Pesos ଷ.ଶ

(11.7)

Aeroméxico’s average airfare in the September quarter 2021 was 2,200 Mexican peso (Airline Intelligence and Research Database 2021). This implies that the airline would need to lift its average airfare by 29.1% to recover the increase in its fuel costs.

11.1.2 Revenue Recovery with Varying Loads 11.1.2.1 Theory and Analytics Equation (11.6) provides a formula for determining the average fuel surcharge that is expected to recover costs for an airline given its expectation about the number of passengers, which is contained in the denominator of (11.6). One of the problems with the fuel surcharge for cost recovery approach that is summarised by equation (11.6) is that the term in the denominator, PAX, depends on the level of the surcharge. This is because passenger demand depends on the average airfare that is paid by the passenger, which is the sum of both the base fare and the fuel surcharge. When the fuel surcharge increases this increases the average airfare which in turn results in a reduction in passenger demand. To understand the level of the fuel surcharge and the change in the fuel surcharge that is necessary to recover incremental fuel costs, it is therefore necessary to understand how different levels of the fuel surcharge are expected to impact demand. In this section we will build a simple model of an airline’s revenue to find the change in the fuel surcharge that is expected to recover incremental fuel costs taking into consideration the impact of a higher fuel surcharge on passenger demand. To start our model, we first present a simple expression which defines an airline’s passenger revenue: R = PAX u (PB + PS)

(11.8)

where PAX is the number of passengers that fly with the airline, PB is the base fare that passengers pay, and PS is the average fuel surcharge that is paid by passengers. The sum PB + PS is the average airfare that is paid by passengers. Equation (11.8) simply says that the passenger revenue of the airline is equal to the number of passengers that the airline carries multiplied by the average price that those passengers pay, which consists of a base fare component plus a fuel surcharge component. We know, however, that an increase in the average price that passengers pay will result in a reduction in passenger demand and thus a reduction in PAX in (11.8). We describe the reduction in PAX by using the following PAX demand function: PAX = PAX(PB + PS)

(11.9)

The passenger demand function (11.9) simply says that an increase in the price paid by passengers, PB + PS will affect ௗ௉஺௑ passenger demand. We assume that an increase in price will lead to lower passenger demand so that ಳ ೄ < 0. An ௗ൫௉ ା௉ ൯

Chapter 11

286

increase in the base fare will have the same impact on passenger demand as the same magnitude increase in the fuel ௗ௉஺௑ ௗ௉஺௑ surcharge, which we can write as ಳ = ೄ . We wish to understand what happens to the left-hand side of (11.8) when ௗ௉ ௗ௉ we increase PS knowing through (11.9) that an increase in PS will lead to fewer passengers. To determine what happens to the left-hand side of (11.8) as we increase PS we must first totally differentiate (11.8) assuming dPS > 0 and dPB = 0 or the airline increases its fuel surcharge, but it leaves the base fare unchanged. The total derivative of revenue at (11.8) is: ୢ୔୅ଡ଼

dR = PAXudPS + (PB + PS)u

ୢ୔౏

udPS

(11.10)

Equation (11.10) says that revenue changes for two reasons. The first says that a higher fuel surcharge lifts revenue for a given number of passengers, which is the first component on the right-hand side of (11.10) and is positive. The second says that a higher fuel surcharge leads to a reduction in passenger demand and thus revenue, which is the second component on the right-hand side of (11.10). The change in revenue is therefore affected by both a positive price force and a negative volume force. We can simplify (11.10) by finding the common factor dPSuPAX as follows: dR = PAXuቂ1 +

ୢ୔୅ଡ଼ ௉ೄ ୢ୔౏ ௉஺௑

×

௉ಳ ା௉ೄ ௉ೄ

ቃudPS

ௗ௉஺௑ ௉ೄ

(11.11) ௉ಳ ା௉ೄ

{ HS is the elasticity of passenger demand to the fuel surcharge while the term ೄ { DS is The expression ೄ ௗ௉ ௉஺௑ ௉ the inverse of surcharge revenue as a percentage of total passenger revenue. Using these two definitions we can simplify (11.11) as follows: dR = PAXu[1 + ɂୗ × Ƚୗ ]udPS

(11.12)

We are now able to determine the conditions under which an increase in the fuel surcharge will lead to an increase in passenger revenue, and the extent to which passenger revenue will rise. An increase in the fuel surcharge will increase revenue when the expression in square brackets on the right-hand side of (11.12) is greater than zero, which will occur when 1 + HSuDS is greater than zero. As HS is less than zero and DS is greater than zero then we require the absolute value of the term HSuDS to be less than 1. We would expect that demand is relatively inelastic to the fuel surcharge, which means HS is a small negative number, and we expect DS to be a positive number that is greater than 1. The more inelastic is demand to the fuel surcharge and the larger is fuel surcharge revenue as a percentage of total revenue, the more likely it is that an increase in the fuel surcharge will generate an increase in revenue. To determine the increase in the fuel surcharge that is required for the increase in passenger revenue to recover incremental fuel costs, we set (11.12) equal to (11.3) as follows: ୊୳ୣ୪ ୊୳ୣ୪ ൫Pେ୳୰୰ୣ୬୲ െ P୆ୟୱୣ ൯QFuelE = PAXu[1 + ɂୗ × Ƚୗ ]udPS

(11.13)

Rearranging (11.13) so that dPS is the subject yields: dPS =

ూ౫౛ౢ ూ౫౛ౢ ୉ ቀ୔ూ౫౛ౢ ి౫౨౨౛౤౪ ି୔ా౗౩౛ ቁ୕

୔୅ଡ଼ൣଵାக౏ ×஑౏ ൧

(11.14)

Equation (11.14) measures the extent to which the fuel surcharge must increase for there to be full recovery of incremental fuel costs. The increase in the fuel surcharge required for full fuel cost recovery depends on the extent to which fuel costs increase, the elasticity of demand to the fuel surcharge, fuel surcharge revenue as a percentage of total revenue and the current number of passengers. 11.1.2.2 Linear Demand We now analyse the case in which the passenger demand function (11.9) is a linear function. We write this linear function in the following way: PAX = a0 + a1(PB + PS)

(11.15)

The passenger revenue of the airline is (11.15) multiplied by (PB + PS) as follows: R = PAX u (PB + PS) = a0(PB + PS) + a1(PB + PS)2

(11.16)

Aviation Charges, Taxes, and a Price on Carbon

287

We can expand on the passenger revenue function (11.16) by expanding the last term on the right-hand side of (11.16) as follows: R = a0PB + a0PS + a1(PB)2 + 2a1PBPS + a1(PS)2

(11.17)

The passenger revenue function of the airline at (11.17) is a concave down parabola in the passenger surcharge PS because the coefficient attached to (PS)2, which is a1, is less than zero. The revenue function as a function of PS, which is drawn schematically in Figure 11-2 below, starts above the origin, representing the passenger revenue the airline earns from base fares [a0PB + a1(PB)2]. This base fare revenue is earned even if the fuel surcharge were to drop to zero. The revenue function increases at a decreasing rate as the surcharge is increased from zero, reaches a maximum at a surcharge of ܲ෠ ௌ and then begins to fall. The reason the function increases at a decreasing rate and then eventually starts to turn downward is because as the fuel surcharge is increased there is a reduction in passenger demand, which in turn places downward pressure on passenger revenue. For fuel surcharges up to ܲ෠ ௌ passenger revenue increases because the positive impact on revenue of a higher airfare is stronger than the negative impact of a reduction in the number of passengers, however for fuel surcharge increases beyond ܲ෠ ௌ the negative passenger volume effect starts to dominate the positive price effect and revenue begins to fall. The level of ܲ෠ ௌ or the turning point of the revenue function is the point at which the expression 1 + ɂୗ × Ƚୗ from section 11.1.2.1 is equal to zero, or ɂୗ × Ƚୗ = -1. Revenue

A

C B

0

Pଵୗ

Pଶୗ

෡ୗ P

Fig. 11-2: Concavity of Passenger Revenue as a Function of the Fuel Surcharge Let us now find the surcharge figure that generates the point of maximum revenue at A in Figure 11.2. One of the methods that we can use to determine point A in Figure 11-10 is by differentiating (11.17) with respect to PS and setting the result equal to zero. This gives: ୢୖ ୢ୔౩

= ܽ଴ + 2ܽଵ (P ୆ + P ୗ ) = 0

(11.18)

Solving this for PS and we obtain the average fuel surcharge that generates maximum revenue for a given level of the base fare: ෡ ୗ = ିୟబ െ P ୆ P ଶୟభ

(11.19)

For the airline to generate additional revenue by raising the fuel surcharge for a give base fare, the fuel surcharge should not be increased above (11.19). Let us now show how our turning point value of the fuel surcharge ܲ෠ ௌ relates to the analysis of section 11.1.2.1 above, in which we demonstrated that an increase in the fuel surcharge will improve revenue if the condition 1 + ɂୗ × Ƚୗ > 0 holds. In our first step we compute HS, which is the elasticity of demand with respect to the fuel surcharge. This can be found by multiplying the coefficient attached to PS in the demand function (11.15) by PS/PAX as follows: ɂୗ =

ୟభ ×୔౏ ୔୅ଡ଼

(11.20)

Chapter 11

288

If we multiply (11.20) by the ratio of the full fare to the fuel surcharge, DS, we obtain: ɂୗ Ƚୗ =

ୟభ ×୔౏ ୔୅ଡ଼

×

୔౏ ା୔ా ୔౏

=

ୟభ ×(୔౏ ା୔ా ) ୔୅ଡ଼

=

ୟభ ×(୔౏ ା୔ా ) ୟబ ାୟభ ×(୔౏ ା୔ా )

(11.21)

We know that if (11.21) is equal to -1 then a small increase in the fuel surcharge will have no impact on revenue. If we set (11.21) equal to -1 and solve for the fuel surcharge, then we obtain the fuel surcharge consistent with point A in Figure 11-10. Setting (11.21) equal to -1 yields: ௔భ ×(௉ೄ ା௉ಳ ) ௔బ ା௔భ ×(௉ೄ ା௉ಳ )

= -1

(11.22)

Equation (11.22) simplifies to: aଵ × (P ୗ + P ୆ ) = -a଴ െ aଵ × (P ୗ + P ୆ ) This rearranges for PS as: ෡ୗ = P

െa଴ െ P୆ 2aଵ

which is precisely (11.19). For an increase in the fuel surcharge to increase revenue, the current level of the fuel surcharge will need to be located before the turning point of the passenger revenue function A. For an increase in the fuel surcharge to fully recover higher fuel costs, the current level of the fuel surcharge will need to be sufficiently lower than ܲ෠ ௌ to enable this to occur. For example, let us suppose that the current level of the fuel surcharge is ܲଵௌ and the airline increases the fuel surcharge to ܲଶௌ as indicated in Figure 11-2. The distance C minus B, which is the increase in revenue, will need to equal the incremental fuel costs for the fuel surcharge to fully recover costs. What is distance C minus B? To find this distance, we need to find the change in revenue using our revenue expression at (11.17). This change in revenue as a function of the change in the fuel surcharge, assuming the base fare is fixed, is: 'R = a0'PS + 2a1PB'PS +2a1PS'PS = [a0 + 2a1(PB + PS)]'PS

(11.23)

Equation (11.23) will need to equal the change in fuel costs from (11.3) for full recovery of those higher fuel costs, which implies ୊୳ୣ୪ ୊୳ୣ୪ െ P୆ୟୱୣ ൯QFuelE = [a0 + 2a1(PB + PS)]'PS ൫Pେ୳୰୰ୣ୬୲

(11.24)

The change in the fuel surcharge that recovers incremental fuel costs is then found by re-arranging (11.24) so that 'PS is the subject as follows: οP ୗ =

ూ౫౛ౢ ూ౫౛ౢ ୉ ቀ୔ూ౫౛ౢ ి౫౨౨౛౤౪ ି୔ా౗౩౛ ቁ୕

ୟబ ାଶୟభ ൫୔ా ା୔౏ ൯

(11.25)

Equation (11.25) is the extent which the fuel surcharge will have to increase to recover the higher fuel costs, taking into consideration the reduction in demand because of the increase in the fuel surcharge. This condition relies on the level of the fuel surcharge being sufficiently below ܲ෠ ௌ to enable full recovery of costs. This condition is equivalent to equation (11.14) in section 11.1.2.1. 11.1.2.3 Aeroméxico Illustration To illustrate the models presented in sections 11.1.2.1 and 11.1.2.2 we return to the case of Aeroméxico once again. We assume that the passenger demand of Aeroméxico per quarter in millions is described by the following demand function: PAX = 5.7 - 0.00116(PB + PS)

(11.26)

where PAX is defined in millions of passengers per quarter, and PB and PS are both defined in Mexican Pesos. The passenger revenue of the airline is equal to (11.26) multiplied by PB + PS which gives: R = 5.7(PB+PS) - 0.00116(PB+PS)2

Aviation Charges, Taxes, and a Price on Carbon

289

If we expand the terms in parentheses in this expression and simplify then we obtain: R = 5.7PB + 5.7PS – 0.00116(PB)2 – 0.00116(PS)2 – 0.00232PBPS

(11.27)

The base fare of the airline is PB = 2,000 Mexican Pesos. If we substitute the base fare into (11.27) we obtain the following revenue function: R(PB = 2000) = 5.7(2000) + 5.7PS – 0.00116(2000)2 – 0.00116(PS)2 – 0.00232(2000)PS This in turn simplifies to become: R(PB = 2000) = 16,040 + 1.06PS – 0.00116(PS)2

(11.28)

The graph of the passenger revenue function (11.28) is presented in Figure 11-3 below. Aeroméxico Quarterly Passenger Revenue (Pesos m) 16,350

A

16,300 R = 16,282 16,250 16,200

R = 16,173

B

16,150 16,100 16,050 16,000 15,950

PS = 457

PS = 150

15,900

798 777 756 735 714 693 672 651 630 609 588 567 546 525 504 483 462 441 420 399 378 357 336 315 294 273 252 231 210 189 168 147 126 105 84 63 42 21 0

Surcharge (Mexican Pesos) Fig. 11-3: Concavity of the Aeroméxico Passenger Revenue as a Function of the Fuel Surcharge The fuel surcharge that maximises the revenue of the airline is found by differentiating (11.28) with respect to PS and setting the result equal to zero, which yields: ୢୖ ୢ୔౏

= 1.06 െ 2 × 0.00116Pୗ = 0

(11.29)

Solving (11.29) for PS yields the revenue maximising fuel surcharge: ෡ୗ = P

ଵ.଴଺ ଶ×଴.଴଴ଵଵ଺

= 457 Pesos

Let us now show how this is consistent with HSDS = -1. The elasticity of demand with respect to the fuel surcharge is found by differentiating (11.26) with respect to PS and multiplying by PS/PAX as follows: HS =

ି଴.଴଴ଵଵ଺×୔౏ ௉஺௑

Substituting (11.26) into (11.30) yields:

(11.30)

Chapter 11

290

HS =

ି଴.଴଴ଵଵ଺×୔౏

(11.31)

ହ.଻ି଴.଴଴ଵଵ଺×൫୔ా ା୔౏ ൯

If we now multiply (11.31) by DS = (PS + PB)/PS we obtain: DSHS =

ି଴.଴଴ଵଵ଺×(୔ా ା୔౏ )

(11.32)

ହ.଻ି଴.଴଴ଵଵ଺×൫୔ా ା୔౏ ൯

Setting (11.32) equal to -1: ି଴.଴଴ଵଵ଺×(୔ా ା୔౏ ) ହ.଻ି଴.଴଴ଵଵ଺×൫୔ా ା୔౏ ൯

= െ1

(11.33)

Re-arranging (11.33) so that PS is the subject and assuming a base fare of PB = 2,000 gives: ෡ୗ = P

ିହ.଻ ିଶ×଴.଴଴ଵଵ଺

െ P୆ =

ିହ.଻ ିଶ×଴.଴଴ଵଵ଺

െ 2000 = 457 Pesos

which is precisely our answer given above. Let us now assume that the current fuel surcharge of Aeroméxico is 150 pesos as indicated by the circle in Figure 11-11 at point B. We wish to understand the extent to which Aeroméxico can impose a fuel surcharge to recover its higher fuel costs. As indicated in section 11.1.1 the airline’s fuel costs have increased by 2,070m Pesos. The best that Aeroméxico can do is to increase its fuel surcharge to 457 Pesos. This will lead to an increase in revenue of 16,282 – 16,173 = 109m Pesos. This would suggest that the airline would only be able to recover, at best, about 5% (= 109/2,070) of its higher fuel costs if it were to optimally raise its fuel surcharge.

11.1.3 Problems with Fuel Surcharges Fuel surcharges are a problematic tool for recovering incremental fuel costs for several reasons. The first and most important is that competition authorities often take an interest in fuel surcharges because the use of the term surcharge implies that the airline is advertising the fact that it is adding an element to its price to recover additional costs. If fuel surcharges are set in a way that generates more revenue than the cost it seeks to recover, then this could pique the interest of competition authorities. The use of fuel surcharges therefore adds to regulatory risk. The second problem with fuel surcharges is that it is difficult for an airline to estimate the fuel surcharge that recovers additional fuel costs. As indicated by equation (11.6) to estimate the average surcharge to recover fuel costs, the airline must determine values for the base price of jet fuel, the expected quantity of jet fuel to be consumed, the number of passengers it expects to carry on its aircraft, and the value of the local currency price of 1 US dollar. These are exceptionally difficult variables to estimate and forecast. If the airline estimates or forecasts these variables incorrectly and this leads to an over-recovery of fuel costs, then again this may pique the interest of competition regulators. The third difficulty associated with applying fuel surcharges is related to the relationship between passenger loads and the quantity of fuel that the airline consumes. When a fuel surcharge is imposed it leads to a reduction in demand, which in turn leads to lower loads on the aircraft. Lower loads on the aircraft lead to lower levels of fuel consumption because the weight of the loaded aircraft has declined. This reduction in fuel consumption will need to be factored into the computation of the change in fuel costs if the true change in fuel cost is to be estimated. This is not a simple calculation and will be estimated with significant risk of inaccuracy. A fourth problem with fuel surcharges is that it is usually defined on a zone basis as discussed above in Tables 11-1 to 11-3. The airline will define discrete zones that are determined based on a particular distance or geography. Each city pair that is located within the same zone will be subject to the same surcharge. The problem with this zone system of surcharging is that often zones constructed based on geography or distance can include city pairs that have significantly different sector lengths, which in turn results in significantly different levels of fuel consumption per flight. This means that city pairs within the same zone that have a below average sector length will more likely over-recover fuel costs while city pairs that have sector lengths that are longer than the zone average will under-recover. The result is that there will be many city pairs within the zone that are charging an additional price that does not necessarily align with incremental costs. For example, in the case of the Qantas set of international fuel surcharges in Table 11-1, Sydney to Hong Kong with a sector length of 7,372km will be charged the same fuel surcharge as Sydney to Singapore with a sector length of 6,289km. This is despite Sydney to Singapore burning significantly less fuel than Sydney to Hong Kong because Sydney to Hong Kong is a longer distance. Consider another example. When Qantas began charging domestic fuel surcharges, this fuel surcharge was the same for all domestic city pairs in the Australian aviation market. This means that passengers travelling on the Sydney (SYD) to Perth (PER) city pair, which is a distance of 3,284km, were charged the same fuel surcharge as passengers flying on Sydney to Canberra (CBR), which is 236km. This is despite a given increase in jet fuel prices leading to a far greater increase in fuel costs per passenger on SYD-PER than on SYD-CBR because fuel consumption per passenger is much higher on SYD-PER than on SYD-CBR.

Aviation Charges, Taxes, and a Price on Carbon

291

It is also the case that some routes within a zone may be characterised by demand that is much more sensitive to the average airfare than other city pairs within the same zone. The city pairs that are characterised by relatively elastic demand to the average airfare will see passenger numbers decline by proportionately more than those city pairs where demand is relatively inelastic to the average airfare. For example, Sydney to Melbourne, which is one of the busiest routes in world aviation, has a relatively high percentage of business-purpose travel, which is a travel purpose that is less sensitive to an increase in the airfare than leisure travel. The route Sydney to Hamilton Island (HTI), is almost entirely leisure-purpose travel. Passengers on this flight will be relatively sensitive to higher average airfares. If the fuel surcharge is the same on both SYD-MEL and SYD-HTI, which are both domestic services, then we would expect the drop in passenger numbers to be much higher in percentage terms in the case of SYD-HTI because demand is more sensitive to price on that route.

Quiz 11-1 Fuel Surcharges 1.

An airline provides the following details about its fuel surcharges: Zone South-East Asia Europe and South Africa Americas Australia and New Zealand All Other Flights

Surcharge ($) 61 211 318 313 150

Expected Passengers (m) 3.5 2.8 1.4 1.1 3.1

Use this information to calculate the following. (a) Total surcharge revenue that the airline expects to earn in millions. (b) Total number of expected passengers in millions. (c) Average surcharge in $. 2.

Why might an airline charge premium cabins a higher fuel surcharge than the economy cabin? Why might an airline keep fuel surcharges the same across the cabins?

3.

Lufthansa is expected to consume 75m barrels of jet fuel in a year. It assumes that the base price of jet fuel is US$25, and the current price of jet fuel is US$80. The airline expects that the Euro price of 1 US dollar is 0.90. What is the incremental fuel cost of Lufthansa in Euro? How does the incremental fuel cost change if the Euro price of 1 US dollar increases to 1?

4.

Garuda Airlines, the full-service national carrier of Indonesia, wishes to determine the average fuel surcharge which recovers fuel costs. It assumes a base price of jet fuel of US$26 per barrel, the current jet fuel price is US$100 per barrel, it expects to consume 12 million barrels of jet fuel, and it expects to carry 32m passengers. What is the airline’s average fuel surcharge for cost recovery in US dollars? If the airline’s scheduled passenger revenue is US$4,573m, by how much must the airline lift its average airfare for the full recovery of fuel costs?

5. Why did airlines begin to impose fuel surcharges? (a) Because the US dollar price of jet fuel has always cycled around a fixed mean. (b) Because the US dollar price of jet fuel changed from cycling around a fixed mean to cycling around an upward trend. (c) Because the US dollar price of jet fuel has always cycled around an upward trend. (d) Because the US dollar price of jet fuel changed from cycling around an upward trend to cycling around a fixed mean. 6.

7.

If the elasticity of air travel demand with respect to the fuel surcharge is -0.1 and surcharge revenue is 15% of total passenger revenue, will a small increase in the fuel surcharge result in an improvement in passenger revenue? If the current number of passengers is 10m, and fuel costs increase by $500m, then how much should the fuel surcharge increase to recover higher fuel costs taking into consideration a fall in demand in response to a higher fuel surcharge?

The demand for air travel for Garuda domestic is described by the following demand function, QD = 40,000,000 – 250,000 u (PB + PS), where PB is the base fare in US dollars and PS is the fuel surcharge in US dollars. The base airfare of Garuda Domestic is US$90. Use this information to answer the following questions. (a) Find an expression for the elasticity of demand with respect to the fuel surcharge? What is the value of the elasticity at a fuel surcharge of PS = US$10? Interpret your answer. (b) Find the Garuda Domestic revenue function as a function of the base price and the fuel surcharge. Simplify this function by assuming that PB = 90.

Chapter 11

292

(c) Use Microsoft Excel to draw Garuda Domestic revenue as a function of the fuel surcharge. (d) At what level of the fuel surcharge does Garuda Domestic maximise revenue? (e) Over what range of surcharge values would Garuda be able to increase revenue to pay for higher fuel costs? 8. Which of the following is a difficulty in applying fuel surcharges to recover higher fuel costs? (a) It is difficult to determine what the size of the fuel surcharge should be to recover costs. (b) City pairs with small distances within the same fuel surcharge zone will be subsidised by city pairs with relatively large distances within the same fuel surcharge zone. (c) It does not take into consideration the fact that demand could increase in response to the imposition of a fuel surcharge. (d) It does not consider the impact of any capacity changes that may take place in response to an increase in fuel prices. (e) It does not take into consideration the possibility that different city pairs within the same surcharge zone have different average airfare elasticities of air travel demand. (f) There is a circularity problem, specifically the imposition of a fuel surcharge reduces demand which in turn results in the need for a higher fuel surcharge. (g) It does not consider the fact that fewer passengers lead to lower fuel burn which in turn reduces the size of the required fuel surcharge recovery. 9.

The graph below presents a time series of the calendar annual movements in jet fuel prices between 1990 and 2018.

Jet Fuel Prices (US$/bbl)

$140

$126 $128

$124

$120

$113

$100 $80

$72

$60 $40

$123

$32

$26

$20

$24

$22 $21 $21 $26

$90

$83

$70

$66

$64

$48

$36 $24

$89 $81

$52

$35 $29

$21 $30 $17

$0 2018

2017

2016

2015

2014

2013

2012

2011

2010

2009

2008

2007

2006

2005

2004

2003

2002

2001

2000

1999

1998

1997

1996

1995

1994

1993

1992

1991

1990

Turkish Airlines is considering imposing a fuel surcharge in 2020. The airline intends to use a zonal system. The zones are defined in the table below, which also present key Turkish Airlines operating and financial statistics for each of the zones. Fuel Surcharge Zonesa Zone

ASKs ma

RPKs mb

Passengers Carried

Domestic Middle East and Europe Americas Africa Far East

24,095 64,489

20,490 49,495

32,932,100 28,135,103

Route Passenger Revenue (US$ m) $1,374 $5,044

28,495 16,914 42,002

24,967 12,892 35,757

2,734,867 3,427,929 5,445,561

$1,728 $1,202 $2,977

aThese

statistics are actual statistics for Turkish Airlines over the 12 months to October 2018 as obtained from Turkish Airlines monthly reporting.

Use this information to answer the following questions. (a) Find the average sector length for each zone. (b) Turkish Airlines ASKs in millions per barrel of fuel consumed is 4,951. Use this information to find the estimated fuel consumption in each zone.

Aviation Charges, Taxes, and a Price on Carbon

293

(c) From the chart of the calendar annual movement in jet fuel prices, justify an estimate of the base price of jet fuel. Explain how you have arrived at your answer. (d) Compute what you believe is an appropriate fuel surcharge for each zone in 2018. Discuss how each fuel surcharge varies with the average sector length in each zone. Does this make intuitive sense? Explain your answer. (e) Determine the percentage increase in the average airfare resulting from the imposition of the fuel surcharge. (f) Consider now the fuel surcharge you have set for the domestic zone. Explain the issues that are likely to emerge for the route IST-ESB compared to the route IST-VAN? (g) The airfare elasticity of demand for the different routes are given in the table below. Price Elasticities of Demand Zone Domestic Middle East and Europe Americas Africa Far East

Airfare Elasticity of Demand -0.3 -0.5 -0.75 -1.1 -0.9

Use this information to determine the reduction in passenger numbers that result from the imposition of the fuel surcharge. What implications does this have for the size of the fuel surcharge? Use this information to explain how variations in the base airfare provides the airline with greater flexibility in raising prices to recover higher fuel costs. (h) Now suppose that you set the fuel surcharge 20% higher. What issues might the Turkish Competition Authority have with this higher fuel surcharge? Explain your answer.

11.2 Departure Taxes 11.2.1 What is a Departure Tax? A departure tax is a tax that is charged by a government when a local or foreign resident departs the country in question. Different countries often call the departure tax a different name. In Australia it is called the Passenger Movement Charge (Australian Border Force 2019), in the UK it is called the Air Passenger Duty or the APD (Her Majesty’s Treasury 2019), and in South Africa, Germany, Austria and Norway it is called the Air Passenger Tax (SARS 2019). In most cases the airline is charged the departure tax, and this is passed onto passengers via ticket prices (as is the case for the Passenger Movement Charge in Australia, the Air Passenger Duty in the UK, and the Air Passenger Tax in Germany). In other cases, the departure tax must be paid at the airport prior to departure, or by some other prepayment method. In some cases, the departure tax amount depends on the distance that is flown by the passenger, as is the case in the UK, Germany, and Austria. In this case, the further a passenger travels the more tax that is paid. Airlines that fly a long distance from the departure point are relatively worse-off compared to airlines that fly a relatively short distance from their departure point. In the case of the UK Air Passenger duty, the level of the tax also depends on the class of travel. There are three rates for three classes of travel. They are the reduced rate, the standard rate and the higher rate - refer to Table 11-4 below for those rates from April 1, 2020. Band Band A (0 – 2,000 miles)

Reduced Rate £13

Standard Rate £26

Higher Rate £78

Band B (Over 2,001 miles)

£80

£176

£528

Source: Her Majesty’s Treasury 2019

Table 11-4: The UK Air Passenger Duty Rates The reduced rate is for travel in the lowest class of travel available on the plane for seat pitches less than 1.016m (40 inches). This is essentially the economy class rate. The standard rate is for any class of travel where the seat pitch is more than 1.016m (40 inches). This rate is for premium economy, business, and first-class travel. The higher rate is for travel in aircraft that weigh 20 tonnes or more and equipped to carry fewer than 19 passengers. Essentially this latter category captures the business jet market.

Chapter 11

294

11.2.2 A Model of the Impact of a Departure Tax 11.2.2.1 Theory and Analytics In this section we build a model to understand the impact of a departure tax on the net tourism outcomes of a country. The net tourism of a country is the spending that is brought into the country because of foreign resident visitors less the spending that leaves the country because of local residents visiting overseas countries. Let us assume that the number of people who travel out of a particular country for tourism reasons, QO, depends on the average price that is paid for outbound travel, PO plus the average departure tax paid, T. We can write this in the following way: QO = g(PO + T)

(11.34)

We also assume that the number of people who travel into the country for tourism reasons, QI, depends on the average price that is paid for inbound travel, PI, plus the average departure tax as follows: QI = h(PI + T)

(11.35)

Both the functions g(PO+T) and h(PI + T) are smooth and continuous functions so that we can differentiate them with respect to PO, PI, and T. As T is added to both the inbound and the outbound travel prices then the way that the passenger reacts to an increase in travel prices is the same as the way that the passenger reacts to the departure tax. We can write డொೀ

డொೀ

డொ಺

డொ಺

and ಺ = . All four terms are less than zero, indicating that as tourism prices or the departure tax this as ೀ = డ௉ డ் డ௉ డ் increase this reduces the demand for both inbound and outbound tourism. The net tourism of the country in question is (11.35) times PI minus (11.34) times PO as follows: NT = PIh(PI + T) – POh(PO + T)

(11.36)

Equation (11.36) measures the money entering the country from tourism less the money leaving the country from tourism. To measure the impact of the departure tax on a country’s net tourism outcome we simply differentiate (11.36) with respect to the departure tax T as follows: ୢ୒୘ ୢ୘

= P୍

ୢ୕౅ ୢ୘

െ P୓

But we know that ୢ୒୘ ୢ୘

= P୍

ୢ୕౅ ୢ୔౅

ௗொ಺ ௗ்

െ P୓

ୢ୕ో

(11.37)

ୢ୘

is the same as

ௗொ಺ ௗ௉಺

and

ௗொೀ ௗ்

ୢ୕ో

is the same as

ௗொೀ ௗ௉ೀ

, which means we can write (11.37) as: (11.38)

ୢ୔ో

By manipulating the right-hand side of (11.38) we can re-write it in terms of inbound and outbound price elasticities of demand as follows: ୢ୒୘ ୢ୘

= Q୍

୔౅ ୢ୕౅ ୕౅ ୢ୔౅

െ Q୓

୔ో ୢ୕ో ୕ో ୢ୔ో

= Q୍ ɂ୧ െ Q୓ ɂ୓

(11.39)

where HI is the inbound price elasticity of demand and HO is the outbound price elasticity of demand. Equation (11.39) says that the net tourism outcome for a country in response to a small change in the departure tax is equal to the number of inbound tourists multiplied by the inbound price elasticity of demand less the number of outbound tourists multiplied by the outbound price elasticity of demand. If the inbound and outbound price elasticities of demand are both equal to H, then we can write (11.39) as: ୢ୒୘ ୢ୘

= ɂ(Q୍ െ Q୓ )

(11.40)

Knowing that H < 0 because an increase in the tourism price leads to less tourism demanded, then (11.40) is greater than zero, meaning that the net tourism of the country improves in response to an increase in the departure tax, when QI < QO. This is the case when the country in question is a net tourism importer prior to the increase in the departure tax, which means inbound tourism spending is less than outbound tourism spending. Conversely, the net tourism of the country falls in response to an increase in the departure tax when QI > QO or the net tourism of the country is positive, which means that the country is a net tourism exporter. Imposing or increasing a departure tax for a country is therefore beneficial to that country when it is a net tourism importer. The departure tax revenue that is earned by the country is equal to the volume of inbound passengers plus the volume of outbound passengers multiplied by the departure tax as follows:

Aviation Charges, Taxes, and a Price on Carbon

295

TR = T[h(PI + T) + h(PO + T)]

(11.41)

The departure tax rate that maximises the departure tax revenue received by the government is found by differentiating (11.41) with respect to T and setting the result equal to zero, which yields: ୢ୘ୖ ୢ୘

= Q୍ + Q୓ + T ቀ

ୢ୕౅ ୢ୘

+

ୢ୕ో ୢ୘

ቁ=0

(11.42)

We can-re-write the condition (11.42) by manipulating the terms inside the parentheses on the right-hand side as follows Q୍ + Q୓ + ቀ

ୢ୕౅ ୘ ୢ୘ ୕౅

Q୍ +

ୢ୕ో ୘ ୢ୘ ୕ో

Q୓ ቁ = 0

If we assume the elasticities of inbound and outbound demand to the tax rate are equal simplifies to:

(11.43) ௗொ಺ ் ௗ் ொ಺

Q୍ + Q୓ + (Q୍ + Q୓ )ɂ୘ = (Q୍ + Q୓ )(1 + ɂ୘ ) = 0

=

ୢ୕ో ୘ ୢ୘ ୕ో

= HT then (11.43)

(11.44)

As QI + QO > 0 then the condition (11.44) is satisfied when HT = -1. This means that the government will maximise the taxation revenue it receives from departure taxes when it sets the tax rate at a level at which the elasticity of inbound and outbound travel to the departure tax equals -1. We will demonstrate this condition in the numerical example to following in the next sub-section. 11.2.2.2 Linear Demand Let us now assume that the outbound demand function and the inbound demand function are given by the following linear functions of the average prices paid for trips to the tourism destination plus the departure tax: QO = a0 + a1(PO + T)

(11.45)

QI = b0 + b1(PI + T)

(11.46)

The parameters a1 and b1 are both less than zero, indicating that if it is more expensive for a local resident to travel abroad, the demand for outbound travel falls, and if it is more expensive for foreign residents to travel to the country in question then there will be less inbound travel. The net tourism outcome for the country in question is equal to inbound spending, which is (11.46) times PI, less outbound spending, which is (11.45) times PO. This represents tourism export receipts less tourism import spending and can be written as: Net Tourism (NT) = b0PI + b1(PI + T)PI - a0PO - a1(PO + T)PO

(11.47)

The net tourism at (11.47) can be written in the following quadratic form in travel prices: Net Tourism (NT) = (b0 + b1T)PI + b1(PI)2 – (a0 + a1T)PO - a1(PO)2

(11.48)

The net tourism outcome at (11.48) indicates that both outbound and inbound tourism spending are concave down parabolas in the price of travel as indicated in Figure 11-4. The outcome for net tourism depends on whether the outbound function sits higher than the inbound function as is the case in Figure 11-4 or the inbound function sits higher than the outbound function. It also depends on the price of inbound travel relative to the price of outbound travel. If both the inbound and outbound prices are the same as is the case for ܲଵூ and ܲଵை in Figure 11-4, and the outbound function sits higher than the inbound function, then the tourism position of the country in question will be in net deficit (NT1 < 0) as indicated in Figure 11-4 by point A sitting higher on the outbound curve than point B sits on the inbound curve. It is possible, however, that the country can be in net tourism surplus even though the outbound function sits higher than the inbound function if the inbound price is set to maximise inbound spending at ܲଶூ while the outbound price is set suboptimally and too low at ܲଶை , which generates a net tourism surplus of C minus D or NT2 > 0. The impact of an increase in T on the net tourism outcome can be found by differentiating (11.48) with respect to T, which yields: ௗே் ௗ்

= b1PI – a1PO

(11.49)

As both b1 and a1 are less than zero, than (11.49) is greater than zero when the following condition holds:

Chapter 11

296

Tourism Spending

A

Tଵ୓ Tଶ୍

C

NT1 < 0 NT2 > 0

Tଵ୍

B

Tଶ୓

Outbound

D

Inbound

0

Pଶ୓

Pଵ୍ =Pଵ୓

Pଶ୍

Price of Travel

Fig. 11-4: Net Tourism as a Function of the Price of Travel – Outbound Sits Higher than Inbound PO >

ୟభ ୔౅

(11.50)

ୠభ

Let us see how we can reconcile this result with (11.39) above. We first calculate the price elasticities of outbound and inbound demand using (11.45) and (11.46): HI =

ୠభ ୔౅

HO =

୕౅

ୟభ ୔ో ୕ో

Substituting these into the condition (11.39) and we obtain: ୢ୒୘ ୢ୘

= Q୍ ɂ୧ െ Q୓ ɂ୓ = b1PI – a1PO > 0

which is precisely (11.49) above. The departure tax revenue of the country in question is the tax rate T multiplied by the sum of (11.45) and (11.46), which is: TR = T[a0 + a1(PI + T) + b0 + b1(PO + T)] = (a0 + a1PI + b0 + b1PO)T + (a1 + b1)T2

(11.51)

As described in Figure 11-5 below, the departure tax revenue function is a concave down parabola as a1+ b1, which is the coefficient attached to T2 in (11.51) is less than zero. The tax revenue maximising departure tax rate is found by differentiating (11.51) with respect to T and setting the result equal to zero: ୢ୘ୖ ୢ୘

= a0 + a1PI + b0 + b1PO + 2(a1 + b1)T = 0

(11.52)

Solving (11.52) for T yields T* =

௔బ ା௔భ ௉಺ ା௕బ ା௕భ ௉ೀ ିଶ(௔భ ା௕భ )

(11.53)

The optimal departure tax rate depends on the strength of the underlying demand for inbound and outbound travel, the elasticity of inbound and outbound demand to tourism prices and the value of those prices. There are strong incentives for a government to lift the departure tax when it is located to the left of the optimal departure tax rate T* in Figure 11-5, and the country in question is a net tourism importer. An increase in the departure tax in this case is likely to raise revenue for the government as well as improve the balance on tourism, both of which help to support an improvement in a country’s income.

Aviation Charges, Taxes, and a Price on Carbon

297

Tax Revenue

0

௔బ ା௔భ ௉಺ ା௕బ ା௕భ ௉ೀ

T* =

T

ିଶ(௔భ ା௕భ )

Fig. 11-5: Departure Tax Revenue Function 11.2.2.2 Application to the UK Market According to the Office for National Statistics in the UK, there were QO = 93m U.K. residents who travelled overseas in 2019 and QI = 41m foreign residents who travelled to the U.K. in 2019. U.K. residents spent around £60b when they travelled overseas, and foreign residents spent around £30b in the U.K. in 2019 (Office for National Statistics 2020). The spending per U.K. resident for outbound travel and the spending per visitor in the case of inbound travel in 2019 is therefore PO = 60,000/93 = £645 and PI = 30,000/41 = £732. The demand for outbound travel by U.K. residents and the demand for inbound travel to the U.K. by foreign residents are described by the following linear functions: QO = 200 - 0.2(PO + T)

(11.54)

QI = 120 - 0.1(PI + T)

(11.55)

where QO and QI are both measured in millions of travellers and PI and PO are measured in pound sterling. Equations (11.54) and (11.55) are equivalent to (11.45) and (11.46) respectively. The net tourism outcome for the U.K. is PI multiplied by (11.55) less PO multiplied by (11.54), which is: NT = PI(120 – 0.1T) – 0.1(PI)2 – PO(200 – 0.2T) – 0.2(PO)2

(11.56)

This is equivalent to (11.48). If we assume that the departure tax is set at T=15 then we can draw the inbound and outbound spending functions that comprise (11.56) – refer to Figure 11-6 below. We can see in Figure 11-6 that the outbound function sits above the inbound function for most values of the spend per visitor. At current levels of expenditure per tourist of PO = £645 and PI = £732 the expenditure on outbound is estimated to be £43,860m and the expenditure on inbound is estimated to be £33,160m resulting in a net tourism deficit of £10,700m. The impact of an increase in the departure tax on net tourism at (11.56) is found by differentiating (11.56) with respect to T as follows: ୢ୒୘ ୢ୘

= െ0.1P ୍ + 0.2P ୓

(11.57)

An increase in the departure tax will result in an improvement in the net tourism spending in the U.K. when (11.57) is great than zero. This will be the case when: PO > 0.5PI

(11.58)

Based on Office for National Statistics in the U.K. numbers, PO at £645 is greater than half PI which is £366, which means we would expect an increase in the U.K. departure tax to improve the net tourism deficit in the U.K. An increase in the departure tax will improve the net tourism of the U.K. because it results in outbound tourism spending falling by

Chapter 11

298

Tourism Spending (GBP m) Inbound

60,000

Outbound

50,000

43,860 40,000

33,160

NT2019 = -10,700

30,000 20,000 10,000

PO = 645

0

PI = 732

898 875 852 829 806 783 760 737 714 691 668 645 622 599 576 553 530 507 484 461 438 415 392 369 346 323 300 277 254 231 208 185 162 139 116 93 70 47 24 1

Tourism Spend per Visitor (GBP) Fig. 11-6: U.K. Inbound and Outbound Tourism Spending Functions 2019 more than inbound tourism spending, which is largely because the starting position is one of a net tourism deficit, or outbound spending exceeds inbound spending. This is indicated in Figure 11-7 below. We can see in Figure 11-7 that the doubling of the departure tax from T=15 to T=30 has shifted the spending curves for both inbound and outbound down. The outbound line shifts down by more than the inbound line at the current levels of inbound and outbound spending per visitor, causing the net tourism deficit to improve from -£10,700 up to -£9,863. The total taxation revenue from the departure tax for Her Majesty’s Treasury is (11.54) plus (11.55) multiplied by T as follows: TR = T[200 - 0.2(PO + T) + 120 - 0.1(PI + T)]

(11.59)

If we evaluate these functions at current values for PO and PI, we obtain the following total revenue function: TR = T[117.8 – 0.3T] = 117.8T – 0.3T2

(11.60)

The U.K. departure tax revenue function is therefore a concave down function of the tax rate as illustrated in Figure 118 below. The turning point of the concave down tax revenue function is found by differentiating (11.60) with respect to ܶ and setting the resulting equation to zero, which yields: ܴ݀ܶ = 117.8 െ 0.6ܶ = 0 ݀ܶ This equation solves for the following optimal level of the U.K. departure tax: T* =

ଵଵ଻.଼ ଴.଺

= £196.3

(11.61)

From an average departure tax rate of T=15 Her Majesty’s Treasury could both improve its tax collections and its balance on tourism if it were to increase the average departure tax.

Aviation Charges, Taxes, and a Price on Carbon

299

Tourism Spending (GBP m) 60,000

Inbound (T=15)

Outbound (T=15)

Inbound (T=30)

Outbound (T=30)

50,000

41,925 40,000

NT2019 = -9,863 32,062

30,000

Inbound 20,000

Outbound 10,000

PO = 645

0

PI = 732

898 875 852 829 806 783 760 737 714 691 668 645 622 599 576 553 530 507 484 461 438 415 392 369 346 323 300 277 254 231 208 185 162 139 116 93 70 47 24 1

Tourism Spend per Visitor (GBP) Fig. 11-7: U.K. Inbound and Outbound Tourism Spending Functions 2019 with a Departure Tax Increase

U.K. Departure Tax Revenue (GBP m) 14,000

12,000

10,000

8,000

6,000

4,000

2,000

0 245 238 231 224 217 210 203 196 189 182 175 168 161 154 147 140 133 126 119 112 105 98 91 84 77 70 63 56 49 42 35 28 21 14 7 0

Departure Tax Rate (GBP) Fig. 11-8: U.K. Departure Tax Revenue Function

Chapter 11

300

Quiz 11-2 Departure Tax 1. (a) (b) (c) (d)

What is a departure tax? A tax on the number of passengers in transit at an airport and about to depart. A tax on the total enplaned passengers of an airline. A tax on the number of passengers who depart an airport. A tax on the number of passengers who arrive at an airport.

2. (a) (b) (c) (d)

What is the name of the departure tax that is imposed by Her Majesty’s Treasury? Airport tax. Passenger Movement Charge. Air Travel Tax. Air Passenger Duty.

3.

The following are the rates of departure tax for a passenger that departs from an airport in the United Kingdom:

(a) (b) (c) (d) (e)

Band Band A (0 – 2,000 miles)

Reduced Rate £13

Standard Rate £26

Higher Rate £78

Band B (Over 2,001 miles)

£80

£176

£528

Use this information to determine the rate of departure tax that would be paid by the following types of passengers over the following journeys. An economy class passenger departs from London Heathrow for Frankfurt. A business class passenger departs from London Gatwick for Shanghai. A business-jet passenger departs from London Heathrow for New York. An economy class passenger travels from London Stansted Airport to Amsterdam Airport Schiphol. An economy class passenger travels from London Luton Airport to Charles De Gaulle Airport.

4. What is the net tourism of a country? (a) The number of foreign resident tourists entering the country less the number of resident tourists leaving the country. (b) The spending of foreign resident tourists entering the country less the spending of resident tourists leaving the country. (c) The spending of resident tourists leaving the country less the spending of foreign resident tourists entering the country. (d) The number of resident tourists leaving the country less the number of foreign resident tourists entering the country. 5.

Explain why a departure tax is more likely to be considered by countries that are net tourism importers rather than net tourism exporters.

6.

Consider the following graph of the inbound and outbound tourism outcomes of a country. Tourism Spending

E A F B C D

0

Inbound Outbound

Pଵ୍

Pଵ୓

Price of Travel

Aviation Charges, Taxes, and a Price on Carbon

301

The initial prices that are paid by inbound foreign resident tourists is ܲଵூ and the initial price paid by outbound resident tourists is ܲଵை . Use this information to answer the following questions. (a) At the initial prices, is the country in net tourism surplus or deficit? What is the size of the surplus or deficit? What would be the impact of an increase in the departure tax? Explain your answers. (b) If inbound foreign residents and outbound residents paid the same tourism price, what would be the outcome for the net tourism position? Explain your answer. (c) If tourism travel prices were set so that inbound and outbound expenditure were both maximised, what would be the outcome for net tourism? (d) If the outbound tourism price remained at the initial price, and the inbound tourism price was set at a level that maximised inbound tourism spending, what would be the outcome for net tourism? 7. When would a country find it unambiguously optimal to increase the departure tax rate? (a) When the actual departure tax rate is less than the revenue maximising rate and the country is a net tourism importer. (b) When the actual departure tax rate is less than the revenue maximising rate and the country is a net tourism exporter. (c) When the actual departure tax rate is higher than the revenue maximising rate and the country is a net tourism importer. (d) When the actual departure tax rate is higher than the revenue maximising rate and the country is a net tourism exporter. 8.

(a) (b) (c) (d) (e) (f) (g) (h)

The outbound demand function for Australian resident travel overseas is described by the linear function QO = 14,984,836 – 766PO while the inbound demand function for foreign resident travel to Australia is described by the linear function QI = 13,062,898 – 836PI. The current set of inbound and outbound tourism prices are PO = 6,520 and PI = 5,211, both denominated in Australian dollars. Both inbound and outbound Australian travellers pay a departure tax of T. Use this information to answer the following questions. Construct a concave down parabolic expression for both inbound spending by foreign visitors to Australia and outbound spending by Australians travelling abroad as a function of the tourism prices and the departure tax. Determine the net tourism function for Australia as a function of the departure tax and inbound and outbound prices. Graph the inbound and outbound spending functions as a function of the tourism prices assuming a departure tax of T = 20. Find the inbound price that maximises inbound spending and the outbound price that maximises outbound spending. Is Australia a net tourism importer or exporter at current prices? Does this result change if we determine inbound and outbound spending at the spending maximising prices? What is the impact on the net tourism deficit or surplus if the departure tax is doubled to T = 40? Demonstrate your answer in a graph. Find the departure tax revenue function of the Australian Treasury as a function of the departure tax. Evaluate the function at the current inbound and outbound prices. Find the departure tax that maximises the revenue of the Australian Treasury. What incentives does the Australian Treasury have to increase or decrease the departure tax?

11.3 Carbon Taxes and Emissions Trading Schemes 11.3.1 Carbon Taxes Imposed on Aviation Markets A carbon tax is a tax on the volume of carbon that is emitted by a company in a relevant jurisdiction, or it is a tax on the quantity of a good or service that is emissions intensive. It is intended to place a price on carbon emissions so that companies face the full social cost of the items they are producing rather than simply the private cost, and consumers face the full social cost of the goods and services that they are consuming. A carbon tax works in one of two ways. When it is imposed on companies that emit carbon, it works by increasing the costs of production of the company that is emitting the carbon. The increase in cost is passed through into higher prices which in turn reduces demand and production. The reduction in demand and production reduces carbon emissions. When it is imposed on goods that are carbon intensive, it works by increasing the price of those goods, which causes a reduction in the demand for those goods and services. This in turn causes producers to supply fewer of these emission intensive goods, which in turn reduces emissions. To illustrate the two different ways that a carbon tax is imposed consider the following. An airline generates emissions by consuming jet fuel as a part of its flying missions and the diesel that it consumes using land equipment at airports and between the airport and its crew base. A carbon tax is imposed on airlines based on the quantity of emissions that it produces. The carbon tax adds to the airline’s variable costs, which it passes through to its passengers in the form of higher airfares. Fewer passengers fly because of the higher airfares, forcing seat factors to fall, which ultimately forces the airline to cut capacity. A reduction in capacity results in a reduction in the jet kerosene that it consumes which in turn reduces its carbon emissions.

Chapter 11

302

A worker in the city uses a car to drive into work five times per week. The gasoline that is used in that car is subject to a carbon tax. Each time the worker fills the car up with gasoline the worker pays a carbon tax. The carbon tax has the effect of increasing the price that the worker pays for gasoline. In response to the higher price, the worker decides to car-pool for the commute to work with a neighbour, who also works in the city. This reduces the quantity of gasoline that is demanded by the two workers by a half which in turn reduces the number of emissions produced by the workers. At the time of writing, carbon taxes were imposed in a range of countries and states – refer to Table 11-5 below. Country/States

Time Period

Finland Poland Norway Sweden Denmark Slovenia Estonia Latvia Switzerland Liechtenstein Iceland Ireland Ukraine Japan Australia France Mexico Spain Portugal Alberta Chile Colombia Argentina Singapore Newfoundland and Labrador Prince Edward Island South Africa

1990 to Present 1990 to Present 1991 to Present 1991 to Present 1992 to Present 1996 to Present 2000 to present 2004 to Present 2008 to Present 2008 to Present 2010 to Present 2010 to Present 2011 to Present 2012 to Present 2012 to 2014 2014 to Present 2014 to Present 2014 to Present 2015 to Present 2017 to Present 2017 to Present 2017 to Present 2018 to Present 2019 to Present 2019 to Present 2019 to Present 2019 to Present

Tax as of 1 April 2019 (US$/CO2-eq) 70 ‫ݍ‬ത௜

(11.73)

The compound cost function says that if the airline carries fewer passengers than implied by the emissions cap than its total costs will be its private production costs as described by equation (11.72). If the airline carries more passengers than implied by the emissions cap then the airline faces a carbon cost, which is added to the private costs at (11.72) to give the social cost function described by (11.73). If we sum the ith airline compound function over all n airlines that operate on the route, then we obtain the market compound cost function: C = cതQ + K

ഥ for Q d Q

(11.74)

C = cതQ + PCJW + K

ഥ for Q > Q

(11.75)

The market compound cost function implies that the market marginal passenger cost is ܿҧ for market output that is below the aggregate cap for the route, ܳത , and ܿҧ + PCJW, for market output that is greater than the market passenger cap. We illustrate the impact on marginal passenger costs in Figure 11-10 below. We can see in Figure 11-10 that the horizontal market marginal cost line begins as a firm horizontal line but stepsup vertically at the point of the market cap on carbon emissions and passengers carried, ܳത . In this example the carbon emissions cap is not binding because the cap is set too low, with the equilibrium level of passengers carried Q* lower than the cap, occurring at equilibrium point E where the private market marginal cost (firm horizontal line) is equal to the market marginal revenue (dashed downward sloping line). In Figure 11-11 below we have a situation of a binding carbon emissions cap. In this case the carbon emissions cap is set sufficiently low, that the social market marginal cost line meets the dashed market marginal revenue at a level of passengers carried, Q** that is more than the constrained passengers carried, ܳത . The number of passengers Q** is cleared by the market at an average airfare of P**. Absent the ETS, the market equilibrium passengers carried would have been higher at Q*, where the private market marginal cost line meets the market marginal revenue line, and the equilibrium airfare would have been lower at P*. As the level of passengers carried is higher absent the ETS then the ETS has the effect of reducing carbon emissions. P(Q)

P*

n(ܿҧ +PCJW) E

ncത

ଵ

MR = P(Q) ቂn + ቃ க

0

Q*

ഥ Q

P(Q) Q

Fig. 11-10: n-Player Cournot Model with a Non-Binding Emissions Trading Scheme

Aviation Charges, Taxes, and a Price on Carbon

307

P(Q)

P** P* E1

n(ܿҧ +PCJW)

nܿҧ

ଵ

MR = P(Q) ቂn + ቃ

0

க

P(Q)

ܳത Q** Q*

Q

Fig. 11-11: n-Player Cournot Model with a Binding Emissions Trading Scheme

Quiz 11-3 Carbon Tax and Emissions Trading Scheme 1.

What are the two ways in which carbon taxes are imposed?

2.

When a carbon tax is imposed on a company that generates carbon emissions in its production process, such as an airline, how does the tax result in lower emissions? It raises the production costs of the producer, which are passed through into lower prices. This in turn results in lower demand and production, and thus lower emissions. It lowers the production costs of the producer, which are passed through into higher prices. This in turn results in lower demand and production, and thus lower emissions. It raises the production costs of the producer, which are passed through into higher prices. This in turn results in lower demand and production, and thus lower emissions. It raises the production costs of the producer, which are passed through into higher prices. This in turn results in higher demand and production, and thus lower emissions.

(a) (b) (c) (d) 3.

Name three countries that impose a carbon tax. Research how the tax is imposed (for example is it imposed on upstream, midstream or downstream producers?), on what gas it is imposed, and the rate of tax.

4.

Describe how a carbon tax may have distributional effects.

5.

Describe how a carbon tax that is imposed in some countries but not in others may have little impact on global emissions because it is not competitively neutral.

6.

Two airlines compete on a route in Cournot competition. The airlines pay an average private marginal passenger cost of $300. The price elasticity of demand at the market level for the route is -1.2. Both airlines consume 0.75 of a barrel of jet fuel per passenger on the route. Each barrel of jet fuel burned generates 500 kilograms of CO2e. A carbon tax is imposed on both airlines of T = 25 per tonne of CO2-e. What is the absolute change in the average airfare because of the imposition of the carbon tax? What is the percentage change in the average airfare, passenger demand and carbon emissions? Demonstrate your results in a suitable graph.

7.

Describe the key differences between an emissions trading scheme and a carbon tax.

CHAPTER 12 THE ECONOMICS OF OIL AND JET FUEL MARKETS

The most important economic exposures of the airline business are the state of the economy, the spot price of jet fuel, exchange rates, interest rates and inflation. While the state of the economy has the biggest impact on the airline business (think the Global Financial Crisis), it is not the economic exposure that keeps airline Treasury Risk management awake at night for most nights. This title goes to the jet fuel price risk of airlines. The reason the jet fuel price exposure worries airline Treasury Risk management more than the state of the economy is because the spot price of jet fuel is more volatile, it can move to higher levels quickly, in some cases overnight, and it has a significant impact on airline costs and profitability. This is unlike the economic exposure which is significantly less volatile, with periods of recession or severe economic weakness occurring just once or at most twice every ten years. Given the jet fuel price is the most important exposure of the airline business, airlines dedicate resources to understanding its historical movements and anticipating its movements in the future, to protect the airline from the possibility that jet fuel prices may increase, and to minimise the impact of higher jet fuel prices once they have increased. The most important input into the production of jet fuel is crude oil, which is converted into jet fuel and other enduser products such as gasoline and diesel in the oil refining process. As such the price of oil is the most important force on which airline Treasury Risk management analysts must focus if they are to better understand, anticipate and react to higher jet fuel prices. The price of oil is determined in the global market for oil according to movements in the demand for and the supply of oil. The demand for oil depends on the volume of global production of goods, which depends on the size of the global population and the average income of that population. Oil demand also depends on the price of oil itself, with an increase in the price of oil causing a reduction in oil demand. The production of crude oil depends on the costs of finding, extracting, storing and transporting oil and refining it into end-user products. Like the demand for oil, the supply of oil also depends on the price of oil, with an increase in the price of oil incentivising oil producers to supply more oil to the market. It is the interaction of these demand and supply forces, combined with the structure of the supplyside of the oil market, which determines the movement in the market price of crude oil over time. In this chapter you will learn how to combine these demand and supply-side forces to build an economic model of the crude oil price. You will also learn how to build an economic model of the process of refining crude oil into jet fuel and using that model to predict the likely outcome for the price of jet fuel as a function of the price of crude oil. If you are an analyst in the Treasury Risk management area of an airline, or if you aspire to be one, then this chapter is a must read.

12.1 The Demand for Oil Crude oil is the basis for a wide variety of goods that are produced in the global economy. While some of these products are well known, such as gasoline for motor vehicles, diesel for trains and marine vessels, bitumen for roads, and kerosene for warming homes and powering aircraft, many are virtually unknown. Table 12-1 presents a list of products that are less known as having oil as a basis. Solvents Dyes Soap Tool racks Tennis racquets Heart valves Golf balls Life jackets Antiseptics Insect repellent Sunglasses Shoes Footballs

Ink Shampoo Petroleum Jelly Bicycle tyres Hair curlers Dentures Balloons CD players Basketballs Fertilizers Parachutes Refrigerant Detergents

Wax Crayons Insecticides Umbrellas Combs Anaesthetics Luggage Drinking cups Purses Aspirin Dishes Electrical tape Contact lenses

Source: Ranken Energy Corporation 2021

Table 12-1: Products that have Oil as a Basis

Nail Polish Toothpaste Upholstery Nylon rope Vaporizers Fan belts Toilet seats Ballpoint pens Deodorants Awnings Artificial limbs Shaving cream Tents

Perfumes Paint Curtains Water pipes Fishing lures Refrigerators TV cabinets Golf bags Panty hose Paint brushes Folding doors Oil filters Telephones

Shoe Polish Lipstick Putty Guitar strings Hand lotion Model cars Dyes Toolboxes Trash bags Upholstery Clothesline Surf boards Cameras

The Economics of Oil and Jet Fuel Markets

309

The fact that crude oil is used in the production of a wide variety of goods tells us that we should expect the global demand for crude oil to be tightly linked to the world economy’s total production of goods. Figure 12-1 below presents the movement over the past 5 decades in the growth in global oil consumption, which is information that is obtained from the BP Review of World Energy Statistics, one of the best energy databases in the world. We can see in Figure 121 that between 1966 and 1985 global oil demand growth was in trend decline falling from growth rates of around the 8% mark down to growth rates of around 2%. Between 1985 and 2019 the growth rate has been more stable, averaging around 1.5%, with a high of 3.4% and a low of -1.2%. Global Oil Demand Growth (%) 10% 8% 6% 4%

2.2%

2% 0% -2% -4% -6%

2018

2016

2014

2012

2010

2008

2006

2004

2002

2000

1998

1996

1994

1992

1990

1988

1986

1984

1982

1980

1978

1976

1974

1972

1970

1968

1966

Source: BP Statistical Review of World Energy 2020

Fig. 12-1: Global Oil Demand Growth between 1966 and 2019 The swings in oil demand growth are largely attributable to swings in the strength of the global economy. As can be seen in Figure 12-2 below, there is a high, positive correlation between global economic growth and global oil demand growth. This tells us that the underlying or organic demand for oil is significantly affected by the global economy’s growth in production and income, consistent with the fact that oil is used as a key resource in the production of a wide variety of goods. Global Oil Demand Growth (%)

Oil Demand

GDP

World GDP Growth (%)

4

6

3

5

2 4 1 0

3

-1

2

-2 1 -3 0

-4

-1

-5 2019 2018 2017 2016 2015 2014 2013 2012 2011 2010 2009 2008 2007 2006 2005 2004 2003 2002 2001 2000 1999 1998 1997 1996 1995 1994 1993 1992 1991 1990 1989 1988 1987 1986 1985 1984 1983 1982 1981 1980 Source: BP Statistical Review of World Energy 2020, International Monetary Fund WEO database 2020

Fig. 12-2: Global Oil Demand and World GDP Growth

Chapter 12

310

As indicated in Figure 12-3 below, the top ten consumers of crude oil over calendar 2019 in terms of the number of barrels consumed per day (bpd) are the U.S.A. with 19,400,000 bpd, China with 14,056,000 bpd, India with 5,271,000 bdp, Japan with 3,812,000 bdp, Saudi Arabia with 3,788,000 bpd, Russia with 3,317,000 bpd, South Korea with 2,760,000 bpd, Canada with 2,403,000 bpd, Brazil with 2,398,000 bdp, and Germany with 2,281,000 bpd. The consumption of crude oil by country will depend on many forces, but the two most significant forces are the size of the population and its income per capita. Combined, these two forces represent the income or GDP of a country. The higher the combined population and income per capita of a country the higher the consumption of oil expected in that country. This is because higher income levels mean that consumers spend more on goods which require oil as an input into production (see Table 12-1), while they are also more likely to own a vehicle, which uses petroleum and diesel, and fly in aircraft, which uses jet kerosene. Another variable that is important in explaining why oil consumption varies across countries is the retail price of petroleum and oil in each country. This will depend heavily on the taxes that the relevant government imposes on the consumption of oil products. Governments such as the Saudi Arabian Government do not impose any taxes on the consumption of diesel and petroleum, which means that the price of petrol and diesel in Saudi Arabia is one of the cheapest in the world. According to Global Petrol Prices 2020, the price of petrol in Saudi Arabia as at November 16, 2020, was just 42 US cents per litre, which is the tenth cheapest price in the world and compares to US$2.26 in Hong Kong, US$1.79 in the Netherlands and US$1.64 in Italy. Consumption of oil in a particular country will also depend on the quality and the quantity of road infrastructure and the number of motor vehicles owned per capita in that country. The greater is the quality and the quantity of the road infrastructure in a country and the more vehicles that are owned by households and business, the more likely it is that oil is consumed in significant quantities for road transport, resulting in high levels of oil consumption per capita. According to the CIA World Factbook Roadwork 2021, the country with the greatest length of road in the world is the U.S.A. with 6,586,610km of road in 2012, followed by China with 4,960,600km of road in 2017, India with 4,699,024km of road in 2015 and Brazil with 2,000,000km of road in 2018. NationMaster Car Ownership 2021, indicates that the U.S.A. has the third highest level of car ownership in the world with 797 vehicles per 1,000 people in 2014, while China comes in at 111th with 83 vehicles per 1,000 people in 2014 and India is ranked 158th with 18 vehicles per 1,000 people in 2014. These road kilometre and car ownership data contribute to explaining the oil consumption rankings presented in Figure 12-3. The share of world oil consumption has changed dramatically over the past 5 decades. As indicated in Figure 12-4 below, China’s share of world oil consumption was just 1% in 1965, non-China Asia was 9% and the Middle East and Africa was 4%, while that for the U.S.A. was 44% and Europe was 31%. World oil consumption was therefore dominated by the U.S.A. and Europe, or Western economies, in 1965. If we fast-forward to 2020, China’s share of world oil consumption has increased to 15%, non-China Asia has increased to 22% and the Middle East and Africa has increased to 13%. The share for the U.S.A., however, has fallen to 21% and Europe to 18% as indicated in Figure 12-5 below. There has been a dramatic shift in the main source of oil consumption from the U.S.A. and Europe to China and other parts of Asia over the past 5 decades, consistent with the fact that a large proportion of world production of goods has been transferred to China and non-China Asia from the U.S.A. and Europe.

Quiz 12-1 Oil Demand 1.

Name the top 5 country consumers of oil in the world in 2019 and how much they consume on a daily average basis.

2.

(d)

What is the most likely explanation for the strong positive correlation between global oil demand growth and global GDP growth? Because oil is used in the production of a wide variety of goods. Because higher income levels result in households buying more goods that require the use of refined oil products. Because an increase in GDP results in an increase in household and business income, which in turn causes an increase in the production of goods and an increase in energy demand, including oil demand. All the above.

3. (a) (b) (c) (d)

Which of the following is NOT likely to drive the difference in crude oil consumption across countries? GDP. Car ownership per capita. Kilometres of roads. The spot price of oil.

(a) (b) (c)

Fig. 12-3: Oil Consumption by Country in 2019

0

5,000

10,000

15,000

20,000

25,000

2019 Oil Consumption ('000 barrels per day)

Source: BP Statistical Review of World Energy 2020

311 The Economics of Oil and Jet Fuel Markets

Iceland North Macedonia Estonia Latvia Trinidad & Tobago Uzbekistan Slovenia Cyprus Luxembourg Lithuania Croatia Slovakia Bulgaria Azerbaijan Sri Lanka Belarus Ireland Denmark Turkmenistan Bangladesh New Zealand Hungary Finland Norway Czech Republic Switzerland Ukraine Romania Ecuador Portugal Israel Peru Austria Sweden Morocco Oman Greece Qatar Colombia Kazakhstan Venezuela Chile Hong Kong Kuwait Pakistan Algeria Philippines Vietnam South Africa Argentina Poland Belgium Iraq Egypt Netherlands Malaysia Taiwan Turkey United Arab Emirates Australia Italy Spain Singapore Thailand France United Kingdom Indonesia Mexico Iran Germany Brazil Canada South Korea Russian Federation Saudi Arabia Japan India China US

Chapter 12

312

Oil Consumption Shares 1965 50%

44%

40% 31% 30% 20% 10%

10%

9% 4%

1%

1%

China

Oceania

0% USA

Europe

Non-US Non-China Americas Asia

ME and Africa

Source: BP Statistical Review of World Energy 2020

Fig. 12-4: Oil Consumption by Region in 1965 Oil Consumption Shares 2020 25%

22%

21% 18%

20%

15%

15%

13% 10%

10% 5%

1%

0% Non-China Asia

USA

Europe

China

ME and Africa

Non-US Americas

Oceania

Source: BP Statistical Review of World Energy 2020

Fig. 12-5: Oil Consumption by Region in 2020

12.2 Oil Supply The supply of oil is the quantity of oil that is offered to the crude market by oil producers. This supply is drawn from the inventory of oil producers, or it can be drawn from their current production. The production of oil is an important component of the supply of oil. Figure 12-6 below presents the growth over time in the world’s production of crude oil between 1966 and 2019, as well as the average growth rate over this period. Growth in World Oil Producion 15% 10% 5%

2.1%

0% -5% -10% 2018

2016

2014

2012

2010

2008

2006

2004

2002

2000

1998

1996

Fig. 12-6: World Oil Production Growth 1966 to 2019

1994

1992

1990

1988

1986

1984

1982

1980

1978

1976

1974

1972

1970

1968

1966

Source: BP Statistical Review of World Energy 2020

The Economics of Oil and Jet Fuel Markets

313

We can see in Figure 12-6 that oil production growth soared to double digit levels in the early 70’s then fell sharply during the oil price shocks of 1975 and 1980, which was part of a period of trend decline in production growth between 1966 and 1984. Thereafter production growth returned to more stable levels, with growth varying between -2% and +4% for most of that period. The production and the supply of crude oil is driven by both the same and different forces to the demand for crude oil. Both the demand for and the supply of crude oil are driven by the price of oil, albeit in different directions. An increase in the oil price causes oil demand to fall, but it also attracts oil producers such as Aramco and Rosneft to supply more crude oil to the global oil market. The oil production and supply decisions of oil producers will also depend on the forces that influence the underlying demand for crude oil, most notably GDP, and those forces that affect the cost of finding, extracting, and transporting crude oil, including the cost of equipment and wages. Figure 12-7 below presents oil production across countries in 2019. We can see in Figure 12-7 that the dominant producer of crude oil in 2019 was the U.S.A. with 17m barrels of crude oil per day. This is followed by Saudi Arabia with 11.8 million barrels per day, and Russia which produced 11.5m barrels per day in 2019. Production by country drops-off sharply after Russia, with Canada producing less than half of Russia’s oil output with 5.6m barrels per day and Iraq producing 4.8m barrels per day. To the extent that the countries which feature to the left of Figure 12-7 produce more oil than they consume then the excess of production over consumption of oil will be exported. This is most certainly the case for Saudi Arabia and Russia, which are both big exporters of crude oil. 2019 Oil Prodution ('000 barrels per day) 18000 16000 14000 12000 10000 8000 6000 4000 2000 -

Syria Tunisia Uzbekistan Romania Trinidad & Tobago Italy Yemen Sudan Denmark Brunei Chad South Sudan Peru Equatorial Guinea Gabon Vietnam Turkmenistan Republic of Congo Thailand Australia Ecuador Argentina Malaysia Egypt Azerbaijan Indonesia India Colombia Venezuela Oman United Kingdom Libya Angola Algeria Norway Qatar Mexico Kazakhstan Nigeria Brazil Kuwait Iran China United Arab Emirates Iraq Canada Russian Federation Saudi Arabia US Source: BP Statistical Review of World Energy 2020

Fig. 12-7: Oil Production by Country Figures 12-8 and 12-9 below present the changing pattern of crude oil production by region between 1965 and 2020. Oil Production Shares 1965 35% 30%

32% 26%

25% 20%

15%

15%

14% 7%

10% 5%

3%

3%

Asia Pacific

Europe

0% North Middle East Euroasia America Source: BP Statistical Review of World Energy 2020

Fig. 12-8: Oil Production by Region in 1965

Latin America

Africa

Chapter 12

314

Oil Production Shares 2020 35% 30% 25% 20% 15% 10% 5% 0%

32% 26% 15% 9%

Middle East

North America

Euroasia

Africa

8%

Asia Pacific

6%

Latin America

4% Europe

Source: BP Statistical Review of World Energy 2020

Fig. 12-9: Oil Production by Region in 2020 We can see in Figure 12-8 that the dominant producer of crude oil in 1965 was the United States at 32% of world production, followed by the Middle East with 26% and then Eurasia with 15%. If we move forward to 2020 the first two positions have switched, with the Middle East producing 32% of global oil production and the United States producing 26%. Africa has slightly lifted its share of crude oil production from 7% to 9% and the Asia Pacific has more than doubled its production share. The large lift in the share of production by Middle East producers and the sharp increase in the demand for crude oil by the Asia Pacific and North America evidences the importance of the transport of crude oil across the globe from the major crude oil producing zones to the consumption zones. An important player in the global supply of crude oil is OPEC. OPEC stands for the Organisation of Petroleum Exporting Countries. It was founded in Baghdad, Iraq, with the signing of an agreement in September 1960 by five countries – the Islamic Republic of Iran, Iraq, Kuwait, Saudi Arabia, and Venezuela. These countries were the founding members of OPEC and were later joined by Qatar (1961), Indonesia (1962), Libya (1962), the United Arab Emirates (1967), Algeria (1969), Nigeria (1971), Ecuador (1973), Gabon (1975), Angola (2007), Equatorial Guinea (2017) and Congo (2018). OPEC was formed so that the member countries as a single entity have greater power over the outcome for the market price of oil by controlling a high proportion of global oil production and supply. By having greater power over output, the member countries could secure the greatest possible price of oil which in turn would help to support the government finances and the economic outcomes for those economies, particularly the developing economies that are members of OPEC (OPEC History 2021). Figure 12-10 below presents the movement over time in OPEC production of crude oil and Figure 12-11 presents the percentage of global output that is produced by OPEC. OPEC Oil Production ('000 bbl per day) 40,000 35,000 30,000 25,000

25,995

20,000 15,000

15,731

10,000 2019 2017 2015 2013 2011 2009 2007 2005 2003 2001 1999 1997 1995 1993 1991 1989 1987 1985 1983 1981 1979 1977 1975 1973 1971 1969 1967 1965 Source: BP Statistical Review of World Energy 2020

Fig. 12-10: OPEC Oil Production – 1965 to 2019

The Economics of Oil and Jet Fuel Markets

315

OPEC Oil Production (% World Production) 55%

50.6%

50% 45% 40% 35% 30%

27.4%

25% 20%

2019 2017 2015 2013 2011 2009 2007 2005 2003 2001 1999 1997 1995 1993 1991 1989 1987 1985 1983 1981 1979 1977 1975 1973 1971 1969 1967 1965 Source: BP Statistical Review of World Energy 2020

Fig. 12-11: OPEC Oil Production as a Percentage of Total World Production We can see in Figure 12-10 that OPEC oil production has increased from just under 15m barrels per day in 1965 up to just over 35m barrels per day in 2019. Major cuts to OPEC production can be seen between 1980 and 1985, with OPEC output falling from 30 million barrels per day to around 16 million barrels per day. There was also a major cut in production by OPEC in 1975 from 30m barrels per day to 26m. In Figure 12-11 we can see that OPEC production as a percentage of global production has fallen from a high of over 50% in 1973 down to 37% by 2019, with an absolute low of 27% in 1985 in response to the sequence of OPEC cuts to output over the early 1980s. While OPEC’s importance in world oil production has fallen over the period examined in Figures 12-10 and 12-11 it remains a dominant contributor to global oil production and supply.

12.3 Oil Pricing Benchmarks There are many key benchmarks that are used to describe the market price of oil in the global economy. They include Brent crude oil, which is a key crude oil pricing benchmark used in Europe, West Texas Intermediate or WTI crude oil, which is a key crude oil pricing benchmark used in the Americas, Dubai crude oil, which is a key crude oil pricing benchmark used in the Middle East, Nigerian Forcados, which is a key crude oil pricing benchmark used in Africa, and Tapis crude oil, which is a key crude oil pricing benchmark used in Asia. These benchmarks can be interpreted as the spot price or the market price of crude oil in the respective locations or geographies around the world. They are used to set prices in contracts between oil producers and oil refineries, and they are used by airlines in contracts to hedge against the movement in jet fuel prices. The different characteristics of these crude oil benchmarks are presented in Table 12-2 below. Crude Benchmark

API Gravity

Specific Gravity 0.827

Sulphur Content 0.24%

Description

Origin

West Texas Intermediate Brent Crude Dubai Crude OPEC Reference Basket Minas Tapis

39.6

Light and Sweet

Southwestern USA

38.1 31 32.7

0.835 0.871

0.37% 2% 1.77%

Light and Sweet Light and Sour Light and Sour

35 45.2

0.8498

0.08% 0.0343%

Light and Sweet Very light and very sweet Light and Sweet Light and Sour Light and Sweet

North Sea UAE Weighted Average of OPEC Members Island of Sumatra Malaysia

Bonny Light Isthmus-34 Light Nigerian Forcados

32.9 33.74 35

0.16% 1.45% 0.2%

Nigeria Mexico Nigeria

Source: George and Breul 2014.

Table 12-2: Characteristics of Different Crude Oil Benchmarks Table 12-2 compares the crude oil benchmarks in terms of their API gravity, specific gravity, sulphur content and the origin of the benchmark. The specific gravity of crude oil is the ratio of the mass of the crude oil benchmark compared to the mass of an equal volume of water. A specific gravity of 0.8 for a crude product means that the ratio of the mass

Chapter 12

316

of the crude oil relative to the mass of the same volume of water is 0.8 or the weight of the crude oil is 80% of the weight of water. API Gravity stands for American Petroleum Institute gravity. This is also a measure of how heavy crude oil is relative to the weight of water. If the API gravity of an object is greater than 10 then it floats on water or it is lighter than water, while if it is less than 10 then it is heavier than water and sinks. The higher is the API gravity the lighter is that product relative to water. It follows that a crude product that has a low specific gravity and a high API gravity is referred to as a light crude product, and a crude product that has a high specific gravity and a low API gravity is a heavy crude product. The sulphur content refers to the percentage weight of the oil benchmark that is attributable to the element sulphur. If a crude oil benchmark has relatively high sulphur content, it is referred to as sour and if it has relatively low sulphur content it is referred to as sweet. When a crude oil benchmark is sour or has a high sulphur content this is undesirable because, when burnt, it generates sulphur oxides which are released into the atmosphere. This sulphur must be removed as a part of the refining process, which means that the use of sour crude oil is more expensive for oil refineries to refine. West Texas Intermediate crude oil is a high-quality crude oil product because it is light in weight with a specific gravity of 82.7% and has a low sulphur content of just 0.24%. For these reasons, it is often referred to as light, sweet crude oil. These properties make it excellent for making gasoline and is the key reason why it is the major benchmark crude oil in the Americas. Brent crude oil is a combination of crude oil from more than a dozen different oil fields in the North Sea. It is less light and sweet than WTI, with specific gravity of 83.5% and sulphur content of 0.37% but still excellent for making gasoline. It is refined in Northwest Europe and is the primary benchmark for crude oil in Europe or Africa (Chen 2020). Dubai crude oil is a medium sour crude that is extracted from the oil fields of Dubai. It is used as a benchmark for pricing oil that is produced in the Persian Gulf and destined for Asia. It has a specific gravity of 87.1% which makes it heavier than Brent and WTI and a sulphur content of 2% which makes its sulphur content the highest of all benchmarks in Table 12-2 (Platts Dubai Crude 2021). Tapis crude oil is a Malaysian crude oil that is extracted offshore in the Malay Basin. Tapis is a very light, sweet crude oil, having the lowest sulphur count of all the oil benchmarks in Table 12-2 at 0.0343% and an API gravity of 45.2, which is the highest of all the crude benchmarks in Table 12.2. Malaysian Tapis is often used as a benchmark in Asian and Australian oil trade and is usually priced higher than WTI and Brent because it is lighter and sweeter, which means that it can be used to produce higher value refined products. Its lower sulphur content also means that the use of Malaysian Tapis requires less processing than other crudes that have a higher sulphur content (Energy Information Administration Malaysian Tapis 2021). Nigerian Forcados crude oil is a very light, sweet crude oil with an API gravity of 35 and a sulphur content of 0.2%. It is extracted from the Southern region of Nigeria near to the Bight of Benin. It is one of several oil benchmarks that are used in Nigeria and wider Africa, including Bonny Light crude oil, which is another light and sweet crude product from Nigeria (Nigerian Forcados Crude 2021). In general, the price of the crude oil benchmarks presented in Table 12-2 are highly correlated, but they can deviate from one another periodically if the crude oil supply and demand conditions in particular geographic catchments change dramatically – refer to Figure 12-12 below. Crude Oil Benchmark Prices (US$/bbl) $120

Dubai

Brent

Forcados

WTI

$100 $80 $60 $40 $20 $0 2019 2018 2017 2016 2015 2014 2013 2012 2011 2010 2009 2008 2007 2006 2005 2004 2003 2002 2001 2000 1999 1998 1997 1996 1995 1994 1993 1992 1991 1990 1989 1988 1987 1986 1985 1984 1983 1982 1981 1980 1979 1978 1977 1976 Source: BP Statistical Review of World Energy 2020

Fig. 12-12: Crude Oil Global Benchmark Prices We can see in Figure 12-12 that the price of all crude oil benchmarks cycled around a fixed mean of around US$20/bbl until early 2000, and then started to cycle around an aggressive upward trend thereafter. The long run or trend growth rate in oil prices after the early 2000 period (between 2003 and 2019) is estimated to be 5.5% in the case of Dubai crude, 5.1% in the case of Brent crude, 5.2% in the case of Nigerian Forcados, and 4.0% in the case of WTI crude oil. This means that the long run growth rate in these four oil price benchmarks between 2004 and 2019 has been strong and is a key reason behind airlines struggling all over the world to consistently make money. Oil prices are

The Economics of Oil and Jet Fuel Markets

317

influenced by both global and local demand and supply-side forces. The global forces tend to dominate the movement in oil prices for most periods of time, but in early 2011 through to 2014 local demand and supply forces disrupted the close relationship between the WTI price and the other crude oil prices. WTI crude oil is extracted from the oil fields of Texas, Louisiana, and North Dakota, in the main part, and is then transported via a pipeline to Cushing, Oklahoma. The local demand and supply conditions in the case of WTI therefore pertain to the conditions in North America, with WTI being the major crude oil benchmark for North America. Brent crude is extracted from the oil fields of the North Sea and is a blend of four different crude oils – Brent, Forties, Osberg and Ekofisk. Brent crude is the benchmark for Africa, Europe, and the Middle East. Its value therefore fluctuates with the local demand and supply conditions in this part of the world. Over the period 2011 to 2014, the Brent crude oil price shot upward and the price of WTI was relatively stable. The higher Brent price was due to significant political turmoil in Egypt, Libya, Yemen, and Bahrain, in conjunction with supply problems in Europe, including lower North Sea production. This resulted in excess demand for Brent which placed upward pressure on the Brent price. There was also continued threats by Iran to close a major shipping lane for the transport of oil from the Middle East to other parts of the world because of the trade sanctions imposed on it, which placed even greater upward pressure on the Brent price. It is also the case that there was significant excess supply of light, sweet crude oil in North America because of the rapid increases in production of North American shale oil. The refinery infrastructure in North America could not keep pace with the production of shale oil, resulting in a large build-up of light, sweet crude oil in this part of the world. This excess supply of light, sweet oil placed initial downward pressure on the WTI price which stabilised at lower levels. The higher Brent price and the slightly lower WTI price resulted in a significant gap opening-up between the Brent price and the WTI price.57

Quiz 12-2 Crude Oil Price Benchmarks

57

1. (a) (b) (c) (d)

Which of the following is NOT a crude oil benchmark? Brent Tapis WTI Gulf Coast

2. (a) (b) (c) (d)

What do we mean by a sweet crude oil? It is sweet to taste. It is combined with sugar in the refining process. It is low in sulphur. It is high in heavy metals.

3. (a) (b) (c) (d)

What do we mean by a light crude oil? It has a relatively low specific gravity. It has a relatively low API gravity. It has a low sulphur content. It has a high sulphur content.

4. (a) (b) (c) (d)

Which of the following represents an estimate of the specific gravity of oil? Divide the weight of 1 gallon of oil by the weight of 1 gallon of water. Divide the volume of 1 kg of oil by the volume of 1 kg of water. Divide the weight of 1 gallon of water by the weight of 1 gallon of oil. Divide the volume of 1 kg of water by the volume of 1 kg of oil.

5. (a) (b) (c) (d)

Which of the following crude products is the sweetest? Brent West Texas Intermediate Tapis Dubai

6. (a) (b) (c) (d)

Which of the following crude products is the lightest? Brent West Texas Intermediate Tapis Dubai

If you would like to know more about the difference between the movements in Brent and WTI crude oil prices between 2011 and 2014, read the interesting articles by Hecht, 2019, Amadeo 2019, and Bradfield 2018.

Chapter 12

318

7. (a) (b) (c) (d)

Which of the following crude products is based out of Europe? Brent West Texas Intermediate Tapis Dubai

8. (a) (b) (c) (d)

Which of the following crude products is based out of Asia? Brent West Texas Intermediate Tapis Dubai

9.

Provide an explanation as to why the price of West Texas Intermediate crude fell below the other main crude oil benchmarks between 2011 and 2014.

12.4 An Economic Model of Price Outcomes in the Oil Market In this section I present an economic model of the oil market. The model is built to obtain a better understanding of the forces that are likely to determine outcomes for the market price of oil and the quantity supplied to the global oil market. The economic model will borrow from much of the content presented in sections 12.1 through to 12.3 of this chapter to shape the parts of the model that describe the demand for oil and competition in the oil market.

12.4.1 Theory and Analytics In this economic model, we assume that the global oil market consists of one large oil producing company and a group of smaller oil producers. The single, large producer attempts to simulate OPEC in the market for oil and the large number of small producers captures all other oil producing companies located in countries outside of OPEC. The economic field of industrial organisation refers to this type of model as a Dominant Firm/Fringe Firm model of competition (Shepherd 1997, 206). We assume that the global oil demand function is described by the following linear function of global GDP and the oil price: DOil = a0 + a1GDP + a2POil

(12.1)

This global oil demand function says that the global demand for oil depends on global GDP and the price of oil. We expect the coefficient attached to global GDP, a1, will be positive, meaning that an increase in GDP leads to an increase in the demand for oil, and the coefficient attached to the price of oil, a2, is expected to be negative, meaning that an increase in the oil price leads to a reduction in the demand for oil. The small producers of oil are price-takers in the global oil market. This means that an increase in oil production by any one of these individual producers is not expected to have an impact on the market clearing global oil price. The supply of oil by the aggregate of the small oil producers is described by the following linear function: QS = a3 u POil

(12.2)

Equation (12.2) simply says that the quantity of oil supplied by the competitive fringe producers of oil depends on the price of oil. We expect that the coefficient attached to the price of oil in (12.2) is positive or a3 > 0, meaning that as the oil price increases this attracts the fringe producers of oil to supply more to the market. The demand that is faced by the dominant firm is the market demand (12.1) minus the supply of the oil producing fringe. This is referred to as the residual demand function of the dominant firm and is equal to (12.1) minus (12.2) as follows: Dୖ୓୧୪ = a0 + a1GDP + a2POil – a3POil = a0 + a1GDP + (a2 – a3)POil

(12.3)

The production cost of the dominant firm is equal to the oil production of the dominant firm multiplied by the marginal ோ cost of oil cOil plus fixed costs KOil. Interpreting ‫ܦ‬ை௜௟ as the output of the dominant firm then the cost of the dominant firm can be represented by the following linear function: CDominant = cOil u Dୖ୓୧୪ + KOil

(12.4)

Multiplying the residual demand function (12.3) by the oil price generates the dominant firm’s revenue function. Subtracting the cost function (12.4) from the dominant firm’s revenue function derives the profit function of the dominant firm as follows:

The Economics of Oil and Jet Fuel Markets

SDominant = [a0 + a1GDP + a2POil – a3POil](POil – cOil) – KOil

319

(12.5)

We can see that the profit function of the dominant firm depends on the price of oil. In fact, given that a2 is less than zero, equation (12.5) says that the profit function of the dominant firm is a concave down function of the price of oil. This means that as the price of oil increases the profit of the dominant firm increases, reaches a maximum and then decreases. This implies that we can find the price of oil that maximises the dominant firm’s profit function at (12.5). The market clearing or equilibrium oil price is the oil price that maximises the profit function (12.5) of the dominant firm. To find this price, the first step is to differentiate (12.5) with respect to the oil price and set the result equal to zero as follows: ୢ஠ ୢ୔ో౟ౢ

= a଴ + aଵ GDP + 2aଶ P୓୧୪ െ 2aଷ P୓୧୪ െ (aଶ െ aଷ )c୓୧୪ = 0

(12.6)

The profit maximising or equilibrium condition equation (12.6) can be written in the following form: a଴ + aଵ GDP + 2(aଶ െ aଷ )P୓୧୪ = (aଶ െ aଷ )c୓୧୪

(12.7)

The left-hand side of the equilibrium condition (12.7) is the marginal revenue of the dominant firm. This measures the response of the dominant firm’s revenue to a change in the oil price. The right-hand side of (12.7) is the marginal oil cost of the dominant firm. The marginal oil cost represents the impact of an increase in the oil price on residual demand, and thus variable costs, of the dominant firm. As marginal cost is negative because (a2 – a3) < 0, then marginal revenue (MR) must also be negative in equilibrium. For MR to be negative this requires the left-hand side of (12.7) to be negative, which in turn requires the following condition to hold: a଴ + aଵ GDP + 2(aଶ െ aଷ )P୓୧୪ < 0

(12.8)

For (12.8) to hold this requires the oil price to be sufficiently high, as described by the following condition: ‫כ‬ > P୓୧୪

ୟబ ାୟభ ୋୈ୔

(12.9)

ଶ(ୟయ ିୟమ )

The condition (12.9) effectively says that the dominant firm must set the oil price at a sufficiently high level to make its residual demand elastic to the oil price. If the dominant firm were to set a price that is below the level of (12.9) then it would earn more revenue by restricting oil production and increasing the oil price. If we solve first order condition (12.9) for the price of oil, we obtain the following profit maximising oil price: POil =

ୡో౟ౢ (ୟమ ିୟయ )ିୟబ ିୟభ ୋୈ୔ ଶ(ୟమ ିୟయ )

= Ⱦ଴ + Ⱦଵ GDP + Ⱦଶ c୓୧୪

(12.10)

We use the E parameters on the right-hand side of (12.10) to simplify the expression for the equilibrium oil price, with these E terms taking on the specific definitions E0 { -a0/[2(a2 – a3)] and E1 { -a1/[2(a2 – a3)]. You should try and make sure that you know where these E terms come from in your free time. We can see from (12.10) that the dominant drivers of the equilibrium oil price are the component of the underlying demand for oil that is driven by world economic activity (a1GDP), the component of the underlying demand for oil that is driven by all other forces aside from economic activity, such as attitudes to the environment and the number of motor vehicles owned per capita (a0), the marginal cost of finding, extracting, transporting and storing crude oil (cOil), the sensitivity of oil demand to a change in the price of oil (a2), and the sensitivity of the supply of the non-OPEC producers to a change in the price of oil (a3). Let us now demonstrate how we find the equilibrium oil price by using diagrams. Consider Figure 12-13 below. In Figure 12-13 the marginal revenue of the dominant OPEC firm is described by the downward sloping line. The slope of this line is 2(a2 – a3) and the vertical intercept is a0 + a1GDP. The horizontal line is the marginal cost which has a slope of zero and a vertical intercept of (a2 – a3)cOil. The point of equilibrium or maximum profit is the point where the marginal revenue line meets the marginal cost line at E. You will note that the marginal revenue line meets the marginal cost line at a point where marginal revenue is negative, or the MC line is below the horizontal axis. This satisfies the condition presented at (12.9) above, which says that the price of oil must be sufficiently high in equilibrium to ensure marginal revenue is negative (or oil demand is elastic to the price). Let us now use the concepts in Figure 12-13 to describe what happens to the equilibrium oil price when GDP is stronger in the global economy, increasing from GDP0 to GDP1 – refer to Figure 12-14 below. An increase in GDP results in the marginal revenue curve shifting to the right from MR0 to MR1 (the firm downward sloping line becoming the dashed downward sloping line in Figure 12-14). For a given net marginal cost curve (the horizontal MC line) the ଴ ଵ point of equilibrium shifts from E0 to E1 and the equilibrium oil price rises from ܲை௜௟ to ܲை௜௟ . An increase in global GDP which increases the underlying demand for oil results in an increase in the oil price.

Chapter 12

320

MR and MC a0 + a1GDP MR = a0+a1GDP+2(a2-a3)POil

0

a଴ + aଵ GDP c୭୧୪ െ൤ ൨ 2(aଶ + aଷ ) 2 ܽ଴ + ܽଵ ‫ܲܦܩ‬ 2(ܽଷ െ ܽଶ )

POil MC = (a2 – a3)cOil

E

Fig. 12-13: Equilibrium Price in the Dominant-Firm/Fringe Firms Oil Model

MR and MC a0 + a1GDP1 MR1 a0 + a1GDP0 MR0 0

଴ ܲை௜௟

ଵ ܲை௜௟

E0

E1

POil

MC = (a2 – a3)cOil coil

Fig. 12-14: Impact on Equilibrium Oil Prices of an Increase in GDP In the case in which the competitive fringe increases supply for a given increase in the oil price, or a3 increases, this results in a change in the slope of the MR line and a shift in the MC line as presented in Figure 12-15 below. In Figure 12-15, the marginal revenue line of the dominant OPEC firm pivots inward from MR0 to MR1 while the marginal cost line of the dominant OPEC firm shifts downward from MC0 to MC1. This results in the new equilibrium point moving ଴ ଵ to ܲை௜௟ . from E0 to E1 and the equilibrium oil price falling from ܲை௜௟

12.4.2 Numerical Illustration The global demand for crude oil defined in millions of barrels of crude consumed per day is described by the following linear function: DOil = 14.74 + 1.126GDP – 0.23POil

(12.11)

where GDP is world GDP defined in trillions of US dollars and POil is the price of oil in US dollars per barrel. Non-

The Economics of Oil and Jet Fuel Markets

321

MR and MC a0 + a1GDP

MR0

MR1

଴ ܲை௜௟

ଵ ܲை௜௟

0

E0

E1

POil

MC0

MC1

Fig. 12-15: Impact on Equilibrium Oil Prices of More Sensitive Fringe Firm Supply OPEC oil supply is described by the following function: QS = 0.43POil

(12.12)

where QS is defined in millions of barrels per day. The residual demand function of the dominant oil producing firm, OPEC, is equal to the demand function (12.11) minus the fringe supply (12.12), which is equal to: Dୖ୓୧୪ = 14.74 + 1.126GDP – 0.23POil - 0.43POil

(12.13)

The marginal cost of production of OPEC is described generally by the parameter cOil and fixed costs by KOil. This information combined with (12.13) can be used to derive the profit function of the dominant OPEC firm: S = (14.74 + 1.126GDP – 0.23POil - 0.43POil)(POil - cOil) – KOil

(12.14)

The first order condition for maximising OPEC profit is found by differentiating (12.14) with respect to POil and setting the result equal to zero: ୢ஠ ୢ୔ో౟ౢ

= 14.74 + 1.126GDP െ 1.32P୓୧୪ + 0.66c୓୧୪ = 0

(12.15)

The equilibrium condition (12.15) can be written as the following marginal revenue equals marginal cost condition: 14.74 + 1.126GDP െ 0.46P୓୧୪ െ 0.86P୓୧୪ = െ0.66c୓୧୪

(12.16)

The left-hand side of (12.16) is marginal revenue and the right-hand side of (12.16) is marginal cost. This is described in Figure 12-16 below. The equilibrium oil price is found by solving (12.16) for POil, which gives: ‫כ‬ = P୓୧୪

ଵସ.଻ସାଵ.ଵଶ଺ୋୈ୔ା଴.଺଺ୡో౟ౢ

(12.17)

ଵ.ଷଶ

If we assume that world GDP is US$90 trillion, and the dominant firm marginal cost of oil is cOil = US$15 then the equilibrium oil price is: ‫כ‬ P୓୧୪ =

ଵସ.଻ସାଵ.ଵଶ଺ୋୈ୔ା଴.଺଺ୡో౟ౢ ଵ.ଷଶ

=

ଵସ.଻ସାଵ.ଵଶ଺(ଽ଴)ା଴.଺଺(ଵହ) ଵ.ଷଶ

= US$95.44

Let us now examine what happens if world GDP increases, as described in section 12.4.1 above. If we now assume that world GDP is equal to US$100 trillion, then the new equilibrium price is higher at:

Chapter 12

322

MR and MC 14.74 + 1.126GDP

14.74 + 1.126‫ܲܦܩ‬ ܿ௢௜௟ െ൤ ൨ െ1.32 2

0

POil

-0.66cOil

E MR = 14.74+1.126GDP-1.32POil Fig. 12-16: Impact on Equilibrium Oil Prices of More Sensitive Fringe Firm Supply

‫כ‬ P୓୧୪ =

ଵସ.଻ସାଵ.ଵଶ଺ୋୈ୔ା଴.଺଺ୡో౟ౢ ଵ.ଷଶ

=

ଵସ.଻ସାଵ.ଵଶ଺(ଵ଴଴)ା଴.଺଺(ଵହ) ଵ.ଷଶ

= US$103.97

In response to an 11.1% increase in world GDP, this results in an increase in the equilibrium oil price by 8.9% according to this numerical model. What happens if the Non-OPEC competitive fringe become more reactive to the price of oil? For example, what happens if more firms enter the competitive fringe? To answer this question, we assume that the non-OPEC supply function becomes: QS = 0.77POil

(12.18)

You will notice that the coefficient attached to the oil price in (12.18) is bigger than the same coefficient in the earlier Non-OPEC competitive fringe supply equation at (12.12). The first order condition for a maximum becomes: ୢ஠ ୢ୔ో౟ౢ

= 14.74 + 1.126GDP െ 2P୓୧୪ + 0.66c୓୧୪ = 0

(12.19)

Rearranging (12.19) for the price of oil assuming World GDP = 90 yields: ‫כ‬ = P୓୧୪

ଵସ.଻ସାଵ.ଵଶ଺(ଽ଴)ା଴.଺଺(ଵହ) ଶ

= US$62.99

This indicates as the competitive fringe become more reactive this leads to a lower outcome for the equilibrium oil price.

Quiz 12-3 The Oil Market 1.

What is the explanation behind the price of Brent crude oil rising relative to WTI crude oil between 2011 and 2014?

2. (a) (b) (c) (d)

Which countries were the world’s three biggest producers of crude oil in 2019? Iraq, Saudi Arabia, Kuwait. U.S.A., Saudi Arabia, Russia. Saudi Arabia, UAE, Kuwait. Saudi Arabia, Iraq, Iran.

3. (a) (b) (c) (d)

What does OPEC stand for? Organisation of the Petroleum Exporting Countries. Organisation of the Petroleum Extracting Countries. Oil Producing Exporting Countries. Organisation of the Petroleum Exporting Corporations.

The Economics of Oil and Jet Fuel Markets

323

4. (a) (b) (c) (d)

In which country was OPEC founded? Saudi Arabia Kuwait Iraq Iran

5. (a) (b) (c) (d)

Which country is NOT a member of OPEC? Venezuela Iraq Kuwait U.S.A.

6. (a) (b) (c) (d)

What is a key characteristic of the dominant firm/fringe firm model of pricing in the oil market? There is a single dominant firm and a single fringe firm producing oil. There are many dominant firms and a single fringe firm producing crude oil. There is a single dominant firm and many fringe firms producing crude oil. There are many dominant firms and many fringe firms producing crude oil.

7. (a) (b) (c) (d)

What is the residual demand function in the dominant firm/fringe firm model of the oil market? Market demand less the demand of the dominant firm. Market demand less the supply of the competitive fringe. Market demand less the demand of the competitive fringe. Demand of the dominant firm less the demand of the competitive fringe.

8.

Which of the following assumptions best describes each firm in the competitive fringe in the dominant firm/fringe firm model? They are price-takers. They are price-makers. They are price-setters. They set price as a mark-up over unit costs.

(a) (b) (c) (d) 9. (a) (b) (c) (d)

In the dominant firm/fringe firm model of the oil market the market demand curve is D = a0 + a1P and the supply of the competitive fringe is QS = bP. What is the residual demand function of the dominant firm? DR = a0 + (a1 + b)P DR = a0 + (a1 – b)P DR = a0 + (b - a1)P DR = a0 + a1P

10. Which of the following is NOT likely to be a direct determinant of the equilibrium oil price in the dominant firm/fringe firm model of the oil market? (a) The underlying demand for oil. (b) The marginal oil producing cost of the dominant firm. (c) The supply responsiveness of the competitive fringe. (d) Fixed costs of oil exploration. The global market demand for oil is described by the following equation defined in millions of barrels per day DOil = 120 - 0.4POil, where POil is defined in terms of US$ per barrel. There is one dominant firm in the oil market and many firms that represent the competitive fringe. The supply of the competitive fringe defined in millions of barrels per day can be described by the following linear function SF = 0.8POil. The marginal oil production cost of the dominant firm is $15 per barrel while the fixed costs are K. Use this information to answer the following 7 questions. 11. (a) (b) (c) (d)

What is the interpretation of the coefficient attached to the price of oil in the global market oil demand function? A US$0.4 per barrel increase in the price of oil causes a 1m barrel per day decrease in oil demand. A 1% increase in the price of oil causes a 0.4% decrease in the demand for oil. A US$1 per barrel increase in the price of oil causes a 400,000 barrel per day decrease in the demand for oil. A US$0.4 per barrel increase in the price of oil causes a 400,000 barrel per day decrease in the demand for oil.

12. (a) (b) (c) (d)

What is the residual demand function of the dominant firm? Dୖ୓୧୪ = (120 – 0.8) – 0.4POil Dୖ୓୧୪ = 120 – (0.8 - 0.4)POil Dୖ୓୧୪ = 120 – (0.4 - 0.8)POil Dୖ୓୧୪ = 120 – (0.4 + 0.8)POil

Chapter 12

324

13. (a) (b) (c) (d)

What is the revenue function of the dominant firm? ଶ Dୖ୓୧୪ = 120 – (0.4 + 0.8)P୓୧୪ ୖ ଶ D୓୧୪ = 120POil – 0.4P୓୧୪ Dୖ୓୧୪ = 120POil – (0.8 - 0.8)POil ଶ Dୖ୓୧୪ = 120POil – (0.4 + 0.8)P୓୧୪

14. (a) (b) (c) (d)

What is the cost function of the dominant firm? C = 15Dୖ୓୧୪ + K C = 15DOil + K C = 15DOil C = (15 + K)DOil

15. What is the profit function of the dominant firm? ଶ ଶ (a) S = 120P୓୧୪ – (0.4 - 0.8)P୓୧୪ - 15Dୖ୓୧୪ – K ଶ (b) S = 120POil + (0.4 + 0.8)P୓୧୪ - 15Dୖ୓୧୪ – K ଶ (c) S = 120POil – (0.4 + 0.8)P୓୧୪ - 15Dୖ୓୧୪ – K ଶ (d) S = 120 – (0.4 + 0.8)P୓୧୪ - 15Dୖ୓୧୪ – K 16. What is the first order condition associated with the maximisation of the dominant firm’s profit? ୢ஠ (a) = 120 – 2 u (0.4 + 0.8) u POil – 15 = 0 (b) (c) (d) 17. (a) (b) (c) (d)

ୢ୔ో౟ౢ ୢ஠ ୢ୔ో౟ౢ ୢ஠ ୢ୔ో౟ౢ ୢ஠ ୢ୔ో౟ౢ

= 120 – 2 u 0.4 u POil – 15 u (0.4 + 0.4) = 0 = 120 – 2 u (0.4 + 0.8) u POil – 15 u (0.4 + 0.8) = 0 = 120 – (0.4 + 0.8) u POil – 15 = 0

What is the solution for the equilibrium oil price? ‫כ‬ = 42.50 P୓୧୪ ‫כ‬ P୓୧୪ = 50.00 ‫כ‬ P୓୧୪ = 51.75 ‫כ‬ P୓୧୪ = 45.30

12.5 The Jet Fuel Market 12.5.1 Refining Process Crude oil has little use in the form in which it is extracted from the beneath the surface of the earth. To be converted into a form that can be used, such as gasoline, diesel, or jet fuel, it must be sent to an industrial facility called an oil refinery. The first step in the conversion of crude oil into a refined product is the transportation of the crude oil to the refinery by ships or pipeline. Once the crude oil arrives at the refinery, it is then heated at different temperatures in a piece of infrastructure called an atmospheric distillation unit. When the crude oil is heated at different temperatures it separates the crude oil into what is called fractions. In simple refineries, these different fractions or components are processed further to remove impurities such as sulphur. After the impurities have been removed the fractions are then blended with other products and additives to generate final refined products. The refined products are stored in large tank farms, ready to be transported to end users by a combination of pipelines, rail, road, and waterways. Figure 12-17 below presents the different refined products that are generated from heating crude oil, and the boiling temperature at which the different refined products are separated. We can see in Figure 12-17 that the product with the lowest boiling point is the light product butane. This is followed by gasoline, naphtha, jet fuel and kerosene. Diesel and heating oil require a higher boiling point of 350 to 450 degrees Fahrenheit, while heavier products such as heavy gas oil and residual fuel require the crude oil to be heated at even higher temperatures. According to the Energy Information Administration Refining Process 2020, the conversion of crude oil into refined products at U.S. refineries in 2020 resulted in the share of refined products presented in the bar chart of Figure 12-18 below. The bar chart indicates that motor gasoline represented the highest share of refined product in the U.S., market at 43%, followed by distillate fuel oil at 30%, jet kerosene at 7%, coke at 5% and still gas at 4%.

The Economics of Oil and Jet Fuel Markets

325

Boiling Range Lighter (Low Boiling Point)

Products

< 85 F

Butane & Lighter

85 - 185 OF

Gasoline blending components

O

Naphtha

185 - 350 OF Crude Oil

Distillation Unit

Heavier (High Boiling Point)

350 - 450 OF

Kerosene, Jet Fuel

450 - 650 OF

Distillate (diesel, heating oil)

650 - 1050 OF

Heavy gas oil Residual fuel oil

> 1050 OF

Source: Energy Information Administration Refining Process 2021

Fig. 12-17: Stylised Version of the Distillation Process US Refined Production Shares 2020 43% 30%

4%

4%

3%

2%

1%

1%

Other

Gas Liquids

Asphalt

Naptha

Lubricants

Jet Kerosene

Distillate Fuel Oil

Motor Gasoline

5%

Still Gas

7%

Coke

50% 45% 40% 35% 30% 25% 20% 15% 10% 5% 0%

Source: Energy Information Administration Refining Process 2021

Fig. 12-18: U.S. Refined Product Production in 2020

12.5.2 Jet Fuel Benchmark Prices There are three main jet fuel benchmarks in the global economy that are used to price the jet kerosene that is consumed by the world’s airlines. The first is the Asian benchmark Singapore Jet Kerosene, which is used throughout Asia by airlines and refineries to settle jet kerosene contracts. The Singapore Jet Kerosene price is usually defined on a US dollars per barrel basis. Oil refineries in Singapore together represent the fifth largest refineries in the Asia Pacific. Singapore itself is one of the global leaders in oil refining and oil-trading. The second major jet fuel benchmark is the Barges FOB (Full on Board) ARA (Amsterdam-Rotterdam-Antwerp) jet fuel, which is denominated in US dollars per tonne. This benchmark is used by the major airlines in Europe to settle jet fuel consumption contracts. ARA refers to the ports of Amsterdam, Rotterdam, and Antwerp in the Northwest Europe refining hub market. The price of products that pass through these ports are used as a benchmark because they have deep, active trading markets and are close enough geographically to be considered as interchangeable. The Northwest Europe markets are active in both ocean-going refined product as well as river barges. A barge is a non-ocean-going tanker that is used to transport jet kerosene, amongst other refined oil products, in coastal waters and inland waterways. Barges are smaller than ocean tankers and have flat bottoms that allow them to navigate through relatively shallow

Chapter 12

326

water. Areas with active barge markets such as Northwest Europe will have spot market transactions quoted for barge refined products, which are usually priced higher because barges are smaller than tankers. The third popular jet kerosene price benchmark is the U.S. Gulf Coast FOB price, which is used to set prices in the Americas. This jet kerosene product is usually denominated in US dollars per gallon. The U.S. Gulf Coast is one of the major refining supply centres and spot market trading hubs in the world. The U.S. Gulf Coast typically includes the refineries along the U.S. coast of the Gulf of Mexico running from southern Texas to southern Alabama. The Gulf Coast spot trading market generally includes trading of jet kerosine products in the ports of Houston, Corpus Christi, and Beaumont/Port Arthur. This includes transactions associated with tanker and barge jet kerosine in the case of water transportation, and transactions at the injection point of key product pipelines such as Colonial, Plantation and Explorer Pipelines in the case of land transportation. The Colonial Pipeline is one of the major refined product pipelines that moves jet fuel from the U.S. Gulf Coast to the U.S. East Coast market. The Plantation Pipeline is one of the major refined product pipelines that moves jet fuel from the U.S. Gulf Coast to the Southeast U.S. market. The Plantation Pipeline links the refining capacity in Louisiana to Southeast U.S. markets, including Georgia, North Carolina, South Carolina, and Virginia. The Explorer Pipeline is one of the major refined product pipelines from the U.S. Gulf Coast to the U.S. Midwest market. The pipeline at its Houston origin is a major point for spot trading of jet fuel, enabling the determination of the Gulf Coast Pipe price assessment.

12.5.3 Historical Movements in the Jet Fuel Price Figure 12-19 below presents the movement over time in the price of the Gulf Coast jet kerosene benchmark over the calendar annual period 1990 through to 2020 (the last point of which is the average for calendar 2020 over the period 1 January 2020 through to 23 November 2020). U.S. Gulf Coast Jet Fuel Prices (US$/bbl) $140 $120 $100 $80 $60 $40 $32 $20 $0

$128

$36

$45

2020 2019 2018 2017 2016 2015 2014 2013 2012 2011 2010 2009 2008 2007 2006 2005 2004 2003 2002 2001 2000 1999 1998 1997 1996 1995 1994 1993 1992 1991 1990 Source: Energy Information Administration Spot Prices 2021

Fig. 12-19: U.S. Gulf Coast Jet Fuel Prices (US$ Per Barrel) We can see in Figure 12-19 that like the price of crude oil, the spot price of jet kerosene cycled around a long run average of US$24 per barrel for the decade period starting 1990, thereafter the price cycled around an upward trend between 2000 and 2012 after which it has cycled around a downward trend. The biggest driver of the price of jet kerosene is the price of the biggest input into jet kerosene production, and thus the biggest source of jet fuel refining costs, which is the price of crude oil. As indicated in Figure 12-20 below, the movement in jet kerosene (U.S. Gulf Coast) and crude oil (West Texas Intermediate) prices is highly positive correlated, with the correlation between 1990 and 2020 estimated to be +95.5%. The difference between the spot price of jet kerosene and the spot price of crude oil is used to pay for the non-crude oil costs associated with refining crude oil into jet kerosene, such as the fixed costs of the refining infrastructure, the costs associated with the energy and materials required to refine the oil and the storage and transportation of the crude oil and refined products.

12.5.4 The Demand for Jet Fuel As indicated in Figure 12-21 below, the global consumption of jet kerosene was 6.917 billion barrels per day in 2017 up from 2.1b barrels per day in 1980. The growth in global jet fuel consumption over the four-decade period 1980 to 2017 is presented in Figure 12-22 below.

The Economics of Oil and Jet Fuel Markets

Gulf Coast Jet and WTI Oil Prices (US$/bbl) $140 $120 $100 $80 $60 $40 $20 $0

Gulf Coast Jet

327

WTI Crude Oil

$45.15 $38.32

2020 2019 2018 2017 2016 2015 2014 2013 2012 2011 2010 2009 2008 2007 2006 2005 2004 2003 2002 2001 2000 1999 1998 1997 1996 1995 1994 1993 1992 1991 1990 Source: Energy Information Administration Spot Prices 2021

Fig. 12-20: Gulf Coast Jet Fuel and West Texas Intermediate Crude Oil Prices World Barrels of Jet Fuel Consumption (m per day) 7,900

6,917

5,900 3,900 2,111

1,900

2017 2016 2015 2014 2013 2012 2011 2010 2009 2008 2007 2006 2005 2004 2003 2002 2001 2000 1999 1998 1997 1996 1995 1994 1993 1992 1991 1990 1989 1988 1987 1986 1985 1984 1983 1982 1981 1980 Source: BP Statistical Review of World Energy 2020

Fig. 12-21: World Jet Fuel Consumption

Growth in World Jet Fuel 30% Consumption 20% 10% 3.4%

0% -10%

-3.8%

-2.5%

-1.6%

-4.6%

2017 2016 2015 2014 2013 2012 2011 2010 2009 2008 2007 2006 2005 2004 2003 2002 2001 2000 1999 1998 1997 1996 1995 1994 1993 1992 1991 1990 1989 1988 1987 1986 1985 1984 1983 1982 1981 1980 Source: BP Statistical Review of World Energy 2020

Fig. 12-22: Growth in World Jet Fuel Consumption We can see in Figure 12-22 that jet fuel consumption averaged growth of 3.4% between 1980 and 2017, which is a similar rate of long run growth to global aviation capacity. Jet fuel consumption growth cycles around this long run average, with low points in growth clearly coinciding with the world’s major global economic downturns. The global economy experienced weak growth or recessions at the start of each decade for the past four decades – 1982 (global jet fuel consumption growth of -4.8%), 1991 (global jet fuel consumption growth -2.5%), 2001 (global jet fuel consumption growth -1.6%), and 2009 (global jet fuel consumption growth -4.6%). At a more micro-level, jet fuel consumption growth will also closely track airline capacity growth. In Figure 12-23 below, we present the growth in the number of available seat miles of one of the globes most successful low-cost carrier, Southwest Airlines, and the jet fuel consumption of Southwest Airlines.

Chapter 12

328

Jet Fuel and ASM Growth Southwest

Jet Fuel

ASM

25% 20% 15% 10% 5% 0%

-5% -10% 2019 2018 2017 2016 2015 2014 2013 2012 2011 2010 2009 2008 2007 2006 2005 2004 2003 2002 2001 2000 1999 1998 1997 1996 1995 1994 Source: Airline Intelligence and Research Database 2021

Fig. 12-23: Growth in Southwest Jet Fuel Consumption Versus Capacity Growth We can see from Figure 12-23 that the firm line representing the growth in Southwest jet fuel consumption is highly positively correlated with the dashed line, which is Southwest capacity growth between 1994 and 2019, with the extent of the correlation estimated to be +98.1% over this period. While there are other forces that will influence the growth in jet fuel consumption of an airline aside from capacity growth, such as the passenger and freight load on the aircraft (the load factor), the average age of its fleet (younger aircraft with more modern engine and fuselage technology consume less fuel) and the price of jet fuel, the dominant driver over time is airline capacity.

12.6 A Model of the Jet Fuel Market In this section I build a model of the market for jet fuel with a view to deriving a demand function for jet fuel. This demand function will describe the forces that should be considered in determining the extent to which the demand for jet fuel falls in response to an increase in the price of jet fuel.

12.6.1 Theory and Analytics The demand for jet kerosene is driven by an airline’s supply of air transport services. To model jet kerosene demand we therefore need to model the capacity decisions of airlines. We start our analysis of such a model by assuming that there is a group of n airlines which compete in Cournot competition on a particular route (hopefully you will recall that we analysed the n-player Cournot model in Chapter 9, section 9.6). The market inverse demand for air travel on the route in question is described by the following function: P = f(S)

(12.20)

where P is the market revenue per seat, f(S) is a smooth and continuous function which means that it can be differentiated, and S represents market capacity for the route in question. The market inverse demand function describes the extent to which yields change on a route when the market capacity on that route increases. We assume that an increase in S ௗ௉ coincides with a decrease in P, or < 0, which also means the inverse demand function (12.20) is downward sloping. ௗௌ Market capacity is equal to the sum of the capacity decisions of the individual airlines. We can write this in the following way: S = σ୬୧ୀଵ s୧ = s1 + s2 + ……. + sn

(12.21)

The notation σ୬୧ୀଵ s୧ is referred to as Sigma Notation. It simply means that we sum over all si terms from the starting number 1, which is at the bottom of the sigma summand, through to the end number n, which is located at the top of the summand. Each airline that operates on the route consumes jet kerosene according to the following linear production relationship: Fi = a0 + a1si i = 1,…… n

(12.22)

This production relationship simply says that the jet fuel consumption of the ith airline in barrels depends on the capacity of the airline, si and some other element that is independent of capacity, ao. The a0 will depend on a variety of factors

The Economics of Oil and Jet Fuel Markets

329

including the load on the aircraft flown by the airline, and the age of the aircraft used to operate services on the route, amongst other potential factors. The fuel cost function of each airline on the route is equal to the into-plane jet fuel price measured in US dollars per barrel multiplied by the quantity of fuel in barrels at (12.22), which gives: C୧୊୳ୣ୪ = P୎୉୘ F୧ = P୎୉୘ (a଴ + aଵ s୧ ) i = 1,…… n

(12.23)

where PJET is the into-plane price of jet fuel per barrel in US dollars (as defined in chapter 6). The non-fuel cost function of each airline is described by the following linear relationship: C୧ = c୧ s୧ + K ୧

i = 1,……, n

(12.24)

where ci is the non-fuel variable cost of the ith airline per seat and Ki is the fixed cost of the ith airline or those costs that do not vary with capacity. The profit function of the ith airline is the pricing function (12.20) multiplied by the capacity of the ith airline si, less fuel costs at (12.23) and less non-fuel costs at (12.24), which yields: Si = P൫σ୒ ୧ୀଵ s୧ ൯s୧ െ P୎୉୘ (a଴ + aଵ s୧ ) െ c୧ s୧ െ K ୧

i = 1,…… n

(12.25)

We assume that the market inverse demand function (12.20) is the following linear function: P = A0 + A1S = A0 + A1σ୬୧ୀଵ s୧

(12.26)

We also assume that the coefficient A1 is less than zero in (12.26) so that an increase in market capacity, S causes a decline in revenue per seat P. We choose a linear function for the market inverse demand function to simplify the problem, and so that we may obtain a solution for the demand for jet fuel that is in closed-form. This means that we can write the solution for the demand for jet fuel as a function of all other variables. Each airline attempts to choose the level of capacity si to maximise profit. To do this, they differentiate (12.25) with respect to si and set the result equal to zero, which yields: ୢ஠౟ ୢ୯౟

= A0 + A1S + siA1 – a1PJET – ci = 0

i = 1, …., n

(12.27)

If we sum (12.27) over all n airlines in the market, we obtain the following aggregate market marginal revenue equals marginal cost condition: nA0 + nA1S + A1S = na1PJet + σ୬୧ୀଵ c୧

(12.28)

If we factor S out of the front of the left-hand side of (12.26) and simplify we obtain: SA1[n + 1] = na1PJet + σ୬୧ୀଵ c୧ – nA0

(12.29)

If we divide both sides of (12.27) by A1[n +1] this yields the airline market supply of capacity in equilibrium: S* =

୬ୟభ ୔ెు౐ ା σ౤ ౟సభ ୡ౟ ି୬୅బ ୅భ [ଵା୬]

(12.30)

Equation (12.30) simplifies to: S* =

୬ൣୟభ ୔ెు౐ ାୡି୅బ ൧ ୅భ [ଵା୬]

(12.31)

where ܿ is the average non-fuel marginal costs of the airlines for the route in question. If we sum (12.22) over all airlines, we obtain the aggregate jet fuel consumption of the airline route: F = Na0 + a1S

(12.32)

Substituting (12.31) into the right-hand side of (12.32) yields the market demand function for jet fuel: ୬ൣୟభ ୔ెు౐ ାୡି୅బ ൧

F* = na0 + a1

୅భ [ଵା୬୅భ ]

(12.33)

Chapter 12

330

The market demand function for jet kerosene depends on seven parameters - the level of jet fuel consumption that does not depend on passenger capacity (a0), the number of airline competitors (n), the inverse of fuel marginal product (a1), the into-plane spot price of jet fuel (PJET), the average non-fuel capacity varying costs per unit of capacity (ܿ), the underlying demand for air travel, (A0), and the sensitivity of inverse demand to an increase in airline capacity (A1). We can also write (12.33) in the following form to emphasise the fact that it is a jet fuel demand function as follows: F* = G0 + G1PJET

(12.34)

where G0 { na0 – a1

௡஺బ

஺భ [ଵା௡஺భ ]

and G1 {

௡௔భమ

. While we have derived this jet fuel demand function at (12.34) with a

஺భ [ଵା௡஺భ ]

particular route in mind, we can think of this route as representative of the set of all airline routes. If we do this, we can interpret (12.34) as a global jet fuel demand function. The sensitivity of jet kerosene to a change in the spot price of jet fuel in equilibrium is found by differentiating (12.34) with respect to PJET, which gives: ୢ୊‫כ‬ ୢ୔ెు౐

=

୒௔భమ ୅భ (ଵା୒୅భ )

= Ɂଵ

(12.35)

This represents the slope of the jet fuel demand function. This slope is less than zero when A1, the slope of the inverse demand function for air travel, is less than zero, which implies the denominator of (12.35) is less than zero. The elasticity of optimal jet fuel consumption to a change in jet fuel prices is (12.35) multiplied by the jet fuel price and divided by (12.34) which gives: ୢ୊‫ె୔ כ‬ు౐ ୢ୔ెు౐ ୊‫כ‬

୒௔భమ ௉಻ಶ೅

=

ೌభ ొቂೌభ ು಻ಶ೅ శ೎షಲబ ቃ

୅భ (ଵା୒୅భ )൥ே௔బ ା

ಲభ [భశಿಲభ ]

(12.36) ൩

Under the assumption that there is no component of fuel consumption that is independent of capacity (a0 = 0) then (12.36) simplifies to: ୢ୊‫ె୔ כ‬ు౐ ୢ୔ెు౐ ୊‫כ‬

=

௔భ ௉಻ಶ೅

(12.37)

௔భ ௉಻ಶ೅ ା௖ି஺బ

Equation (12.37) says that the drivers of the elasticity of jet fuel consumption to the jet fuel price are the underlying demand for air travel and jet fuel costs as a percentage of total variable costs. We will provide an illustration of how to apply the formula at (12.37) in section 12.6.2 to follow.

12.6.2 Numerical Illustration Let us consider the domestic U.S. airline market. This market consists of the airlines presented in Table 12-3 below, which also contains an estimate of the jet fuel consumption of those airlines and their jet fuel productivity (available seat miles per gallon of fuel consumed) over the 12 months to September 2020. U.S. Airline (1) Alaska Airline Allegiant Airlines American Airlines Delta Air Lines JetBlue Airlines Skywest Airlines Southwest Airlines Spirit Airlines United Airlines Total

Jet Fuel Consumption (m gallons) (2) 561 159 1,951 1,700 529 294 1,512 328 1,541 8,575

Available Seat Miles (m) (3) 44,130 13,738 134,024 113,945 40,288 20,291 119,705 30,380 97,768 614,269

Fuel Productivity (3)/(2) = (4) 78.663 86.194 68.689 67.039 76.159 68.985 79.170 92.692 63.434 71.635

Source: Airline Intelligence and Research Database 2021

Table 12-3: Jet Fuel Consumption, Capacity and Jet Fuel Productivity in the US Domestic Market – 12 Months to September 2020

The Economics of Oil and Jet Fuel Markets

331

We can see in Table 12-3 that the jet fuel consumption of the domestic U.S. airlines totals 8.6 million gallons over the 12 months to September 2020, which is generated from 614 billion available seat miles of airline capacity. This implies fuel productivity for the U.S. domestic market of 71.635 available seat miles per gallon of jet fuel. We use this information to specify a relationship between aviation capacity and fuel consumption in gallons in the U.S. domestic market as follows: F=

ୗ ଻ଵ.଺ଷହ

= 0.01396S

(12.38)

where S is the number of annual available seat miles in millions and F is the number of annual gallons of jet fuel consumed by the U.S. domestic airlines in Table 12-3.58 This expression is the equivalent of (12.32) above under the assumption that a0 = 0. We assume that the inverse demand function at (12.26) for the U.S. domestic market is: P = A0 + A1S = 0.15 - 0.00000008S

(12.39)

where P is defined as the revenue per available seat mile for the combination of airlines presented in Table 12-3. We assume that the average non-fuel marginal cost of airlines in the U.S. domestic market in US dollars per available seat mile is ܿ = 0.07. Substituting these values into (12.33), including n = 9, and we obtain the estimated jet fuel demand function for the U.S. domestic market: F* =

୒ൣୟభ ୔ెు౐ ାୡି୅బ ൧ ୅భ [ଵା୒]

=

ଽൣ଴.଴ଵଷଽ଺୔ెు౐ ା଴.଴଻ି଴.ଵହ൧ ି଴.଴଴଴଴଴଴଴଼[ଵାଽ]

= 900,000 െ 157,050P୎୉୘

(12.40)

Using Microsoft Excel, the U.S. jet fuel demand curve is presented in Figure 12-24 below. At an into-plane jet fuel price of US$1.5 per gallon, the estimated jet fuel consumption in the U.S. domestic market per annum is: F* = 900,000 െ 157,050(1.5) = 664,425 million gallons U.S. Domestic Market Jet Fuel Consumption (m gallons) 1,000,000 900,000 800,000 700,000 600,000 500,000 400,000 300,000 200,000 100,000 0

664,425

$1.5 0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2 2.2 2.4 2.6 2.8 3 3.2 3.4 3.6 3.8 4 Price of Jet Fuel (US$ per gallon)

Fig. 12-24: U.S. Domestic Market Jet Fuel Demand Curve This point is also marked in Figure 12-24. We will now use our jet fuel demand function (12.40) to estimate the price elasticity of jet fuel demand at different jet fuel prices. The jet fuel demand elasticity is equal to the slope of the jet fuel demand function in (12.40) multiplied by PJET and divided by (12.40), which gives: U.S. Domestic Elasticity of Jet Fuel Demand to Price =

ିଵହ଻,଴ହ଴୔ెు౐ ଽ଴଴,଴଴଴ିଵହ଻,଴ହ଴୔ెు౐

(12.41)

Figure 12-25 below graphs the elasticity of U.S. jet fuel consumption to price at (12.41) against the price of jet fuel.

58 In this numerical illustration we use available seat miles rather than airline seat kilometres as the volume metric. This does not change the substance of the analysis. Using seats, available seat miles or available seat kilometres are interchangeable when deriving the jet kerosene demand function.

Chapter 12

332

U.S. Domestic Market Elasticity of Jet Fuel Consumption to Price 0.0 -0.5 -1.0 -1.5 -2.0 -2.5 0

0.2 0.4 0.6 0.8

1

1.2 1.4 1.6 1.8

2

2.2 2.4 2.6 2.8

3

3.2 3.4 3.6 3.8

4

Price of Jet Fuel (US$ per gallon) Fig. 12-25: U.S. Domestic Market Jet Fuel Elasticity of Demand We can see that jet fuel demand becomes more elastic to the jet fuel price as the jet fuel price is increased. At a price per gallon of US$1.5, the elasticity of jet fuel demand to price is equal to -0.35. This means that a 10% increase in the jet fuel price leads to a 3.5% reduction in jet fuel demand.

12.7 Jet Refining Margins The jet fuel refining margin is the difference between the spot price of jet fuel and the spot price of crude oil. We can describe this by the following simple expression: Jet Fuel Refining Margin = Spot Price of Jet Fuel – Spot Price of Crude Oil = PJET - POIL

(12.43)

Another name for the jet fuel refining margin is the jet crack spread, or simply the jet crack. It is referred to as the crack spread because the process of breaking crude oil down into fractions is also referred to as cracking. When we calculate the refining margin, we tend to use jet fuel and crude oil benchmarks located in the same geographic zones. For example, the following calculations are used as refining margins: Americas Refining Margin is the Spot Price of Gulf Coast Jet minus the Price of West Texas Intermediate, the Europe Refining Margin is the FOB Barge ARA Price minus the Brent Price, and the Asia Refining Margin is the Singapore Jet Kerosene Price minus the Malaysian Tapis Price. As indicated in Figure 12-20 above, over calendar year 2019 the jet crack spread between Gulf Coast jet fuel and West Texas Intermediate crude oil was: Americas Refining Margin = 45.15 – 38.32 = US$6.83 per barrel The relativity between the spot price of jet fuel and the spot price of crude oil can also be determined as a ratio, called the crack ratio. The crack ratio is the ratio of the spot price of jet fuel and the spot price of crude oil, which can be described in the following way: Jet Fuel Crack Ratio =

ୗ୮୭୲ ୔୰୧ୡୣ ୭୤ ୎ୣ୲ ୊୳ୣ୪ ୗ୮୭୲ ୔୰୧ୡୣ ୭୤ େ୰୳ୢୣ ୓୧୪

=

୔ెు౐ ୔ో౅ై

The reason the crack spread is often represented as a ratio is because the crack margin tends to share a positive relationship with the absolute level of the crude oil price. In other words, the crack margin goes up as the oil price goes up and goes down as the oil price goes down. This can be seen in Figure 12.26 below. In Figure 12-26 we see that the firm crude oil price line and the dashed jet crack margin share a positive relationship, with the correlation between them 90.9% between 1990 and 2020. The rationale behind the positive relationship between the jet crack margin and the price of oil is that part of the refining costs of oil refineries is attributable to the storage of crude oil, which is classified by refineries as an inventory asset. As the price of oil changes the value of these inventory assets change, which must be passed through into the costs of operations of the business. This change in costs is then passed through into the prices of the refined products. Figure 12-27 below presents the crack ratio in the case of Gulf Coast jet fuel prices relative to West Texas Intermediate crude oil prices.

The Economics of Oil and Jet Fuel Markets Jet Crack Margin and WTI Oil Prices (US$/bbl)

333 Jet Crack Margin (US$/bbl)

WTI Crude Oil

Jet Crack Margin

$104

$36

$94

$31

$84

$26

$74 $64

$21

$54

$16

$44

$11

$34 $24

$6

$14

$1 2020

2018

2016

2014

2012

2010

2008

2006

2004

2002

2000

1998

1996

1994

1992

1990

Source: Energy Information Administration Spot Prices 2021

Fig. 12-26: West Texas Intermediate Crude Oil Prices and the Jet Crack Margin Gulf Coast Jet and Brent Oil Crack Ratio 1.45 1.40 1.35 1.30 1.25 1.21

1.20

1.18

1.15 1.10 1.05

2020 2019 2018 2017 2016 2015 2014 2013 2012 2011 2010 2009 2008 2007 2006 2005 2004 2003 2002 2001 2000 1999 1998 1997 1996 1995 1994 1993 1992 1991 1990 Source: Energy Information Administration 2021

Fig. 12-27: Gulf Coast Jet Fuel Price to West Texas Intermediate Crude Oil Price Crack Ratio We can see in Figure 12-27 that the long run average crack ratio is equal to 1.21, or the jet fuel spot price is 21% above the West Texas Intermediate price, with the crack ratio cycling around this long run average between 1990 and 2020. The extent to which the crack ratio cycles above and below the long run average of 1.21 will depend on decisions that oil refineries make about the price at which they sell their refined product price to retailers. The next section will provide some insights into the economic drivers that may cause the crack ratio to cycle above and below long run average levels.

12.8 An Economic Model of the Jet Fuel Crack Margin 12.8.1 Theory and Analytics In this section I will present an economic model that I have built to describe the decisions that oil refineries must make about how much crude oil is required to produce refined products, the impact that those decisions have on the amount of refined product that is produced, and the prices of those refined products relative to the price of crude oil.

Chapter 12

334

An oil refinery consumes Q barrels of crude oil and converts this into two products – Jet Kerosene (J) and a composite product that consists of all other refined products (O), which will include such products as gasoline, diesel, kerosine and heating oil. The price that the refinery pays for the crude oil is POil plus transport costs cT which is defined on a per barrel basis. The price that the refinery receives for its jet kerosene product is PJ per barrel and the price it receives for its other refined products is PO per barrel. We assume that the barrels of jet kerosene that are derived from refining barrels of crude oil is defined by the following relationship: QJ = GJQ

(12.44)

where QJ are the number of barrels of jet kerosene. The barrels of other refined product, such as gasoline and diesel, derived from refining barrels of crude oil are defined by the following relationship: Q O = G OQ

(12.45)

where QO are the number of barrels of other refined product. We can think of these relationships as production relationships with fixed production parameters GJ and GO. The refinery can influence the price it receives for jet kerosene and other refined products by altering the volume of jet kerosene and other refined products that it produces. This is described by the following linear inverse demand functions for jet kerosene and the other refined products: PJ = a0 + a1QJ = a0 + a1GJQ

(12.46)

PO = b0 + b1QO = b0 + b1GOQ

(12.47)

We assume that the oil refinery does not face any competition for simplicity. We also assume that the oil refinery fixed costs are described by K. We assume that there are additional variable costs associated with refining crude oil into jet fuel and other refined products, including cO per barrel for other refined products and cJ per barrel in the case of jet fuel. The profit function of the oil refinery is equal to (12.46) multiplied by the number of barrels of jet fuel sold plus (12.47) multiplied by the number of barrels of other refined product sold less refinery costs as follows: S = (a0 + a1GJQ – cJ)GJQ + (b0 + b1GOQ – cO)GOQ – (POil + cT)Q – K

(12.48)

The oil refinery decides to purchase Q barrels of crude oil to maximise profit. The first order condition for maximum profit is found by differentiating the profit function with respect to Q and setting the result equal to zero as follows: ୢ஠ ୢ୕

= ൫a଴ + aଵ Ɂ୎ Q െ c୎ ൯Ɂ୎ + Ɂ୎ Qaଵ Ɂ୎ + (b଴ + bଵ Ɂ୓ Q െ c୓ )Ɂ୓ + Ɂ୓ Qbଵ Ɂ୓ െ (P୓୧୪ + c୘ ) = 0 (12.49)

We can re-write first order condition (12.49) as a marginal revenue equals marginal cost condition: a଴ Ɂ୎ + b଴ Ɂ୓ + 2Q(aଵ Ɂଶ୎ + bଵ Ɂଶ୓ ) = Ɂ୎ c୎ + Ɂ୓ c୓ + (P୓୧୪ + c୘ )

(12.50)

The left-hand side of (12.50) represents the marginal revenue of the oil refinery. This describes the impact on refinery revenue of a small increase in crude oil consumption, which is refined into jet fuel and other refined products. The small increase in oil refining output has two impacts. It results in greater production of both jet fuel and other refined products, which for given jet fuel and other refined product prices results in greater revenue. The increase in refined product production, however, also results in lower jet fuel and other refined product prices, which reduces revenue for given volumes of jet fuel and other refined product. The net impact on revenue is dependent on the relative size of these positive volume and negative price impacts of greater oil refining. The right-hand side of the equilibrium condition (12.50) represents the marginal cost of the oil refinery. This has three component parts – jet fuel refining marginal costs, other product refining marginal costs and the cost of buying and transporting crude oil. The marginal costs of the oil refinery are always positive. As marginal costs are always positive then marginal revenue must be positive at the point where profit is maximised. For marginal revenue to be positive this requires the left-hand side of (12.50) to exceed zero, or: a଴ Ɂ୎ + b଴ Ɂ୓ + 2Q൫aଵ Ɂଶ୎ + bଵ Ɂଶ୓ ൯ > 0 For the condition at (12.51) to hold this requires the following restriction on output: Q