Financial Modeling For Equity Research 3e [3 ed.]

Citation preview

Financial Modeling For Equity Research A step-by-step guide to earnings modeling and stock valuation for investment analysis Third Edition

John Moschella, CFA, CPA

Gutenberg Publishing Company Copyright © 2019 Gutenberg Research LLC All rights reserved.

TEXTBOOK ORIENTATION Opening Thoughts From the Author

I believe that throughout our lives, who we choose to educate us, and what form that education takes, is among the most fundamentally important decision we make. Each day we learn more, and are directly influenced by the information we allow into our world: the news channels we watch, the communities we live in, the websites we visit. How you choose to educate yourself is a critical decision in all aspects of your life. Your education in accounting, finance, investments, and financial modeling is no different. To help you decide to what extent this book, and our website GutenbergResearch.com, will be included in your universe of educational material, I have summarizes my views on some key components within this orientation section. These thoughts shape the concepts which form the core approach taken in this book, as well as all of our other Gutenberg Research programs.

About the Author—John Moschella, CFA, CPA I have spent nearly 15 years analyzing companies in various capacities. After earning a BSBA in Finance, MS in Accounting and MBA from Northeastern University, I began my professional career at PricewaterhouseCoopers (PwC) in New York as an Assurance Associate in the Financial Services practice. During my time at PwC I participated in a rotational assignment within the Financial Service Research Institute at PwC where I studied bank mergers throughout the financial crisis, drawing on my experience as an intern bank examiner with the Federal Deposit Insurance Corporation (FDIC) during my undergraduate career. After PwC, I spent five years at UBS Investment Bank where I worked first as a Capital Specialist, and then as an Equity Research Associate. In my research role I built and maintained earnings models, contributed to research reports, and participated in client conferences, covering the Semiconductor and Semiconductor Capital Equipment Industries. I then moved to General Electric Capital Corp in 2014 as a Risk Analyst where I built regression models to predict asset losses based on various macroeconomic scenarios. After the sale of the majority of GE Capital’s assets, I started a consulting firm which provides capital planning support to banks, in addition to running Gutenberg Research, a crowdsourced earnings modeling community.

Thoughts on Modeling

At its core, financial modeling is about educating ourselves on the potential prospects of a particular company. This education is derived through the consideration of past events, as well as potential upside and downside possibilities, which could change the company’s financial path. Our modeling efforts represent an attempt to organize the chaos of variables and forces which drive our theories of the future. At the center of modeling and forecasting, is the concept that as research market participants, we seek to have a reasonable basis for our ideas, whether we use them to make investment decisions, publish articles, communicate in an interview, or tweet a comment. I am confident that anyone can create a model. In fact, we all model everyday. Whether we formalize our efforts in a spreadsheet is a separate point. When someone asks your opinion about a product, another person, or a company, in your mind you will consider the different aspects which go into forming your view, before giving your response. This is what modeling means to me. The spreadsheet simply acts as a vessel for the various factors we have used to form our opinions. My modeling approach does not encompass advanced spreadsheet features or automation. Nor do I employ complex quantitative theories. I believe these tools drive people away from modeling, which contradicts the natural progression

of knowledge. We all benefit from inclusion. Mankind’s greatest achievements have come from many people collectively looking at the same problem (in this case the future prospects for a company). The collective thought and effective challenge is what enables us to reach a more accurate answer as a community of researchers. I tend to favor simplicity over complexity. Sometimes we fall trap to the fallacy that complexity equals certainty. It does not. After reading through all the steps in my modeling approach, you may begin to think that I have violated this basic concept; however, the detail used in my models is in fact relatively simple compared to the models of sell-side analysts at investment banks. After our model has been created, it will empower us to cut through the noise, and draw very simple conclusions about whether we expect a company’s earnings and cash flows to grow or shrink in the future, and how that will compare to competitors or the broader market. These are the simple answers we seek. How we decide to achieve them is up to us individually.

Thoughts on Valuation

Your belief on valuation may differ from mine. I believe that most developed equity markets are highly efficient, meaning all current information is incorporated into equity prices during market hours, at times of sufficient liquidity 1. As a result, I believe that stocks are never overvalued or undervalued, in the true definition of the terms. They are always fairly valued based on the information available at the time. The evidence of this point can be demonstrated by making a transaction. The transaction price proves that the true value of any asset (a security, commodity, real estate, art, anything) is what someone is willing to pay for it now. If you believe, as I do, that the market is efficient, why would you waste time modeling earnings for a company? The answer is that the market incorporates all information into its price discovery efforts: macroeconomic, political, competitive, and of course company specific upside and downside forecasts. The latter point is where I see the added value proposition. I believe that the market’s development of a future forecast includes a broad range of potential upside and downside cases. We can approximate the market’s view using the consensus analyst estimates, although this represents a very narrow sample of market participants, typically with a broad range of outcomes. You could talk to 20 different analysts about the same stock, with 20 different earnings estimates, 3 different recommendations 15 different target prices, and 10 different approaches to how they reached their conclusions. By entering your opinion about the earnings capacity of a company into an earnings model, you are forming your own view, which may lie below or above the average. Naturally, this will produce a future valuation which is different from the current fair value. This is the reason we forecast, besides the pure joy of investigation and analysis which may be enough for you (as it is for me). Valuation has a short shelf life. It is a very delicate concept which can disappear in the wind, the minute a geopolitical risk, economic downturn, emerging technology, or any other number of developments capture the market’s attention. My view is we should not fight the market by implying things are over or undervalued. Instead we should continually challenge our own views and biases, stay true to our analysis, but incorporate new information as it is released. This is why I emphasize the modeling of earnings and cash flow, and deemphasize exact valuations. As you read through the chapters within this book, I encourage you to challenge my approach in your mind, seek out other methodologies from other analysts, writers, training programs, and develop your own ideas of how efficient the market is, and how you should think about valuation.

The nature of our Gutenberg community members is to debate all points. In the back of my mind I can hear my fellow analysts saying “what about market manias, or steep recessions when logic falls to the wayside? Are valuations fair during these times.” This is a topic for discussion, and is the reason I qualify my opinion with the point “at times of sufficient liquidity.” I am not implying that market participants always act rationally. The idea is not to get sidetracked with semantics, but to understand the point of view, and why the valuation approach is logical. If your belief differs from mine, incorporate adjustments into your model to suit your needs.

1

iv

Textbook Orientation

Thoughts on Equity Research

Over time the Equity Research Industry has adapted to changing circumstances. In the early 2000s information equality took a leap forward with the passing of Regulation Fair Disclosure (Reg FD), which prohibited the dissemination of material information to select analysts or investors. This changed the nature of the relationship between analysts and the companies they cover. At the same time technological advances continued to progress, making it easier for companies to communicate directly to investors, analysts, and news outlets. In general, the availability of data has also come a long way in the last 10-20 years, which has led to the advent of contributor-based stock analysis websites, investment blogs, and even crowdsourced earnings estimate services. The latest development in sell-side research at investment banks has come in the form of a required change in pricing. Banks typically package their research products with equity trading services. Now regulators in certain markets are pushing to break the services apart, in an effort to improve cost transparency. With the pricing change comes new pressure to cut research costs at banks. As market forces act on the industry, the quality of research, which has already declined relative to what is currently available free of cost, will continue its descent. Equity research reports are beginning to blend together, and it is getting more and more difficult to recognize the value proposition of multiple identical reports that track the earnings preview/review cycle. Analysts are to some extent incentivized to follow the pack in order to protect a stable, yet unremarkable, career. The pricing change will amplify this effect as many cling to a bygone era of growth and prosperity in research. These forces have led the industry to become reactive instead of proactive. There is nothing more frustrating then watching all analysts cut their earnings and price targets after company management has decreased their guidance. What is the value in this research, when clients had no chance to execute on the recommendation prior to the announcement? The improvement in information flow, rise of the independent blog-style stock analyst, and changing price environment for sell-side research, has brought the Equity Research Industry to a pivotal point in its history. I believe that the market for equity research will shift to favor those who are best suited to provide it. In many cases this will remain top sellside analysts whose research is valued at or above the equilibrium point of its cost. However, sell-side research must shift to providing only primary high-value research, and leave the lower-end earnings preview/review style reporting to firms that can provide it at the lowest cost possible. I believe these developments will drive the age of the truly independent analyst. While individuals have already proved successful in gaining recognition through investment articles and blogs, the next logical step is to move up the traditional equity research value chain, and add the sell-side’s most valuable tool to the mix: earnings models.

About Gutenberg Research

Gutenberg Research is a crowdsourced earnings modeling community founded by two brothers, John Moschella and Brandon Cannon, to facilitate the next leg of equity research transformation. Our management guidance and consensus-based models will provide the basis for discussion, while our contributors will provide the voices, and the best analysts will rise to the top. Gutenberg is creating the platform for analysts, regardless of where they come from, to be recognized in all aspects of their work: Written reports/articles, backed by detailed analysis within their models, and explained in commentary. • • • •

Our Purpose: is to drive the evolution of equity research through financial modeling. Our Goal: is to make earnings models for all publicly traded companies available to all research participants. Our Vision: is an ultra-efficient research market where all participants are able to contribute their opinions in a challengeable environment backed by qualitative and quantitative support. Our Mission: is to grow the Gutenberg community, seeking out like minded analysts who share our vision for the future of research, and educate and empower those who wish to join our efforts.

The Gutenberg name and philosophy are inspired by the fifteenth century visionary and inventor of the printing press, Johannes Gutenberg. Gutenberg's press forever altered the state of communication and flow of information through the mass production of books, changing literacy from a luxury of an elite few, to a right of all. Now, taking a page from Johannes Gutenberg’s book, we are making earnings models available to the masses, rather than tools available only to the highest paying clients. v

We believe that our community’s collective knowledge will provide the best forecasting insight; however, the complexity of hundreds of forecasting possibilities, coupled with thousands of different investing theories must be tamed to facilitate the discussion among analysts. To maintain order and cut through the convoluted web of potential forecasting approaches, we provide spreadsheet templates for analysis: our management guidance and consensusbased earnings models which represent a “base-case” scenario. Our community members can then download these base-case scenario spreadsheets, and input their own assumptions to add their opinion related to a particular stock’s earnings and valuation prospects. Our ultimate goal is to have an inventory of models for all publicly traded companies with many variations of bull- and bear-cases for each. At the heart of our goal, is the commitment to financial modeling education. Our Certificate in Financial Modeling Program, and the self-study guide of our textbook Financial Modeling for Equity Research drives the execution of this educational effort. We are growing our network of contributors, first seeking out those with prior modeling experience to assist in the build of our consensus-based model inventory, and then expanding the network to capture the opinion of a wide group of market participants. We have a long road ahead of us, but we are off to a great start…stay tuned! Gutenberg Research Structure: Our parent company, Gutenberg Research LLC, operates under three segments: •

GutenbergResearch.com: Internet-based community of financial modeling contributors. Members can access our consensus-based models, and share their model forecasts on our site. The Gutenberg Financial Modeling Virtual Intern Program also runs on GutenbergResearch.com, providing a modeling introduction to students interested in fundamental analysis, and creating a platform for model developers to showcase their skills to prospective employers.

•

The Equity Research Institute at Gutenberg Research: Provides live classroom and virtual modeling training including the Certificate in Financial Modeling Program.

•

Gutenberg Publishing Company: Publishes and distributes course content and the Financial Modeling for Equity Research textbook.

About This Textbook

Subjects Covered: This book begins with an introduction to modeling in Chapter 1, covering some of the basic concepts which lay the foundation of the modeling process used throughout the curriculum. The core financial statement model building steps run from Chapter 2 through Chapter 4, which will cover the Income Statement, Balance Sheet, Cash Flow Statement, and forecasting. Chapter 5 covers the inputs for the Discounted Cash Flow (DCF) valuation, including the calculation of the market Equity Risk Premium (ERP) using interest rates, volatility and equity market return data, deriving beta, and calculating the required return on equity. Chapters 6 and 7 cover share valuation. After Chapter 7 you will have completed the initial model build, including the forecasted share valuation. Chapter 8 then demonstrates how to use and maintain your model. This includes a simulated earnings release. Appendix 1 shows how to use regression analysis to project inputs for your model. The remaining Appendix entries include the latest updates to the subject company model, as well as the external data used as inputs: volatility, interest rates, equity market returns, beta, the Equity Risk Premium (ERP), average returns and the standard deviation in returns. As new data is released, the curriculum will continue to be updated with added appendix entries. These entries will be published in later editions, so our readers may continue the learning process as economic conditions, and the competitive landscape changes. Downloading the Spreadsheets: Refer to the "How to Use This Textbook" section of Chapter 1 for instructions on how to access the spreadsheet files. The following Key Concepts will be covered in the curriculum as recurring themes: Key Concept 1—Modeling is a Formalization of Our Opinions This idea demystifies the term “model”, which some may interpret as a black box without transparency over the methodologies employed. Instead a model is simply the mechanism we use to list out our forecast assumptions, for the sake of bringing order to our projection. vi

Textbook Orientation Key Concept 2—The Fundamental Principles of Modeling: Balance and Drive These two points help frame how you should design your model to fit your specific needs. Balance, relates to the level of complexity in your model. You should “balance” the resulting analytical benefit you accrue from each additional layer of complexity, against the added effort to create it. Drive, deals with the selection of metrics used in the model, which should focus on items that are critical to the related accounts in the financial statements (or “drive” the value of the financial statement line item). Key Concept 3—An Approach to Model Any Account, for Any Company There are many different approaches to creating a financial model. The Gutenberg Modeling Framework provides a basic blueprint of tasks to perform during the modeling process, with specific questions to consider as you build your forecast. Key Concept 4—Valuation is in the Eye of the Beholder There are many different opinions of market efficiency and valuation theory. Two approaches are covered in the curriculum, with many opportunities for customization based on your particular views on valuation. Key Concept 5—Models are Living Tools After a model has been built, it should not sit idle. It should be used to perform analysis, debate potential company outcomes, and explore market possibilities. Most importantly, it must be maintained as new information is released. This point is critical. A model that is completed today, will be out of date by tomorrow. Just as weather changes from day-to-day, and is difficult to predict next week, so are earnings and share valuation. Remember to keep your model up-to-date. Key Concept 6—Be a Proactive Analyst A proactive analyst takes risks and adjusts his or her forecast based on the information available. A reactive analyst plays it safe, waits for the subject company to issue results, and potentially revise guidance before going out on a limb with their forecast. Key Concept 7—Own Your Projections One of the most important things you can do as an analyst, is to hold yourself accountable to your projections. There are many behavioral finance concepts working against analysts. We tend to be overconfident in our ability to predict future results, and often do not fully incorporate new data into our existing projections. We also tend to focus on areas where we are correct in our predictions, and tend to forget the areas where we are wrong. Try to keep your psychological biases in mind as you compare your projections to the reported results of the company you are covering. Celebrate the accurate points in your projections, but recognize when you are wrong. If necessary correct your mistake and move on with your new forecast.

Other Gutenberg Programs—The Virtual Financial Modeling Internship

Our Financial Modeling Intern Program is a part-time, virtual internship, which simulates the experience one would receive working as a sell-side Research Associate at an investment bank. We teach our interns how to use models to estimate a company's future earnings, and publish their work to showcase their modeling skills to potential employers. Our intern program has adapted into its current form over the last few earnings seasons. In the initial rounds, our interns created full earnings models from scratch, including the Balance Sheet, Cash Flow Statement, and Income Statement, and calculated estimated share valuation targets based on their earnings model. This proved to be an overwhelming experience for most, as just 10% of the interns were able to complete the exercise. Under the new program format, interns instead enter their own forecast into an existing model before the subject company reports earnings. After the company reports they can review the results to see how close their forecast was. For additional details on our intern program visit: https://www.gutenbergresearch.com/gutenberg-virtual-intern.html

vii

Other Gutenberg Programs—Certificate in Financial Modeling

Our Certificate in Financial Modeling Program uses the Financial Modeling for Equity Research textbook as the core class curriculum. The program is designed for our readers who wish to take their modeling development to a more advanced level with video tutorials, model review from our professors, demonstration of knowledge through examination, and the ability to add modeling to their resumes. In this program our students watch as we build a financial model from the ground up, starting with a blank spreadsheet, and ending with a comprehensive set of interrelated financial statements and share valuation estimates. The program is designed to mirror methods which many sell-side research analysts use in professional practice. Similar to the virtual internship, this program mimics the “on the job” experience one would receive as a new Equity Research Associate at an investment bank. Unlike the intern program, it includes the full-scope of financial modeling from beginning to end. Structure of the Certificate Program: There are five primary components to the program, all designed to add various elements to the training: 1) the written curriculum, 2) the demonstration videos, 3) the supporting spreadsheets, 4) guidance from Gutenberg professors, and 5) the examination. These resources are available through the Certificate Program Education Hub. Program Format: The program is offered in two formats, an online virtual classroom, or live in-class program. Both formats cover the same curriculum. The live course covers the topics as a fast-paced bootcamp over the course of two days. All live course participants will also have access to the full-scope online version of the curriculum, including the online video sessions. Periodic live web conferences are held online where students are able to ask questions, or listen to their collogues’ questions. All students are able to access the web conference sessions (live and on-line). Students may also submit questions through the learning hub, where answer will be posted in short videos. Earning the Certificate: At the completion of the training program students will receive a virtual Certificate in Financial Modeling from The Equity Research Institute at Gutenberg Research, to add to the education section of their resumes. In order to qualify for the certificate, students must pass the financial modeling exam which covers the topics in the program curriculum, and submit an earnings model of their own. Endorsements and Recommendations: For any certificate students interested, the program professor will endorse you for financial modeling and give you a recommendation on LinkedIn, after you complete the program. About the Exam (Required for the Certificate Program): The exam consists of 50 questions, each with an equal weighting. Students must answer 70% or more of the questions correctly to complete the certificate program. The exam includes six cases which highlight various components of the modeling program, followed by a section of multiple choice questions which focus on financial statement relationships. The exam questions are adapted from the Chapter Wrap-Up questions within the textbook. There is a four hour time limit for the exam. All questions are multiple choice. The topics for each of the exam cases are as follows: • Case 1—Primary Topic Covered: Basic model building concepts. Number of Questions: 8 • Case 2—Primary Topic Covered: Model forecasting techniques. Number of Questions: 5 • Case 3—Primary Topic Covered: Equity Risk Premium (ERP) and the required return on equity. Number of Questions: 6 • Case 4—Primary Topic Covered: Cash flow and share valuation. Number of Questions: 14 • Case 5—Primary Topic Covered: Forecasting, scenario and sensitivity analysis. Number of Questions: 4 • Case 6—Primary Topic Covered: Regression analysis. Number of Questions: 5 • Multiple Choice Section—Primary Topic Covered: Financial statement relationships. Number of Questions: 8

viii

Textbook Orientation About the Model Submission (Required for the Certificate Program): In addition to completing the exam, all students must also submit an earnings model to complete the certificate program. Students may select any public company listed on a U.S. exchange, or may submit a version of the FedEx Corp model (subject company of the textbook), which must incorporate a unique forecast (not calibrated to the consensus estimate). Models are graded on a “Pass/Fail” basis, and must achieve 70% of the possible points to receive a passing grade. Model scoring is governed by the following matrix: 1) The structure of the model can differ from the Gutenberg templates (i.e. different worksheets for the various financials); however, the model layout must be logical: 20 points. 2) The model drivers must be intuitive. This point covers the factors used to disaggregate earnings in the model (described in the textbook as “the Earnings Engine”): 20 points. 3) The primary model inputs must be logical based on the student’s projections about the company: 20 points. This does not mean the model reviewer has to agree with the forecast, but that the forecast makes sense based on the developer’s theories about the company, economic conditions, competition, etc. 4) The three primary financial statements must be forecasted including the Income Statement, Balance Sheet, and Cash Flow Statement: 5 points. 5) The model must forecast at least four quarters of projections: 5 points. 6) Cash on the Balance Sheet must equal the ending cash balance from the Cash Flow Statement: 5 points. 7) 8) 9) 10)

The Balance Sheet must balance (assets must equal liabilities plus equity): 5 points. Links between the financial statements must be logical: 5 points. Historic results must tie to SEC filings or company reports: 5 points. Segment results must reconcile to the Income Statement using the same method in the forecast columns as the historic columns: 5 points. 11) At least one share valuation technique is required: 5 points. 12) The model may not include macros. 13) Regression analysis is not required. 14) A target price band is not required. How to Represent the Certificate on a Resume/LinkedIn: Students who have received the certificate should indicate the achievement under the “Education” section of their resumes and LinkedIn profiles. If the subject model is submitted as a contribution to be published on GutenbergResearch.com, students may include this under a “Published Works” section. For additional details on the certificate program visit: https://www.gutenbergresearch.com/certificateprogram.html

Terms of Use

This book, Gutenberg Research Programs, and all associated material, models, files, and content published in text or on GutenbergResearch.com is for demonstration purposes only, and is presented “as is”. Neither Gutenberg Research, the author, nor any Gutenberg Research agents or associates are liable for any errors, delays, incompleteness of data presented, or for actions taken based on reliance on any information contained in this program, associated files and templates, or information presented on GutenbergResearch.com. The information presented is for demonstration and educational purposes only, and does not represent investment advice. Investors should consult a professional investment adviser prior to making investment decisions. Please refer to the full Terms of Use: https://www.gutenbergresearch.com/terms-of-use.html

ix

CHAPTER 1: INTRODUCTION TO FINANCIAL MODELING • What is an Earnings Model

• Timing of Publication

• Fundamental Principles

• Basic Excel Functionality

• How to Use this Textbook

• Types of Models

• The Earnings Cycle

• Anatomy of a Model

Chapter 1 Overview: An earnings model is a representation of a company’s financial position, disaggregated to a level which can be analyzed based on historic results and future expectations. The level of disaggregation depends on the subject company. For example, if someone asks you whether or not FedEx Corp, the global shipping company, will beat the consensus Earnings Per Share (EPS) estimate next quarter, it would be very difficult to answer without thinking about how many packages the company will ship, and the average revenue earned per package. In addition, you would probably think about how competitive pressure, or company initiatives will impact future profitability. An earnings model shows all of these components and how they work together to project a bottom-line EPS estimate. In this chapter we will discuss these fundamental concepts to help you prepare to build your first earnings model.

Key Concept 1—What is an Earnings Model: The Formalization of Opinions

The term model is broadly used to explain many different types of financial representations, which can vary depending on the type of task the model is performing. For example, an economist may use a regression model to predict the likelihood of an economic downturn based on several macroeconomic inputs. On the other hand, an option trader’s model may use various inputs to price a particular option contract. When we use the term model, or earnings model, in the context of equity research, we are simply referring to a spreadsheet which uses a set of assumptions to project the financial statements, for a particular company. Most people use the term “earnings” model since the Income Statement is typically the main focus of the spreadsheet. The primary objective of creating a model is to have a dynamic tool which can be adjusted to determine the impact of a change in assumptions. It is critical to understand this concept before proceeding to the next chapter. To demonstrate, consider the excerpt of the FedEx Corp earnings model in Exhibit 1 below. For now, ignore the fact that the fiscal first quarter of 2019 (F1Q19) has passed, and assume that this represents the forecast for the next quarter (refer to the “Timing of Publication” section for details on the reporting dates used in this textbook). Notice there are two columns for the forecast period: 1) Column “S(old)” is the forecast before making any changes to the model, and 2) Column “S(new)” is the forecast after a change has been made to the inputs. For now, focus on cell S13(old) which represents the total revenue forecast for the next quarter. This cell is not a hardcoded number, but an equation which equals the sum of the individual segment revenue estimates in rows 77 through 171. For simplicity, the components which make up the individual segment revenue forecasts have been hidden, except for the Express Segment U.S. Overnight Box business. The revenue forecast for the U.S. Overnight Box business is based 10

Chapter 1: Introduction to Financial Modeling on an equation which takes the average number of packages shipped in a day, multiplied by the average yield (revenue per package), multiplied by the number of days in the period. Column Q shows the actual result from the last reported quarter financial statements (refer to the “Timing of Publication” section for details on the reporting dates used in this textbook). In column S(old) we use the same calculation to project the next quarter, by inputting assumptions for the average package volume in cell S(old)74, and revenue per package S(old)76. Notice the package volume growth rate forecast has been set at -1.0%, and the package yield growth has been set at 4.0%. Now let us assume you disagree with the estimates in cells S(old)74 and S(old)76, and instead believe that the volume and yield will be higher. You may decide to increase the volume growth rate in cell S(new)74 to 5%, and average yield growth in cell S(new)76 to 6%. After you make these changes, you can go back to the total revenue in cell S(new)13 and see that it has changed from $16.9B to $17.0B based on the new assumptions in the blue cells. The rest of the Income Statement accounts have changed as well, since the cells are linked throughout the model’s equations, although at this stage you will not be able to see this effect yet. This example summarizes the primary purpose of earnings modeling. The goal is to break a company’s complex operations into smaller parts, and use the model as a tool to project future results. If we step back and look at what Exhibit 1 shows at a higher level. The blue cells represent our opinion of what the future package volume and revenue per package will be. Therefore the model has formalized our opinions, which creates transparency for our overall earnings and valuation forecast. This is the crux of Key Concept 1. A model is not a “black box” of uncertainty. It is a simple tool which organizes a set of assumptions.

Exhibit 1—FedEx Corp Package Revenue Model Example

My Experience: At investment banks, equity sales representatives sell the research of equity research analysts to institutional clients (usually packaged with other equity service offerings). This is where the phrase “sell-side” research comes from since investment banks sell to “buy-side” clients. The bank’s clients typically have different tiers of service based on the fees they pay. The highest-tier clients typically receive the analyst’s “working models”, which means the equations remain in the Excel file, so the client has the ability to change the assumptions and see the impact on earnings. Lower tier clients, such as in-house wealth management advisors who generate very little revenue for the research department, receive models which show the “values only” versions of the spreadsheet without the embedded equations. Hard-coded models do not allow for dynamic analysis. In this program, we will be demonstrating how to create working models.

11

FAQ 1—What is the difference between the modeling approach in this textbook and models built by sell-side analysts? The main difference is the level of granularity. Wall Street analysts tend to disaggregate earnings to a level which cannot be validated by company reports alone. The structure of the models in this program is also setup differently. Most Analysts use separate worksheets for each financial statement. In the model we build, we will include all three financial statements in the same worksheet: the Income Statement, Balance Sheet, and Cash Flow Statement. This will drive home the point that the financial statements are all interrelated into one connected model. I want my students to keep this in mind, and be able to simply scroll up or down to instantly see the impact throughout the various financial statements. When you are comfortable with this concept, feel free to split the financial statements into separate worksheets if you wish. Another difference is that we will be linking our valuation forecast to the model. Analysts do not do this. They give a recommendation (buy/sell/hold) and a target share price. While analysts may give potential upside and downside scenarios, they would not want to imply that changing certain attributes would alter their price forecast, so they typically do not link the valuation to their models.

Key Concept 2—The Fundamental Principles of Modeling: Balance and Drive

There are an infinite number of inputs and assumptions which could be used to build an earnings model. To help you frame your thought process as you decide on the variables you would like to include in your model, consider these two guiding principles: the Principle of Balance and the Principle of Drive. The Principle of Balance: This concept relates to the level of granularity you choose to incorporate in your model. The degree of effort required to create and maintain the model, should be balanced against the resulting analytical benefit from adding each additional layer of complexity. To demonstrate this principle, consider how you might project future revenue for a company which sells widgets. You may decide to take a simplistic approach by applying a growth rate to the last reported revenue directly, or you may add an additional layer of complexity by disaggregating revenue by the number of widgets sold and the average price of each widget. If the analytical benefit you derive from the additional widget unit detail in the second approach, outweighs the additional modeling effort, then you should choose the second method. If not, you should keep the model simple by choosing the revenue growth rate method. Maintaining the balance between effort and analytical value will help you to create a model which best fits your needs. The correct balance point is up to you. We all have different reasons for creating models. You may be building a model as a hobby because you are interested in following a particular company. In which case, your balance point would likely fall on the lower complexity side of the spectrum. On the other hand, if you are a sell-side analyst, your model would be more complex to meet the needs of your clients. In this book we will discuss different types of models, each with a different level of complexity; however, if you have purchased this book, I am assuming you have committed to learning the modeling approach which projects all three financial statements and uses multiple valuation methods. As you progress through the chapters consider where your balance point lies. If you find the material to be too complex, you may want to make some basic assumptions to simplify your version of the model. The Principle of Drive: For each financial statement account within our model, we must decide which metric should drive our future period projections. The metrics can be internal to a particular company, external forces, or a combination of the two. The key concept is that we choose metrics which drive the future results. If we go back to our widget company example, it is safe to say that the number of widgets sold will drive future revenue. This would be an example of a metric which is internal to the company. If widgets are a cyclical product, then you could also make an argument that changes in Gross Domestic Product (GDP) could also drive future sales. This is an example of an external metric which could be incorporated into the model (likely an input to determining the number of units, which would then translate into a revenue estimate). You would not, however, use average rainfall to project the revenue forecast since rain does not drive the widget company’s results. It can be difficult to determine which metrics to use when you are creating a model from scratch. As you research your company you should continually be asking what is driving the overall earnings of the company, and what is driving each individual account on the financial statements. Listen for clues from management as they discuss the company’s operations on earnings calls and investor conferences. Draw conclusions from the details provided in their financial statements, or other sources of data.

12

Chapter 1: Introduction to Financial Modeling The FedEx example we will be using for the majority of this textbook is just that, an example. The specific model drivers for FedEx may not be directly relevant for the company you are trying to model, but the methodology is the same. If you can effectively identify the appropriate drivers, than you can build a model for nearly any company.

How to Use This Textbook

This material is designed as a step-by-step demonstration of how to create an earnings model, which you can use as a guide to build a model for a company of your choosing. I have written the text as if I were having a one-on-one conversation with a new member of my research team. This informal approach results in a clear and concise transition of what could otherwise be a complex topic. Throughout the material, I have included my personal experiences in equity research, public accounting and risk management, as well as modeling pitfalls to watch out for, and frequently asked questions I have received from a team of Gutenberg Research contributors. This book will primarily use a FedEx Corporation earnings model as the subject company for our demonstration. The spreadsheets shown throughout the chapters are available free of charge when you register your textbook. FedEx was chosen for this material because the company’s financial reporting is relatively straightforward. The concepts covered are generally applicable to most companies. Using this real-world case study approach rather than a textbookstyle example demonstrates that equity research and earnings modeling can be difficult, and requires a significant number of assumptions. Despite this fact, modeling is a key part of the process of developing a reasonable basis to support an investment thesis. Book Registration: Use the following link to fill out the registration form and gain access to the textbook spreadsheet files: https://www.gutenbergresearch.com/book-registration-third-edition.html or from the GutenbergResearch.com homepage, select “Education” from the website taskbar, then “Book”, and “Book Registration-Third Edition”.

13

After you register, you will be given access to the Template Download Page, which you can access at the following link: https://www.gutenbergresearch.com/templates-third-edition.html or visit GutenbergResearch.com and click on the “Education” drop-down menu from the website taskbar. Then select “Book”, “Book Registration-Third Edition”, and finally “Templates–Third Edition”.

The Earnings Cycle

Earnings models require constant updates for new information. The most common model updates are quarterly earnings releases. Analysts tend to update their models multiple times each quarter. The first update typically occurs just before the quarterly earnings release, then the night of the earnings release, and again after the filing with the U.S. Securities and Exchange Commission (SEC) is submitted, assuming the filing contains additional information required in the model. Other than these predictable model update points, if there are any significant news events or product releases, analysts may also change their models to reflect the latest information throughout the quarter. The image below is a demonstration of the fiscal fourth quarter 2018 earnings cycle for FedEx Corp. Since FedEx follows a fiscal reporting schedule, the quarterly results do not align with calendar quarters (the fiscal fourth quarter ends May 31, 2018). The “consolidation period” represents the time after the reporting date, when the accounting and reporting group is busy preparing the financial statements. This is usually a “quiet period” for the company where management does not typically discuss matters with research analysts or the press until after the financial statements are released. FedEx issued an earnings press release and held an earnings conference call on June 19, 2018. The 10-K was filed with the SEC shortly after the earnings release on July 16, 2018, after the financial statement audit concluded. This marks the completion of the earnings cycle, which will then repeat again in the next fiscal quarter.

Exhibit 2—FedEx Fiscal Fourth Quarter Earnings Cycle

Research analysts at investment banks typically prepare earnings preview research notes and update their models about a week before the press release date. When the press release is posted to the company’s investor relations page, the analysts will update their models to include the actual results for that quarter, and change their future period forecasts based on comments made by management in the press release or on the earnings conference call. After their model is updated they will publish it with a research note to their clients later that night.

14

Chapter 1: Introduction to Financial Modeling My Experience: I prefer to update my models as soon as the press release hits the wires so I can focus on the earnings call when it starts. Some companies start their earnings call by reading what was already written in the press release as part of the prepared remarks. Try not to zone out during this section of the call, and be sure to stay focused for the important parts, like the Q&A with the analysts. Some companies will read a quick summary on the call and then skip directly to the Q&A session (Netflix and Tesla are good examples of this approach). Be sure to listen to a few earnings calls for the company you are covering to gain an understanding of how their calls are conducted, before attending your first live call.

Timing of Publication

This textbook uses a real-world example with actual results disclosed in the FedEx Corp SEC filings. Chapter 1 through Chapter 8, Step 34 reflect details from the FedEx fiscal fourth quarter 2018 results. The first forecast period represented in the text is the fiscal first quarter 2019. Chapter 8 Steps 35 and 36 demonstrate how to update an earnings model for future results, and is based on the FedEx fiscal first quarter 2019 results. As FedEx releases new quarterly financial results in the future, the tables shown throughout this book will not reflect the most recent information. Despite the constant maintenance required to sustain an up-to-date example, the methodologies demonstrated in the material should remain relatively constant over time; Therefore, this book will continue to be a useful demonstration of the earnings modeling process even after FedEx reports future results. To maintain the relevance of the data, entries will be added to the appendix in future editions, which will track the changes made to the FedEx earnings model after the fiscal first quarter 2019 results.

Basic Excel Functionality

As you read through the material you will notice that none of the advanced spreadsheet features (macros, links to external spreadsheets or data sources, etc.) are used in the modeling approach; however, if you are new to Excel, there are some basic features which you should get familiar with before you begin to build your first earnings model. Freeze Panes: It helps to keep the column dates along the top of the spreadsheet, and some users may prefer to keep the target share price in the upper left-hand corner of the screen to see the impact assumption changes have on valuation. To do this select the cell where you want the panes to freeze. Select “View” then “Freeze Panes”. This will let you scroll left or right, up or down, and keep the header with each column’s date, and the account for each row, in place. Grouping Rows and Columns: The row count will increase as variables are added to the model, which can get distracting. You can group sets of rows or columns to make your model more user-friendly. I tend to group columns by year, and rows by product or segment. I also group the Balance Sheet and Cash Flow Statement so I can easily collapse these sections when I am focusing on earnings. To create a group, highlight the rows you would like to include. Select “Data” and “Group”. Notice that a group symbol has appeared in the left-hand column of Excel. Now you can easily collapse the group by clicking the “-” symbol, and then re-expand the group by clicking the “+” symbol. Show Equations: There is a simple and useful trick in Excel which instantly converts all of the cells in the worksheet to show the underlying equations. Hold down control and select the “~” key. To change the view back to normal, hold control and select “~” a second time. Trace Equations: Since you will be using templates with equations that I have created, you may want to trace which cells were used for certain financial statement line items. To determine which cells are used in a calculation select “Formulas” and “Trace Precedents”. Arrows will appear which will clearly show the flow of values through the equations. If you would like to see if a cell impacts other cells, click “Trace Dependents”. When you are finished with your review click “Remove Arrows”. Dragging Equations and Fixed References: Most analysts setup their models to have four quarter columns followed by one column which represents the full year. We will also use this method in our demonstration. Keep in mind that with this style of modeling, you cannot drag columns with quarterly calculations across the spreadsheet as the cell references will be incorrect after you pass through the full year column. Instead of dragging, consider copying and pasting the equations which will maintain the correct cell references. In some cases you may want to keep a fixed reference on a row, column or both. To do this you can add a “$” in your cell references within the equations, by using the F4 key, or manually typing in the “$” symbol. Within the equation

15

point to the cell for which you are looking to fix the reference (for example cell =A1). Press F4 once to lock both the row and column (=$A$1). Press F4 again to lock the row only (=A$1). Press F4 a third time to lock the column (=$A1). Other Excel Functions: Throughout the book I will introduce other important spreadsheet functions in the sections where they are relevant. These functions include: 1) Using data tables for sensitivity analysis (refer to Chapter 8 step 32 for details). 2) Using the goal seek function (refer to Chapter 4 Step 17). 3) Running regression analysis in Excel (refer to Appendix 1).

Types of Models

I use different types of models for different purposes. While models can have an unlimited number of styles and variations, my models generally follow one of three formats which I refer to as the: 1) Back of The Envelope (BOTE) Model, 2) Basic Model, and 3) Advanced Model. These model styles vary based on the type of analysis I am performing. For example, sometimes I want to run a quick sensitivity review on a company’s earnings before a quarterly release. For this type of analysis it does not make sense to populate years of historic data and all three of the financial statements. Instead I create a basic Income Statement with a few assumptions to drive the earnings forecast. This is my Back of The Envelope (BOTE) Model format. In other cases, I may want to add a bit more detail to the model to allow some flexibility in the forecast. For example, in our FedEx demonstration from Exhibit 1 we discussed including package volume and revenue per package estimates. For this type of analysis I would have to add a breakdown of the company’s segments to the Excel file. If I include the Income Statement and the product or segment details, I consider this a Basic Model. If I want to use a discounted cash flow valuation, then I would have to project the Cash Flow Statement and Balance Sheet. For this type of analysis I use the Advanced Model, which links the financial statements together based on my earnings projections and other assumptions. There are many other forms that earnings models can take. I use these three basic styles to frame my analysis based on what I am trying to accomplish. The Advanced Model will be the focus of this book.

Table 1—Types of Earnings Model Description

BOTE Model

Basic Model

Advanced Model

Level of difficulty

Beginner

Intermediate

Advanced

Financial Statements

Summarized version of the Income Statement

Valuation Methodology

Market Multiple

Market Multiple

Forecast Horizon

End of current fiscal year plus four full quarters.

End of current fiscal year plus four full quarters.

• •

Income Statement Segment Details

• Income Statement • Segment Details • Balance Sheet • Cash Flow Statement • Market Multiple • Discounted Cash Flow Approximately five years

Anatomy of a Model

This section covers the primary components included in the Advanced Model, using a completed earnings model for FedEx as an example, shown in Exhibit 4 at the end of this chapter. For the purpose of this section, the rows in the example model have been split into six parts, and some of the cells have been hidden in order to fit the full model on two pages. This section is designed to be a preview of the topics which will be covered throughout the next few chapters. At this point in the process, you may not fully understand the components within the model. All of the concepts will be explained as you work your way through the text. Anatomy of a Model—Color Legend: Keeping track of all the assumptions in a spreadsheet can be difficult. I use a basic color coding process to help sort the different types of cells in the file. Blue cells represent the key assumptions, or 16

Chapter 1: Introduction to Financial Modeling inputs into the model. These are the cells you would want to focus on if you were adjusting one of my models to incorporate your own assumptions. To demonstrate, revisit the FedEx example used earlier in Exhibit 3 below. If you want to change the assumptions for the U.S. Overnight Box Segment volume or yield, you would change the blue cells S74 and S76. Alternatively you could directly input the total U.S. Overnight Box revenue in cell S77; however, this cell is not shaded blue, therefore, it is not an input cell. Instead the cell is dependent on yield and volume growth rates, which are blue input cells. If you manually typed in the number over the equation in S77, you would break the link between the volume, yield, and revenue. By shading the input cells blue, you can quickly identify which cells should be changed and which should be left as is.

Exhibit 3—FedEx Corp Package Revenue Model Example

I also use colors to indicate how a certain metric has been calibrated (refer to the color code in Exhibit 4 cells B3, B4, and B5). For example, orange cells represent financial statement line items which have been calibrated to meet consensus estimates (discussed in Chapter 4, Step 17), while purple cells represent guidance from company management (discussed in Step 18). Notice that the column headers are also color coded (refer to cells Q11 and S11). This is to allow the user to quickly see which columns represent historic results (dark gray columns), and which represent future period forecasts (light gray). The cells within the historic columns generally do not have any color coding applied. This is because historic cells represent actual values which can be found in the company’s past financial statement filings. For example, cell Q46 represents the actual dividend growth rate for the fiscal fourth quarter of 2018, while cell S46 represents the future period projected dividend growth rate. In some cases historic results may include blue cells, which indicate that a particular metric was not disclosed by the company, and the value represents an estimate, even-though it exists in a historic column. If a cell within a future period column is not color coded, than the cell represents an equation derived from other inputs. For example, cell S13 for total revenue uses an equation based on the revenue forecast of each segment rather than a direct input value in the cell. In general, it is best to leave these cells unchanged to maintain the integrity of the embedded equations. Anatomy of a Model—Segment Details and Assumptions: The next section (refer to Exhibit 4 rows 48 through 239) contains the majority of the inputs which feed the Income Statement (rows 11 through 46). This section will be covered in Chapter 2.

17

Anatomy of a Model—Balance Sheet: The Balance Sheet begins on row 241 and continues through row 284. Rows 286 through 301 contain ratio analysis and assumptions related to the Balance Sheet forecast. Most analysts separate the Balance Sheet, Income Statement, and Cash Flow Statement into separate worksheets; however, for our demonstration we will keep all three in the same worksheet to emphasize the point that the financials are interrelated into one integrated projection model. This section will be covered in Chapter 3, Steps 8, 9, and 10. Anatomy of a Model—Cash Flow Statement: The Cash Flow Statement essentially reflects the cash impact of earnings and changes in Balance Sheet accounts (refer to Exhibit 4 rows 303 through 346). Similar to the Balance Sheet, the Cash Flow Statement has a section where the assumptions are entered in rows 348 through 354. This section will be covered in Chapter 3 Step 11. Anatomy of a Model—Valuation Inputs: The valuation summary section shows the inputs used in the share price valuation and target price band (refer to Exhibit 4 cells C6 through C9). The details which feed the summary are include in the market multiple section (cells B356 through C366), the discounted cash flow section (cells B368 through C398), and the risk estimation section (cells B400 through C421). This section will be covered in Chapter 5, 6, and 7.

18

Chapter 1: Introduction to Financial Modeling

Exhibit 4—FedEx Corp Advanced Earnings Model (Income statement/Segments/Ratios)

19

Exhibit 4 (continued)—FedEx Earnings Model (Balance Sheet/Cash Flow/Valuation)

20

Chapter 1 Wrap-Up

CHAPTER 1 WRAP-UP Takeaways •

Earnings models can take many forms. In the context of equity research, the term “model” typically refers to a spreadsheet which houses a collection of assumptions used to predict a company’s earnings and financial statements in future periods.

•

The terms “dynamic” or “working” model, imply that the spreadsheet includes embedded equations which allow the user to change a set of assumptions and instantly see the impact of these changes within the spreadsheet.

•

Key Concept 1—Modeling is a Formalization of Our Opinions: This idea demystifies the term “model”, which some take to mean a black box without transparency over the methodologies employed. Instead a model is simply the mechanism we use to list out our forecast assumptions, for the sake of bringing order to our projection.

•

Key Concept 2—The Principles of Balance and Drive: These two concepts help frame how you should design your model to fit your specific needs. o Balance relates to the level of complexity in your model. You should balance the resulting analytical benefit you accrue from each additional layer of complexity, against the added effort to create it. o

•

Drive deals with the selection of metrics used in the model, which should focus on items that are critical to the related accounts in the financial statements (or “drive” the value of the financial statement line item).

The earnings cycle represents the key dates of the quarterly financial reporting process for U.S. publicly traded companies, marked by three primary stages: o the reporting period which covers the days included in the financial results, o

o

the consolidation period which is the time between the end of the reporting period and the day of the earnings release, and the audit period which is the time between the earnings press release and the filing of the 10-Q/K with the SEC when management finishes their reporting procedures and the external auditors complete their review of the financials.

Concept Quiz Instructions: Answer each of the following questions as “true” or “false” 1) Earnings models which include equations to link primary assumptions are known as regression models. 2) Once an earnings model is created, it will be a useful tool to analyze the subject company indefinitely. 3) Earnings models can come in many different forms with different forecasting and valuation techniques.

21

Instructions: Select the best answer for each of the following questions 4) Flintstone Gravel Corporation, a public SEC registrant, will issue a quarterly earnings press release after the market closes today at 5:00pm EST. What earnings cycle period is Flintstone most likely in? A. The reporting period B. The consolidation period C. The audit period 5) Leo Davis is a top-ranking sell-side research analyst covering Daffy Incorporated, which makes rubber ducks and other popular toys. Leo would like to schedule a conference with his buy-side clients and the investor relations team from Daffy Inc. The Daffy management team declines the meeting. What is the most likely reason for the decline? A. Management of public companies do not attend meetings with clients of analysts. B. Management of public companies do not speak with sell-side analysts directly. C. Daffy is in the consolidation period of the earnings cycle.

Concept Quiz Answers 1) False. Earnings models which include equations to link primary assumptions are known as working models. 2) False. Earnings models require constant maintenance to incorporate the latest earnings releases and other market data. 3) True. Earnings models can come in many different forms with different forecasting and valuation techniques. 4) B. Flintstone is in the Consolidation Period until the press release is issued. 5) C. Company management does attend meetings with sell-side and buy-side analysts. Daffy is likely in the “quiet period” during the consolidation of their financial statements, and therefore does not want to risk disseminating information to a select group ahead of the earnings release, as this would be a violation of Fair Disclosure regulations.

22

CHAPTER 2: BUILDING THE INCOME STATEMENT Step 1: Getting Started

Step 4: Build the Earnings Engine

Step 2: How to Setup Your Model

Step 5: Complete the Income Statement

Step 3: Filling in the Historic Data

Step 6: Adjust for Non-GAAP Items

Chapter 2 Overview: In this chapter we start building our model. If you are new to modeling be sure to move slowly through each step before moving on. This will ensure you do not get overwhelmed later in the process as you move to the more complex areas of the model. We will begin to disaggregate earnings for the primary FedEx segments, and link our future estimates to our defined inputs. As you read through the modeling approach for each account, remember that the basic idea is to identify a driver of the value, and use it to project into the future. The FedEx Income Statement accounts may differ from the company you will ultimately cover; however, if you keep the fundamental “drive” concept in mind, you should be able to develop a model for nearly any company. Note: Throughout this chapter we will be setting up equations to project future values for the Income Statement line items. We will discuss what drives the forecasts when we consider which metrics and equations to use; however, we will not be setting our final future estimates yet. Instead, we will use placeholder until we cover forecasting and model calibration in Chapter 4.

Step 1: Getting Started

Step 1a—Choose a Company: For this textbook we will be building an earnings model for FedEx Corp. After you complete the FedEx model, you may choose to build a model of your own. There are a few things to consider when choosing your next company to cover. If this is your first time building a model, consider choosing a domestic company as foreign based companies listed on domestic exchanges will have an extra currency translation step. This can be somewhat difficult for beginners. Next, make sure your company is not in the process of being acquired, as this adds another layer of complexity. You should also consider the industry. Some industries are inherently difficult to cover such as biotech or energy, while others are more straight forward, such as retail or consumer staples. Finally, consider your level of comfort with the financial modeling process. The easiest companies to cover tend to be well establish organizations with stable earnings, and straight-forward guidance from management. The most difficult tend to be new companies with volatile earnings, and no guidance from management. Step 1b—Get to Know the Company: Before you can begin modeling earnings, you should develop an understanding of the company’s business. The fastest way to do this is by reading the “Management’s Discussion and Analysis” (MD&A) section of the 10-K or 10-Q filings from the U.S. Securities and Exchange Commission (SEC) website (www.SEC.gov). The MD&A summarizes much of the general information you need to know about the company. In fact most companies will explain in great detail how to analyze their results. For example, the FedEx MD&A excerpt below explains the primary drivers of FedEx earnings in just a few easy to read bullets. 23

Exhibit 5—FedEx Corp MD&A Excerpt

Source: SEC.gov, FedEx Corp 10-K, filing date July 16, 2018, retrieved September 24, 2018.

Next, read the latest earnings press release, available on the company’s investor relations page, or from the SEC 8-K filings. This will give you an indication of some of the important items to watch for in the next quarter, and usually includes the guidance management gave for the current quarter. In the fourth fiscal quarter press release below, FedEx management gave key factors to consider in the modeling of 2019 results.

Exhibit 6— FedEx Corp Fiscal Fourth Quarter 2018 Earnings Press Release

Source: SEC.gov, FedEx Corp 8-K, filing date June 19, 2018, retrieved September 24, 2018.

After you read the press release, listen to the last few earnings conference calls. The webcast versions of the calls are typically available on the company’s investor relations page. This will help you get acquainted with how the company conducts the calls, and what the analysts are focused on. Step 1c—Keep Track of the Company: Visit the company’s investor relations page and sign-up for alerts. Some companies use Twitter or other social media outlets to publish updates, so consider following the company to stay upto-date on developments. In addition, you can sign-up for free alert services which will notify you of important company specific news. Keep in mind that these services will send you updates for every piece of content as it is made available, 24

Chapter 2: Building the Income Statement so you will need to get used to sorting out the important developments from the clutter. If you are new to the company and have trouble determining which items are important, one shortcut would be to look at the stock price when the news comes out. If it is not moving much, then the article is probably not meaningful. Some companies issue monthly updates. For example, many retailers publish monthly same store sales statistics. Check your company’s press releases to see if they publish similar reports. If so read the updates when they come out, and consider including the details in your model.

Step 2: How to Setup Your Model

Step 2a—Start with an Existing Template: This textbook includes a blank model template which you can download after you have registered your book: https://www.gutenbergresearch.com/book-registration-third-edition.html The template is a helpful starting point. Keep in mind that different companies will have vastly different financial reporting, so you will need to make some changes to the template. Follow Along in the Spreadsheet: Refer to “File 1–Blank Model” for a template to help you get started. Refer to the "How to Use This Textbook" section of Chapter 1 for instructions on how to access the spreadsheet files. Step 2b—Align the Reporting Schedule in the Header: Not every company’s fiscal year coincides with the calendar year. Take a look at the last press release from FedEx. Note that the report date on the financial statements does not match the calendar quarter since the company’s fiscal year ends on May 31st not December 31st. This means we will have to update the heading sections in the model to coincide with the company’s fiscal quarters (refer to Exhibit 7 below). Notice I have added an “F” in front of the quarter “F1Q18” to indicate that the column represents the company’s “fiscal” quarter. For example, “F1Q18” means this column is the first quarter of the fiscal year ending in 2018. I also use a letter “E” after the year to indicate which columns represent future estimates (light gray columns). The dark gray columns without the letter “E” represent the historic periods which have already been reported. Note: Refer to the “Timing of Publication” section of Chapter 1 for details on the reporting dates used in this textbook.

Exhibit 7—Aligning the Header with the Fiscal Reporting Calendar

Step 2c—Format the Income Statement: Make sure the Income Statement in the model is consistent with the company’s presentation within the financial statements. To do this you will need to rename each of the accounts based on the account names in the SEC filing. The Income Statement for most companies is somewhat similar, although you may need to insert or delete rows from the template to fit the company’s format.

Step 3: Filling in the Historic Data

Next, enter the financial values from the SEC page. Be sure to work your way backward in time, because the company may have made corrections after a previous release. Keep in mind that after each set of four quarters, there is a column which represents the full year. Do not hardcode the values in these full year columns. Instead insert equations which represent the sum of the individual quarters for that year. Input the values for the Income Statement line items in your model. Companies report on a comparable basis, meaning two year-over-year data points are disclosed at a time. As a result you should only need to download four filings to get data for eight quarters (not eight separate filings). For the purpose of this chapter, we will input two years of historic results. Feel free to go back further in time if you would like to include additional data points in your model.

25

Exhibit 8—Formatting the Income Statement With Historic Data

Source: SEC.gov, FedEx Corp 10-K, filing date July 16, 2018, retrieved September 24, 2018.

My Experience: There have been many advancements in financial data automation over the last few years. The Financial Accounting Standards Board (FASB) publishes a taxonomy which consists of eXtensible Business Reporting Language (XBRL) tags for financial statement disclosures. These tags are used by third party data aggregators to make inputting financial data into an Excel file as easy as refreshing the worksheet. If you have access to a data aggregator, it may save you some input time. I usually copy/paste the data from the SEC webpage. Some companies provide historic financial statements in Excel on their investor relation page, which can make this step much easier. Follow Along in the Spreadsheet: Refer to “File 2–Historic Income Statement (Step 3)”. This file will save you some time since the last eight quarters of reported results have already been filled in. Refer to the "How to Use This Textbook" section of Chapter 1 for instructions on how to access the spreadsheet files.

Step 4: Build the Earnings Engine (Revenue Forecast)

Step 4a—Decide What Will Drive Your Earnings Forecast: This part takes some judgment, and is probably the most difficult part of the modeling process. You will need to decide what metrics you are going to use to build the future earnings forecast. For most companies, this will be based on the segment details provided in the press release. For others, you may need to run some calculations to develop the key drivers of your model. The most important aspect is to incorporate any metric management includes in their guidance, since management’s focus indicates that the metric is a key driver of results. In addition, by including these metrics, you can calibrate the model to meet the guidance, or at least identify if your model estimates vary from management’s expectations. FedEx provides investors with a substantial amount of data, which is one of the reasons I selected the company for this book. In addition, management provides the data in an Excel file called the “Statistical Book” which is available at www.fedex.com/us/investorrelations (under the “Financial Information” Section). There are many different methods you can use to breakdown earnings. I will use a detailed approach for demonstration purposes. Feel free to use a higher level of disaggregation to keep your model simple if you prefer. First, I will break revenue down between the primary segments: 1) Express, 2) Ground, 3) Freight, and 4) Services and Other. Then, since the Express Segment represents a 26

Chapter 2: Building the Income Statement large component of total earnings, I will further break this segment down by the two primary sub-segments: Packages and Freight (note that there is a separate non-Express Freight Segment). I will also breakout the Express-Package and Express-Freight into sub-categories. The Express-Package business will have the following six sub-categories: 1) U.S. Overnight Box, 2) U.S. Overnight Envelope, 3) U.S. Deferred, 4) International Priority, 5) International Economy, and 6) International Domestic. The Express-Freight business will have four sub-categories: 1) U.S. Freight, 2) International Priority Freight, 3) International Economy Freight, and 4) International Airfreight. This may seem like a lot of categories for one segment, but keep in mind that the Express Segment makes up more than half of the total revenue. The detailed segmentation will give us the flexibility to alter our future expectations for a specific portion of the business. For example, if a new competitor enters the small package overnight U.S. market, we can alter our growth rates just for that specific sub-category. Or if you would like to forecast a scenario were the U.S. economy strengthens while the rest of the world weakens, you may want the additional segmentation to enter your forecasts for each sub-category. At this stage you should revisit the Balance and Drive concept discussed in the introduction. If you do not require all of the details in your model, you may want to keep it simple and just breakout the primary segments. FAQ 2—Why should we bother disaggregating the results at all? You could apply a simple growth rate to top-line revenue and not disaggregate results by segment. It is important to remember, the point of modeling is to break the results down into smaller pieces to help us understand what the potential future outcomes could be. If you are analyzing FedEx prior to a new earnings release, wouldn’t you want the ability to run sensitivity analysis on the number of packages shipped, and revenue per package earned ahead of the quarterly results? If your model does not break results down by these metrics, you would not be able to do this. Next we must decide how we will project earnings for each of the components. If you listen to the earnings calls, you will hear management discuss the volume of package shipments and the yield, which is the revenue per shipment. These two key factors are also include in the Statistical Book. They are perfect for our model. First, to get an understanding of how these metrics work, we can bring the historic values into our model and reconcile the revenue calculation. We will start with the Express U.S. Overnight Box segment sub-category. Notice in Exhibit 9 I have taken four metrics from the company’s Statistical Book: 1) Average Daily Volume (ADV), 2) yield, 3) U.S. Overnight Box revenue, and 4) operating weekdays. I have put the weekday count in a separate section at the top of my segment breakout, since I will also be using it for the other sub-categories. Next, I can recalculate the revenue to make sure I understand the components correctly. In this case we would like to keep our revenue amounts in millions. ADV is stated in thousands, so we will have to divide our result by 1,000. U.S. Overnight Box revenue = Average Daily Volume × the operating day count × yield ÷ 1,000 = 1,257 thousand packages × 65 days in the quarter × $23.26 per package ÷ 1,000 = Revenue of $1,901M The equation reconciles. If the calculation did not work, meaning if the total revenue we recalculated did not tie to the historic reported results, this would be an indication that we have misunderstood the metrics that management has disclosed, and would have to do further investigation before proceeding to the next step. Since it checks out, we can proceed with filling in the rest of the historic quarters.

Exhibit 9—Reconciling Historic U.S. Overnight Box Metrics

27

Step 4b—Breakout Segments and Input Equations for Future Period Forecasts (Express Segment–U.S. Overnight Box): Now we must consider how we will project the future quarter revenue starting with the quarter ending August 31, 2018 which is the first fiscal quarter of 2019 (Refer to the “Timing of Publication” section of Chapter 1 for details on the reporting dates used in this textbook). The concept of ADV and yield are a bit abstract, so it would be difficult for us to form an opinion of the exact value for these metrics in the next quarter. Instead, we can apply a growth metric relative to the last reported value. This will make the concept less abstract since it links the forecast to the historic value, giving us a reference point for our future projections. We could use a quarter-over-quarter growth rate, however, for companies with seasonal business trends, a year-over-year growth rate is more appropriate. Insert the year-over-year growth rows for ADV and yield, then calculate the historic growth rates. Now use the blue color code for the three rows which will be the inputs to our future forecast: 1) weekday count, 2) ADV growth rate, and 3) yield growth rate. For this step in the process set the future forecast blue cell inputs equal to the average of the last four quarters, except for the day count which you can derive yourself by counting the weekdays on a calendar, or you can take the count from the FedEx Stats Book. We will discuss how to choose the estimates for the growth rate cells later in Chapter 4: Model Calibration & Forecasting. Next we will input the equations for ADV which is: 1 + the year-over-year growth rate × the ADV from the comparable quarter of the previous fiscal year The equation for yield follows the same methodology. Note that the equations for all of the future periods are identical. Keep in mind the comparable quarter reference will change with each future quarter in the forecast. Be sure to reference the correct quarter to apply the growth rate.

Exhibit 10—U.S. Overnight Box Revenue Forecast Equation

Notice that in Exhibit 10 I have entered equations in cells S73 through V73 and S75 through V75 which are driven by the estimates in the blue cells S74 through V74 and S76 through V76. You may be wondering why we use these equations to back into the revenue estimate instead of just typing our revenue estimate in directly. The reason is that we are creating a “working model” (as discussed in the Introduction). Using the equation-based method will allow you to type in different estimates for the two growth rates, and instantly see the impact on earnings. Once you have ADV and yield setup, the revenue calculation is consistent with the historic results: U.S. Overnight Box revenue = Average Daily Volume × the operating day count × yield ÷ 1,000 Before you move on to the other segment sub-categories, let’s make sure you fully understand the equations we are using here. Table 2 summarizes the method we use to create the working model for the U.S. Overnight Box segment sub-category.

Table 2—Summary of Future U.S. Overnight Box Assumption Equations* Line Item

Driver (Assumptions)

Future Period Equation (F1Q19E)

ADV (in thousands of packages, cell S73)

Growth rate (S74)

F1Q18 N73 × (1+ growth rate in S74)

Yield (revenue per package, S75)

Growth rate (S76)

F1Q18 N75 × (1+ growth rate in S76)

U.S. Overnight Box revenue (S77)

Days, ADV, and yield

Days (S59) × ADV (S73) × yield (S75) ÷ 1,000

28

Chapter 2: Building the Income Statement *Note that the cell references in Table 2 correspond to the Excel file in Exhibit 10 above. The growth rate used in the equations above represent the year-over-year rate. Step 4c—Breakout Segments and Input Equations for Future Period Forecasts (All Other Express Segment Categories): Once you are comfortable with the earnings engine for the U.S. Overnight Box sub-category, you can copy and paste these rows for most of the other sub-categories since they will use a similar methodology. You can do this by copying rows 73 through 77 (refer to Exhibit 10), pasting below the U.S. Overnight Box sub-category. After your rows are set, change the category names for each of the following: 1) U.S. Overnight Envelope, 2) U.S. Deferred, 3) International Priority, 4) International Economy, 5) U.S. Express Freight, 6) International Priority Freight, 7) International Economy Freight, and 8) International Airfreight. Step 4d—Breakout Segments and Input Equations for Future Period Forecasts (Express Segment Totals): Before moving on to the other segments, we should create a subtotal section for the Express Segment to aggregate the Package and Freight sub-segments. This will prove helpful when we begin to model the operating expenses after the revenue forecast is complete. Exhibit 11 below shows one way to approach the Express Package and Freight subtotals.

Exhibit 11—Express Segment Total Section

There are two important things to point out for this section. First, notice that the Express Segment has an “Other Express” category to classify revenue which does not fit in the Package or Freight sub-segments. Given that the Other Express revenue is a relatively small value, we can use a simple year-over-year growth rate to project the future quarters (cells S58 to V58). Second, besides the “Other Express” revenue, there is no new modeling going on in this Express Segment subtotal section. For the historic and forecast quarters, the revenue and ADV are simply adding the individual Package and Freight sub-categories, and the composite yield is a weighted average of the sub-category yield. This point is demonstrated in Exhibits 12 and 13 which show the Express Package and Freight sub-segment details and the subtotal section of the model. For example, the total package ADV in cell V51 (refer to Exhibit 12 below) is equal to the sum of: U.S. Overnight Box ADV (V73) + U.S. Overnight Envelope ADV (V79) + U.S. Deferred ADV (V85) + International Priority ADV (V91) + International Economy ADV (V97) + International Domestic ADV (V103) = 1,271 + 539 + 891 +532 + 291 + 2,386 = 5,910 Total package composite yield in cell V51 is calculated as the weighted average of the individual sub-category yields: • • • • • •

U.S. Overnight Box (V73 ÷ V51) × U.S. Overnight Box yield (V75) = 1,271 ÷ 5,910 × $24.29 = $5.23 U.S. Overnight Envelope (V79 ÷ V51) × U.S. Overnight Envelope yield (V81) = 539 ÷ 5,910 × $13.72 = $1.26 U.S. Deferred (V85 ÷ V51) × U.S. Deferred yield (V87) = 891 ÷ 5,910 × $17.02 = $2.56 International Priority (V91 ÷ V51) × International Priority yield (V93) = 532 ÷ 5,910 × $60.98 = $5.48 International Economy (V97 ÷ V51) × International Economy yield (V99) = 291 ÷ 5,910 × $52.61 = $2.59 International Domestic (V103 ÷ V51) × International Domestic yield (V105) = 2,386 ÷ 5,910 × $8.48 = $3.42 o Composite Package Yield = $5.23 + $1.26 + $2.56 + $5.49 + $2.59 + $3.42 = $20.54

Total package revenue in cell V53 is equal to the sum of: U.S. Overnight Box revenue (V77) + U.S. Envelope revenue (V83) + U.S. Deferred revenue (V89) + International Priority revenue (V95) + International Economy revenue (V101) + International Domestic revenue (V107) = $2,007M + $481M + $985M +$2,107M + $995M + $1,315M = $7,891M

29

Exhibit 12—Express Segment Packages Sub-Segment

Exhibit 13—Express Segment Freight Sub-Segment

Step 4e—Breakout Segments and Input Equations for Future Period Forecasts (Ground Segment): The FedEx Ground Segment is much more straight forward compared to the Express Segment. We will not be breaking this one into subcategories, although there are two components: 1) The primary component is a function of Average Daily Freight pound (ADFlb), which is similar to ADV except it is stated in weight instead of number of packages, and yield which is the measure of revenue per freight pound. 2) The secondary component is the “Other Ground” revenue. We can use a simple year-over-year growth rate to forecast this, similar to the “Other Express” revenue classification in Step 4d. The calculation of total Ground Segment revenue as demonstrated below in Exhibit 14 is: Ground Segment revenue = ADFlb × the operating day count × yield ÷ 1,000 + All other ground revenue Similar to the Express Segment, for the modeling methodology, we will use the year-over-year growth rate in ADFlb (cells S136 through V136), and yield (S138 through V138) to drive the Ground Segment revenue projection.

30

Chapter 2: Building the Income Statement

Exhibit 14—FedEx Ground Segment Forecast*

*Reminder: Do not be concerned with what values to enter in the blue forecast cells at this point. The values shown in the blue cells in this section are simply temporary placeholders. We will discuss how to approach entering these values in Chapter 4 which covers model calibration and forecasting. Step 4f—Breakout Segments and Input Equations for Future Period Forecasts (Freight Segment): In the FedEx Stats Book, the key performance metrics for the Freight Segment are based on shipment count and weight. You can choose which metric to use in your forecast. I have used shipments per day and revenue per shipment. The calculation of total Freight Segment revenue as demonstrated below in Exhibit 15 is: Freight Segment revenue = shipments per day × the operating day count × revenue per shipment ÷ 1,000 Similar to the other segments, for our modeling methodology we will use the year-over-year growth rate in shipments per day (cells S153 through V153) and revenue per shipment (S157through V157) to calculate the Freight Segment revenue projection.

Exhibit 15—FedEx Freight Segment Forecast

Step 4g—Breakout Segments and Input Equations for Future Period Forecasts (Services and Other Segment): The two remaining segments are the FedEx Services Segment, and “Corporate, Other and Eliminations Segment,” which includes earnings from the FedEx Trade Networks business, as well as accounting eliminations which take place as part of the segment consolidation financial reporting process. For the services business, I use a simple year-over-year growth rate to project revenue. The “Other” category is difficult to project since it is the catch-all bucket, and includes consolidation entries. For now I have set the forecast equal to the average of the last four reported quarters.

Exhibit 16—FedEx Services and Other Segment Forecast

Step 4h—Link Segment Revenue Detail to the Income Statement: Now that we have our revenue estimates for each individual segment from Steps 4a through 4g, we can link the sum of these individual components to the revenue line at the start of the Income Statement. You can do this by writing an equation which references the revenue of each segment directly in Income Statement, or creating a subtotal in the segment section below the Income Statement. Be

31

sure to add the total revenue equations for each of the forecast quarters you are modeling, as we will be using the total revenue across the entire forecast horizon.

Exhibit 17—Linking Segment Revenue Details to the Income Statement

Follow Along in the Spreadsheet: Refer to “File 3–Segment Details (Step 4)” to see a breakdown of the Segment details used to create the revenue forecast. Refer to the "How to Use This Textbook" section of Chapter 1 for instructions on how to access the spreadsheet files.

Step 5: Complete the Income Statement

For some companies, you may decide to model other components of the Income Statement in the segment section. For example, since FedEx discloses the operating expense detail by segment, we can add these line items to our segment forecast. Before we begin let’s revisit the Balance and Drive concept to help plan our operating expense modeling approach. Refer to the segment operating expense details that FedEx discloses in Exhibit 18 below. Is it really necessary to model all of the expense lines for each of the segments? In reality, even if we did break each item out in our model, we may not be able to form an opinion for all of these future forecast inputs. In some cases it may actually be easier to include all of the detail, since we could simply add each line item to get to the total operating expense for the consolidated Income Statement. I will demonstrate the difficulty in not projecting each line, and you may decide which individual items (or all of the operating expense lines) to include in your version of the model. Let’s focus on the line items where we may be able to make an informed projection. For example, we can use jet fuel and gasoline prices to project fuel expense. We could also use capital spending projections (i.e. aircraft purchases and other capital items) to project changes in depreciation expense. The remainder of Step 5 will demonstrate how to break out a selection of individual items, and how to project the remaining components of operating expense.

Exhibit 18—FedEx Segment Operating Expense Disclosure

Source: SEC.gov, FedEx Corp 10-K, filing date July 16, 2018, retrieved September 24, 2018. 32

Chapter 2: Building the Income Statement Step 5a—Add Sections for Segment Operating Expenses: I have incorporated the Express Segment operating expenses below our Express Segment subtotal revenue section (refer to Exhibit 19 below). Notice that have I split the fuel expense between jet fuel (calculated as the number of gallons used × the price per gallon) and other fuel expense. The additional fuel details are include in the FedEx Stats Book. You can use the Stats Book to input the historic values for the segment operating expenses. For the “All Other Operating Expenses” category I have used a ratio of the total expenses-to-revenue for the future period forecast. We may be able to use comments from management on employee salary expectations (which could change with minimum wage requirements), or changes in purchase transportation costs. We will discuss these concepts further in Chapter 4: Model Calibration & Forecasting. For now, I have set the blue forecast cells equal to the average of the last four reported quarters.

Exhibit 19—Modeling Operating Expenses

After you have completed the Express Segment, you can move on to breakout the operating expenses for the other segments by: 1) fuel expense, 2) depreciation and amortization, and 3) all other operating expenses. There is an additional step to forecasting depreciation expense, since it requires the Balance Sheet value for property and equipment, as well as the capital expenditure estimate from the Cash Flow Statement. For the time being, set the forecast for depreciation expense in the Income Statement equal to last reported value as a placeholder. Refer to Exhibit 20 below. Note that I have set the estimate in F1Q19E (cell S18) equal to the value from F4Q18 (cell Q18). Next, allocate the depreciation expense back to each segment at the same percentage from the last reported value in F4Q18. For example, the Express Segment depreciation expense from F4Q18 was 53.7% percent of the total consolidated depreciation (431 ÷ 802 = 53.7%). The reason we want to distribute depreciation back to the segments is that it is an important part of the segment operating margin. I have highlighted the depreciation line yellow, to remind us to revisit this forecast after the Balance Sheet estimates are complete in Chapter 3, Step 8f.

Exhibit 20—Depreciation Placeholder and Allocation to Segments

33

Step 5b—Link Segment Operating Expense Details to the Income Statement (Fuel): After you have completed the expense details for each segment, you can bring the totals up to the Income Statement, starting with fuel expense, similar to the approach used for revenue in Step 4h. Refer to the demonstration below in Exhibit 21 for the details. Notice in rows 174 through 177 I have included details of the operating expenses which are not allocated to specific segments.

Exhibit 21—Linking Segment Fuel Expense Details to the Income Statement

Step 5c—Link Segment Operating Expense Details to the Income Statement (All Other): Since we did not model each of the remaining operating expense line items in the segment details, we cannot simply add each of the segments to arrive at the total lines for the Income Statement. Instead we must allocate our forecasts of the “All Other Operating Expenses” line to the remaining operating expense lines on the Income Statement. To do this first calculate the weight of each line to the subtotal of salaries, purchased transportation, rental and landing fees, maintenance, and other expense from the comparable historic quarter (so the F1Q19 estimate will be based on the weights from the F1Q18 historic results). Note that I have excluded impairment charges since these are typically one-time events. If management expects impairments in the future and discloses estimates, you can enter those estimated amounts directly on the Income Statement. I have also excluded retirement plan mark-to-market adjustments which will be explained later in Chapter 3, Step 9f.6a. Using this methodology the weights in Exhibit 21 for each of the expense line items are as follows: • • • • •

Salaries = $5,518 (cell N15) ÷ [$5,518 (N15) + $3,445 (N16) + $818 (N17) + $675 (N20) +$2,270 (N23)] = 43.4% Purchased Trans = $3,445(N16) ÷ [$5,518(N15) + $3,445(N16) + $818(N17) + $675(N20) + $2,270(N23)] = 27.1% Rental/Landing = $818 (N17) ÷ [$5,518 (N15) + $3,445 (N16) + $818 (N17) + $675 (N20) +$2,270 (N23)] = 6.4% Maintenance = $675 (N20) ÷ [$5,518 (N15) + $3,445 (N16) + $818 (N17) + $675 (N20) +$2,270 (N23)] = 5.3% Other Expense = $2,270 (N23) ÷ [$5,518 (N15) + $3,445 (N16) + $818 (N17) + $675 (N20) +$2,270 (N23)] = 17.8%

After you have the weights calculated, the expense for each line equals the sum of the segment forecast for the “All Other Operating Expense” line, allocated by the historic weights: • • • • •

Salaries = 43.4% × [$7,222 (S67) + $3,877 (S146) + $1,517 (S163) + $813 (S173) + $293 (S177)] = $5,950 Purchased Trans = 27.1% × [$7,222 (S67) + $3,877 (S146) + $1,517 (S163) + $813 (S173) + $293 (S177)] = $3,715 Rental/Landing = 6.4% × [$7,222 (S67) + $3,877 (S146) + $1,517 (S163) + $813 (S173) + $293 (S177)] = $882 Maintenance = 5.3% × [$7,222 (S67) + $3,877 (S146) + $1,517 (S163) + $813 (S173) + $293 (S177)] = $728 Other Expense = 17.8% × [$7,222 (S67) + $3,877 (S146) + $1,517 (S163) + $813 (S173) + $293 (S177)] = $2,448

34

Chapter 2: Building the Income Statement

Exhibit 22—Linking All Other Operating Expenses to the Income Statement

Step 5d—Enter Income Statement Reconciliation Checks: After you have brought the operating expense line items to the Income Statement, check that your total is equal to the consolidated total from the historic quarters. I typically include a section of check figures which will turn red if my Income Statement calculations do not equal the sum of the individual segments, plus the corporate consolidation elimination entries. These check equations are in Exhibit 23 below, rows 179 through 183. In addition to reconciling the segment details, these equations also confirm the consistency between the historic accounting and forecast approach. Note that I have included each of the metrics broken out separately in the segment details of our model: 1) revenue, 2) depreciation, 3) fuel expense, 4) other operating expenses, and 5) operating income. The equations simply take the line from the Income Statement and subtract the corresponding rows from each of the segment sections. If the total does not tie to the Income Statement, then there must be a mistake in the approach which would require additional investigation.

Exhibit 23—Reconciliation Checks Between Segment Details and the Income Statement

Step 5e—Non-Recurring Items (Impairments): Management periodically tests assets and reporting units for impairment by comparing the carrying value to the estimated future undiscounted cash flows that the asset is expected to produce. If the cash flows are not greater than the carrying value, then the carrying value is written down to fair value, and an impairment loss is taken on the Income Statement. 35

Impairments are inherently difficult to project with the limited information we have available for forecasting, and the subjectivity around cash flow estimates. If management provides some form of guidance, or if it becomes clear that the company has substantial non-performing assets, you could enter an impairment expectation. Otherwise it may be more appropriate to leave the impairment estimate at either zero, or an average of the historic impairment charges. Step 5f—Periodic Adjustments (Retirement Mark-to-Market Adjustment): FedEx typically books retirement plan mark-to-market adjustments in the fourth fiscal quarter based on the actual return of the plan assets compared to the expected return, changes in the discount rate used to value the retirement obligation, or changes in mortality rates. For now we will set the retirement adjustment to ($911M) as a placeholder, and will explain the details of how to project this metric in Chapter 3, Step 9f. Step 5g—Input Equations for Interest Expense, Interest Income, and Other Income: Similar to depreciation expense, the interest income and expense accounts require Balance Sheet items to derive a forecast. Interest income should be based on a ratio of income to interest bearing accounts and investments, or in the case of FedEx, cash and equivalents. Interest expense is a function of the debt balance. For now, set the interest income and expense lines to the last reported value as a placeholder, and highlight the rows yellow as a reminder to come back after the Balance Sheet forecast is complete. Since interest income is not disaggregated from net interest expense in the quarterly filings, take the annual disaggregated interest income and expense lines and divide by four to get the approximate quarterly value. Other income makes up less than one percent of annual net income, so it is far less important compared to the other accounts. You can attempt to model this account as a percentage of revenue, but given the immateriality I have simply set the forecast equal to the average of the last four quarters.

Exhibit 24—Other Income and Expenses

Step 5h—Input Equations for Tax Expense: Calculate an effective tax rate using historic tax expense, divided by the income before income tax line on the Income Statement (refer to cell Q221 in Exhibit 25 below). The FedEx tax rate was impacted by the Tax Cuts and Jobs Act (TCJA) U.S. tax legislation which resulted in a change to the top corporate statutory federal income tax rate from 35% to 21%. The change will be phased in resulting in a U.S. statutory federal rate of 29.2% for 2018 and 21% after that. After the change is legislation the company’s U.S. deferred tax liability was remeasured, which resulted in a negative effective tax rate. The tax rate will likely stabilize going forward. Therefore we can use management’s effective tax rate guidance for future periods. Keep in mind there are additional taxes included in the effective tax rate other than just the U.S. federal income tax.

Exhibit 25—Forecast Equation for Tax Expense

Step 5i—Input Equations for the Future Share Count Estimate: The last assumption needed to calculate the GAAP EPS, is the share count. The number of shares outstanding is impacted by share repurchases, equity awards, and the dilutive impact of convertible securities. Share repurchases tend to be the largest driver of changes in shares. For this reason, I breakout repurchases separately from other share count changes, and use estimates of repurchase amounts and 36

Chapter 2: Building the Income Statement average share price to calculate share count reductions. The remaining share changes are difficult to project, and less material, so I apply a basic average growth rate to project the future estimates. The first step in the process is to confirm that the company is authorized to continue share repurchases. For this we turn to the last SEC filing “Footnote 1: Description of Business and Summary of Significant Accounting Policies” (refer to Exhibit 26 below). According to the disclosure FedEx has an additional 12 million shares remaining for repurchase under the current program. Therefore we will be able to continue modeling repurchases for the foreseeable future.

Exhibit 26—Share Repurchase Program

Source: SEC.gov, FedEx Corp 10-K, filing date July 16, 2018, retrieved September 24, 2018.

Fill in the historic share repurchase details including the share count repurchased, average share price and dollar repurchase amount, from the SEC filings (refer to Exhibit 27 cells Q227, Q228, and Q229). Next calculate an average rate of change in basic and diluted shares, excluding the impact of repurchases (cells Q225 and Q226). For the future forecast equations, set the basic and diluted share count equal to the last report value (Q40 and Q41), plus the estimated share count growth rate (cells S225 and S226, use the four quarter historic average as an estimate), minus the share repurchase estimate which is derived based on an estimate of the average share price (S227, you can use the current market share value), and the estimated share repurchase dollar amount (S228). After you have the next four quarters of basic and diluted share count estimates, you can enter the annual share estimate as a weighted average of the four quarters, weighted by net income ( W40 and W41).

Exhibit 27—Input Equations for Share Count

The share repurchase estimates we have calculated here will also be used in Chapter 3, Step 10 when we forecast the equity section of the Balance Sheet, and Step 11 when we forecast the cash flows from financing activities. After we have a fully integrated model, we will be able to automatically incorporate the impact of a new share repurchase program, by removing the cash needed for repurchases from the Balance Sheet through the Cash Flow Statement. My Experience: If you end up working in research at an investment bank you will most likely be required to submit your models into an automated system which checks for errors based on a series of calculations and comparisons. This can cause many validation problems, one of which is checking the sum of the four quarters of EPS against the annual total. If you use a simple average share count for the year-end column, you will most likely end up with a validation error. To avoid the issue, you should set the equation for the full year share count to be a weighted average based on net income, rather than a simple average. Step 5j—Calculate EPS: At this point you have all the inputs necessary to calculate Earnings Per Share (EPS), which is simply net income divided by shares outstanding. Table 3 summarizes the Income Statement assumptions and calculation from revenue through EPS.

37

Table 3—Summary of Assumptions and EPS Calculation Line Item

Driver (Assumptions)

Future Period Equation

Revenue

ADV, yield, and day count or year-over-year growth estimates

• •

Depreciation expense

Capital expenditure estimates and depreciation-to-Property & Equipment (P&E) ratio

Average Property & Equipment forecast × depreciation-to-P&E ratio

Fuel expense

Jet fuel prices and estimates of fuel usage (in gallons)

Jet fuel price × estimate of gallons used

All other operating expense

Opex-to-revenue ratio

Revenue × opex-to-revenue ratio

Total operating expenses

N/A (subtotal equation)

Depreciation + fuel expense + all other operating expenses

Operating income

N/A (subtotal equation)

Revenue – total operating expenses

Interest expense

Interest rates and debt balance forecast (debt-to-equity ratio)

Debt balance × interest expense-to-average debt ratio

Interest income

Interest rates and investment / cash equivalent projections

Cash equivalents × interest-to-interest bearing accounts ratio

Other income

Immaterial

Historic average

Income before taxes

N/A (subtotal equation)

Operating income – interest expense + interest income + other income

Provisions for income tax

Effective tax rate ratio

Income before tax × effective tax rate

Net income

N/A (subtotal equation)

Income before tax - provisions for income tax

Diluted shares outstanding

Repurchase assumptions & share growth rate

Previous share count × (1 + growth rate) – share repurchase forecast

Diluted EPS

N/A (subtotal equation)

Net income ÷ shares outstanding

ADV × yield × day count Sum of individual segment estimates

Step 5k—Dividend Growth Rate: Dividends are typically reported at the end of the Income Statement in the SEC filings. While dividends represent a distribution of earnings, not an expense, it is important to project future dividend growth as this will impact the cash flow estimate. The equation for the forecast is a simple year-over-year growth rate. Management typically discusses dividend projections on earnings conference calls.

Step 6: Adjust for Non-GAAP Items

Step 6a—Incorporate Non-GAAP Adjustments into the Model: Many companies use non-GAAP adjustments to show what management believes to be a more accurate representation of the company’s true earning capacity. Most nonGAAP adjustments reduce expenses for non-cash items and nonrecurring gains or losses. Non-GAAP adjustments are relatively consistent between companies (other than the recognition of one-time events). These adjustments tend to include adjustments for stock-based compensation, amortization of acquired intangibles, and the tax impact of these

38

Chapter 2: Building the Income Statement items. If the company is going through a restructuring or merger, non-GAAP adjustment will usually strip out the charges related with these events in an effort to show what the comparable financial results may look like in the future. The adjustments are typically divided between reductions in costs of goods sold (which is not relevant in our FedEx example), reductions of operating expense, and adjustments to income tax expense, or directly to net income. Recently, there has been a shift in focus away from non-GAAP measures. For example, Facebook Inc, like many other technology companies, has historically focused guidance on non-GAAP measures. On the first quarter 2016 earnings conference call that approach changed: Before diving into our quarterly financial results I wanted to highlight that beginning this quarter, I will focus my prepared remarks on our GAAP results…The primary difference between our GAAP and nonGAAP metrics is stock-based compensation. Stock-based compensation plays an important role in how we compensate our employees, and therefore we view it as a real expense to the business. -David Wehner, Facebook Inc CFO, first quarter 2016 earnings conference call, April 27, 2016. Soon after Facebook’s announcement, Google parent Alphabet Inc took the GAAP movement a step further by removing the non-GAAP disclosures altogether: ...we are making changes to our non-GAAP reporting. SBC (Stock-Based Compensation) has always been an important part of how we reward our employees in a way that aligns their interests with those of all shareholders. Although it is not a cash expense, we consider it to be a real cost of running our business because SBC is critical to our ability to attract and retain the best talent in the world. Starting with our first quarter for 2017, we will no longer regularly exclude stock-based compensation expense from non-GAAP results. Noncash stock-based compensation will continue to be reported on our Cash Flow Statement, but we will no longer be providing a reconciliation from GAAP to non-GAAP measures that reflects SBC and related tax benefits. -Ruth Porat, Alphabet Inc CFO, Fourth quarter 2016 earnings conference call, January 26, 2017. Despite the recent shift in focus from some notable companies, non-GAAP reporting remains a common practice among the majority of U.S. public entities. Therefore, it is important for new research associates to learn how to incorporate non-GAAP adjustments into their models. The key to successful non-GAAP modeling is to maintain the integrity of the GAAP financial statements, without having to create a separate non-GAAP Income Statement. To do this, add a row after each GAAP metric which the company posts an adjustment to, and run a set of equations down the Income Statement which incorporates all of the non-GAAP items. Although it may appear that the GAAP and non-GAAP Income Statements are comingled in the model, the calculations actually remain separate. This allows us to have two distinct bottom-line EPS estimates side-by-side, one on a GAAP basis, the other non-GAAP. Step 6b—Review the Company’s Non-GAAP Disclosure: FedEx discloses non-GAAP explanations and reconciliations in the press release and SEC filings (refer to Exhibit 28 below for details). In 2018 the most important non-GAAP adjustment for many U.S. companies, including FedEx, was the recognition of the deferred tax liability impact from the Tax Cuts and Jobs Act (TCJA), which resulted in a change to the corporate statutory federal income tax rate from 35% to 21%. Since the tax rate declined significantly, the new estimated deferred tax liability declined from what was previously reported on the Balance Sheet, resulting in a $1.15B tax benefit reflected on the Income Statement (recognized in the fiscal third quarter of 2018). Since this was a one-time benefit to recognize the remeasurement, the non-GAAP adjustment removes the impact from net income. The company’s non-GAAP disclosure includes two other large adjustments: 1) An increase to operating income of $380M to remove a goodwill impairment charge related to the deterioration of the FedEx Supply Chain Business, and 2) A $477M increase to operating income which removes expenses related to the integration of TNT Express (which was acquired by FedEx in May of 2016). Management’s basis for the integration expense adjustment is that after the full integration of TNT, these expenses will no longer exist going forward, although they are expected to continue for the next few years. It is important to watch the level and trend of integration and restructuring expenses since management has some discretion over the classification. Given the size of the TNT acquisition, which is the largest in FedEx history, the integration expenses are probably reasonable.

39

The remaining non-GAAP adjustments are relatively small compared to the company’s total operating income. These include: adjustments for legal expenses which are non-recurring in nature, and the company’s fourth quarter mark-tomarket retirement obligation adjustment.

Exhibit 28—FedEx Corp Non-GAAP Adjustments

Source: SEC.gov, FedEx Corp 8-K, filing date June 19, 2018, retrieved September 24, 2018.

Step 6c—Reconcile the Company’s Non-GAAP Disclosure: At this point in the process we have completed the GAAPbased Income Statement in Step 5. The GAAP results will be used to reconcile the non-GAAP adjustments. Start the reconciliation process by adding a non-GAAP section below the segment details in your model, and list out each of the non-GAAP adjustments the company makes as I have done in rows 230 through 239 in Exhibit 29 below. Notice that I put the financial statement line impacted on each row for reference. Next, enter the amounts for the historic press release disclosures on non-GAAP adjustments (refer to cells N231 through R239).

Exhibit 29—Non-GAAP Adjustment Modeling

Now that you have all the non-GAAP items identified, you will need to link each to the Income Statement. The two Income Statement lines which are impacted by non-GAAP adjustments for FedEx are operating income and net income. Insert rows after each line in the Income Statement. For example, in Exhibit 30 rows I have added rows 26, 27, 38, 39, 40

Chapter 2: Building the Income Statement and 44. Next, insert equations to add the non-GAAP items. Cell Q26 equals the non-GAAP adjustments for operating expenses (“opex”), Q231 + Q233 + Q235 + Q237.

Exhibit 30—Linking Non-GAAP Adjustments to the Income Statement

The final step is to create equations in the Income Statement for non-GAAP operating income, net income, and EPS by incorporating the non-GAAP adjustments with the GAAP measures. Next, compare the resulting non-GAAP net income and EPS from your model calculation (Exhibit 30), to the results reported by the company (Exhibit 28). If they do not reconcile then you have made a mistake, and will need to revisit your calculation. If the values from your model match the non-GAAP disclosure, you can move on to the next step. Follow Along in the Spreadsheet: Refer to “File 4–Income Statement Forecast Equations (Through Step 6)” for the detail of what your model should look like at the end of Chapter 2. Refer to the "How to Use This Textbook" section of Chapter 1 for instructions on how to access the spreadsheet files.

41

CHAPTER 2 WRAP-UP Takeaways •

Setting up an earnings model in a format which is easy to follow and can be replicated across different companies can streamline the earnings forecasting process.

•

Useful resources for learning about a new company you would like to cover include: the “Management’s Discussion and Analysis” (MD&A) section of the 10-K or 10-Q SEC filings, and the investor relations section of the company’s website.

•

The top-line revenue forecast is one of the most important aspects of an earnings model, since it drives many of the other earnings components. Revenue forecast should be disaggregated in the model to include relevant drivers of the future period projections.

•

The subject company’s segment reporting is typically a good starting point for building the earnings engine, which is the primary metrics of the revenue forecast in the model.

•

Due to the fact the Income Statement, Balance Sheet, and Cash Flow Statement are linked through multiple accounts, it is sometimes necessary to incorporate placeholders into your model through the model building process, until the other financial statements are complete.

•

The drivers of the earnings forecast must be properly linked to the Income Statement and reconciled against historic results to ensure consistency in the reporting logic.

Concept Quiz Instructions: Answer each of the following questions as “true” or “false” 1) Research analysts should only consider financial items which have been verified to comply with Generally Accepted Accounting Principles (GAAP). 2) Filling in historic results is a waste of time. Instead Analysts should focus on projecting future results only. Instructions: Use the following summary to answer questions 3 through 10. Rocky Road Mining Supply Corporation manufactures computer graphics cards used to mine digital block-chain currency coins. Nick Magellan, an equity research analyst, has created the following model to project Rocky Road’s earnings for the next quarter. Columns C, D, and E represent the historic financial reporting periods. Column F is the forecast for the next quarter, and blue cells represent the primary assumptions for the forecast.

42

Chapter 2 Wrap-Up

3) What equation should Nick use to project revenue in cell F4? A. F14 ÷ 1,000,000 × F15 B. F14 × F15 C. F14 × F15 ÷ 1,000,000 4) What equation should Nick use to project cost of goods sold in cell F5? A. F4 × F16 B. F4 × F16 × F17 C. F4 × (1 - F16) 5) What equation should Nick use to project gross margin in cell F6? A. F4 – F16 B. F4 – F5 C. F4 + F5 6) What equation should Nick use to project operating expenses in cell F7? A. F4 × F17 B. F4 × F16 C. F4 × F16 – F17 7) What equation should Nick use to project operating margin in cell F8? A. F4 × F17 B. F4 – F7 C. F6 – F7 8) What equation should Nick use to project the provision for income tax in cell F9? A. F4 × F19 B. F8 × F19 C. F8 - F19

43

9) What equation should Nick use to project net income in cell F10? A. F4 – F5 – F8 – F9 B. F8 – F9 C. F8 – F10 10) What equation should Nick use to create a revenue “check” between the Income Statement and the model assumptions across row 20? A. F4 – (F14 ÷ 1,000,000) × F15 B. F4 – F14 × F15 C. F4 – (F14 + F15)

Concept Quiz Answers 1) False. Non-GAAP measures may be relevant in instances where the non-GAAP adjustments provide a more meaningful view of a company’s earnings capacity, or provide additional insight from management. 2) False. Historic results can assist in forming a reasonable basis for future projections. Solution model for questions 3 through 10.

3) A. Revenue should be based on the number of units sold × average selling price per unit. Since the Income Statement is in millions, we will need to convert the number of units. The equation is F14 ÷ 1,000,000 × F15. 4) C. We can back into cost of goods sold using the gross margin assumption. The equation is = F4 × (1 - F16). 5) B. Now that revenue and cost of goods sold have been projected, the equation for gross margin in the forecast period is consistent with the historic period. The equation is = F4 – F5. 6) A. The forecast equation is = F4 × F17. 7) C. Now that gross margin and operating expenses have been projected, the equation for operating margin in the forecast period is consistent with the historic period. The equation is = F6 – F7.

44

Chapter 2 Wrap-Up 8) B. The forecast equation is = F8 × F19. 9) B. Now that revenue, cost of goods sold, operating expenses, and the provision for income tax have been projected, the equation for net income in the forecast period is consistent with the historic period. The equation is = F8 – F9. 10) A. The check equation should compare the revenue from the Income Statement to the units and ASP in the assumptions section. The equation should run across each column, including the historic results. For example, the equation in column C is =C4 – (C14 ÷ 1,000,000) × C15 and column F is = F4 – (F14 ÷ 1,000,000) × F15. Follow Along in the Spreadsheet: Refer to “File 19-Chapter Wrap-Ups”, worksheet “Chapter 2” to access the Excel file used for the questions above. Refer to the "How to Use This Textbook" section of Chapter 1 for instructions on how to access the spreadsheet files.

45

THE GUTENBERG MODELING FRAMEWORK Most new research associates are able to develop forecasts for the Income Statement line items since they are similar across various companies, and earnings projections are generally intuitive in nature. When it comes to the Balance Sheet and Cash Flow Statement, first-time associates sometimes have trouble modeling the accounts as they tend to be a bit more abstract. Before beginning the next chapter, this section will summarize the approach we have used for the Income Statement in the form of overarching steps to the modeling process. Formalizing the approach at this stage should assist with the challenges we will face in the Balance Sheet and Cash Flow Statement sections.

Key Concept 3—An Approach to Model Any Account, for Any Company Develop an Understanding of the Account you are Modeling: Start with a search of the SEC filings for details of what items are included in the account, and any relevant accounting policies that should be considered in your forecast. Next, search the latest press release, company presentations, and earnings transcripts for details about the account or guidance from management. Review Historic Results: Examine the past trends of the account. Are there any one-off events in the past which should be considered in your future forecast? For example, if a new business was acquired this quarter, then the year-overyear growth rates for certain accounts would not be comparable to prior periods; therefore, the growth rates you enter in your future forecast will need to be adjusted away from the historic trends to include the new business. During your review of the historic results you should also examine whether or not the account you are modeling is correlated to another account. For example, depreciation expense is directly related to fixed assets, so the forecast of future period depreciation should be linked to the fixed asset balance. If you cannot determine whether or not a particular account is correlated with another, at least compare it to revenue and total assets to determine if there is a general relationship which should be considered in your forecast. Classify the Account Based on the Future Period Modeling Approach: Now that you understand the account, you can develop an approach to project the future value. Use the following questions to classify the account based on the level of confidence in your ability to project the account in the future: Question 1) Is there a metric available which can be used to develop a reasonable future forecast? Using our FedEx example, if you are trying to model the revenue account using the method described in Chapter 2, Step 4 where we broke out the number of packages (volume) and revenue per package (yield) for each segment. You would answer “Yes” to this question since these metrics would provide a reasonable basis to develop a future period forecast. Similarly, for the fuel expense line you would also answer “Yes” to this question since fuel expense is a function of price and the number of gallons of fuel used during the period. On the other hand, if you are trying to model the “other operating expenses” account, you would answer “No” to this question, as there is not a direct metric which you could use to develop a reliable forecast. Question 2) If you answered “Yes” to Question 1, is there any data available before the company reports results which would provide some degree of comfort in your forecast? For this question consider any guidance management provides, or any data points available outside of the company’s reporting which could assist in 46

The Gutenberg Modeling Framework developing a forecast. For example, in our FedEx revenue forecast, if management provided guidance for expected package volume and yield for each segment, then we would answer “Yes” to this question. Since they do not, we would answer “No”, and classify the revenue account and sub-components of package volume and yield as medium confidence Type 2 (refer to the table below). Despite having a lower level of confidence, we can still use historic trends, comments from management, knowledge about economic factors, and changes in competition to form a reasonable forecast for future periods. Next let’s revisit the fuel expense. Since jet fuel and gasoline prices are published daily, we are able to project one of the components in the equation, that is price, before the company reports results. In terms of the other component (gallons of fuel), we can link our projection with the historic fuel per volume shipped, which should provide a reasonable expectation of the number of gallons used in our forecast. Therefore we would answer “Yes” to this question and classify this account as a high level of confidence Type 1 account. Note that when we refer to the level of confidence under this framework, it is on a relative basis, meaning relative to the other accounts in our forecast. Realistically we cannot possibly have a high degree of confidence in any forecast measure since there are too many unknown variables. This is the nature of financial modeling in general. The relative high/medium/low confidence classifications are still useful in framing our modeling approach. Question 3) If you answered “No” to Question 1, can you determine that seasonality or business activity impacts the account? For FedEx the “other operating expense” account is a good example of an item which fits into this category. The shipping industry is seasonal in nature, and as delivery volumes peak and trough with the seasons, so do the “other operating expenses”. We can use a ratio of other operating expenses to revenue as an approximation of the seasonality impact. For this reason we can classify the “other operating expense” account as Type 3. Question 4) If you answered “No” to Question 1, and “No” to Question 3, is an average of the last few reported values appropriate? The “other income/(expense)” account includes all other corporate consolidated income and expenses which are not classified as operating expenses. This account is a “catch-all” which is difficult to project. Seasonality will not necessarily impact the future value, nor will the level of revenue or total assets. Given the difficulty and the immateriality of the account, setting the future value equal to an average of the last four reported quarters is reasonable. Therefore the “other income/(expense)” line can be classified as Type 4. If you cannot reasonably predict the account with seasonality or other factors, and there is no indication that smoothing the value over four quarters would be a more accurate projection, or if you determine that the last reported value is the most appropriate forecast, then simply set the future value equal to the last quarter. For example, if there is an other asset account with no logical modeling methodology available, it would not make sense to set the value equal to the average of the last four reported periods, since the nature of an asset balance is that the value reported in the last period would still exist in the next period, aside from any future asset acquisitions or dispositions. In this case we would classify the “other asset” balance as a Type 5 item. The table below summarizes these concepts. Keep this framework in mind as we continue building our model in the next few chapters.

47

The Gutenberg Modeling Framework Relative Level of Projection Confidence

Account Modeling Classification

High

Type 1

Medium

Low to no confidence

Low to no confidence

Low to no confidence

Type 2

Type 3

Type 4

Type 5

Description of Modeling Approach •

A metric is available which can be used to develop a reasonable future forecast, and there are reliable ways to predict the metric.

•

Use the metric in your forecast equations, and use the data available to inform your future period equation inputs.

•

A metric is available which can be used to develop a reasonable future forecast; however, there are no reliable ways to predict the metric with a high degree of certainty.

•

Use the metric in your forecast equations, and use seasonality, historic trends, and any other information to form your forecast.

•

No metric is available which can be used to develop a reasonable future forecast.

•

Use this classification when you can determine that a seasonality scaling factor is appropriate, or Balance Sheet Growth scaling factor.

•

Set the forecast equal to the last reported value plus a scaling factor, the comparable quarter value, or set the forecast to a percentage of revenue or total assets.

•

No metric is available which can be used to develop a reasonable future forecast.

•

Use this classification when you cannot determine whether or not a seasonality scaling factor is appropriate, but can determine that the average of the last few reported values is a reasonable expectation for future periods.

•

Set the forecast equal to the average of the last four quarters.

•

No metric is available which can be used to develop a reasonable future forecast.

•

Use this classification when you cannot determine whether or not a seasonality scaling factor is appropriate, and cannot determine that the average of the last few reported values is a reasonable expectation for future periods.

•

Set the forecast equal to the last reported value.

48

CHAPTER 3: BALANCE SHEET AND CASH FLOW MODELING Step 7: Complete the Financials

Step 10: Modeling Equity

Step 8: Modeling Assets

Step 11: Modeling Cash Flows

Step 9: Modeling Liabilities

Step 12: Financial Statement Links

Chapter 3 Overview: Understanding the key relationships between the three financial statements is critically important for research associates. In this chapter, we will demonstrate how to model the Balance Sheet and Cash Flow Statement, and highlight the key links between the statements. Keep in Mind: As you continue building your model, you should start to notice the direction of the equations tend to move downward in the historic periods, and upward in the forecast periods. Another way to think about this is that we tend to calculate the key ratios below the financial statements in the historic columns, then use these ratios to forecast the future results in the forecast columns. This flow of data is very apparent in some of the straightforward accounts, such as with the forecast of spare parts in Step 8c (notice in Exhibit 33 the arrows move downward in the historic columns and upward in the forecast columns). Keep this concept in mind as it will help you anticipate the method for each equation in the model as we continue with the creation of the Balance Sheet and Cash Flow Statement. Quick Note on Calibration: In the previous Chapter we noted that the Income Statement calibration (population of the blue cells) would come in Chapter 4: Model Calibration & Forecasting. Since our forecasting in Chapter 4 will be centered around earnings, the placeholders we use for the blue input cells within the Balance Sheet and Cash Flow Statement will be less likely to change, although the balances presented here will change as our earnings forecast will shift during the model calibration steps.

Step 7: Complete the Historic Financials

In this step, you will need to bring the historic Balance Sheet and Cash Flow Statements from the SEC filings into your model. This is the same procedure performed for the Income Statement in Chapter 2, Step 3. Remember to work your way backward in time, starting with the most recent filing, since the company may have made corrections after a previous release. Keep in mind that after each set of four quarters there is a column which represents the full year. Do not hardcode the values in the full year columns. Instead, insert equations which represent the fiscal fourth quarter balance for the Balance Sheet, and the sum of the individual quarters for the Cash Flow Statement. Pitfall: First-time research associates tend to have trouble with the Cash Flow Statement because most companies present the Cash Flow year-to-date, not quarter-to-date. As a result, to get the quarter cash flow for each line you will need to enter an equation to separate the cash flows for the quarter from the rest of the year, which is the year-to-date number minus the previous quarters of the year. Be sure to pay careful attention to the line descriptions, as they can move up or down the Cash Flow Statement from one quarter to the next.

49

Follow Along in the Spreadsheet: Refer to “File 5–Historic Balance Sheet and Cash Flow (Step 7)” for a version of the file with the historic results already populated. Refer to the "How to Use This Textbook" section of Chapter 1 for instructions on how to access the spreadsheet files.

Step 8: Balance Sheet Modeling—Assets

As you read through this section, consider the fundamental concepts demonstrated in this step. The company you are building a model for may not have some of the same accounts as FedEx. For example, if you are building a model for a bank or an internet company, you would not have a forecast for spare parts. Despite the difference in accounts from one company to the next, the basic modeling concept remains the same. We will be applying the Gutenberg Framework in this Chapter: 1) Start by developing an understanding of the account you are trying to project, 2) Review the historic results to determine if there are any trends or metrics which should be considered, and 3) Classify the account based on level of confidence we have in our forecast, and the approach we will use to project the account in future periods. Step 8a—Cash and Equivalents: If you open one of my completed models, the first thing you should notice in the forecast section of the Balance Sheet is that the cash balance is an equation which links to the Cash Flow Statement. This is because all of the Balance Sheet changes in the model must be paid for with cash, or financed with debt or equity, and the Cash Flow Statement is where these decisions are reflected. If we did not link the Balance Sheet to the Cash Flow Statement, then our financial statements would be out of sync, and our Balance Sheet may not balance (remember assets must always equal liabilities plus equity). For now we will skip the cash and equivalents line until after the Balance Sheet and Cash Flow Statement forecast has been setup. At which point we can link the cash line to the ending cash balance on the Cash Flow Statement. Step 8b.1—Receivables, Less Allowance (Historic Ratio): Accounts Receivable (A/R), is presented net of allowance for doubtful accounts. The receivables account is the first of the working capital items which will need to be projected. Step one is to calculate the A/R turnover ratio for the historic periods as revenue divided by average A/R. This calculation is shown in Exhibit 31 below in cell Q289, which equals revenue (Q13), divided by the average receivable balance (average of Q245 and P245). The A/R turnover ratio measures how many times the company collects on a receivable balance in each period. Next, divide the turnover ratio by the number of days in the quarter (calculated in cell Q290). The number of days represents how long it takes the company to collect its receivables. Notice that I use the average accounts receivable balance in the A/R turnover calculation. This is because the numerator, revenue, is an Income Statement account, whereas the denominator is a Balance Sheet account. Anytime a ratio mixes accounts from these two financial statements, the average of the Balance Sheet accounts should be used to reflect the fact that the Income Statement captures values over a period of time, whereas the Balance Sheet represents a distinct position at the end of the period. In the forecast periods we can set the receivables turnover ratio equal to the comparable quarter from the previous year. For example, in Exhibit 31 set cell S289 equal to N289. This will capture some of the historic seasonality trends between the quarters. Gutenberg Framework Classification (Receivables): Since the receivables forecast includes a metric which can be used to project the future balance (the turnover ratio), but there is not a reliable way to predict what the turnover ratio will be in the future, this account is classified as a Type 2 item with a medium level of projection confidence. This classification may seem counterintuitive since we set the blue input cells equal to the value from the comparable quarter, which would imply a Type 3 classification; however, we are classifying the financial statement line item (receivables) not the metric used to predict it (the turnover ratio). Modeling has a nearly infinite number of possibilities. We could come up with a metric to predict the turnover ratio, and another to predict that metric, but at some point we will reach diminishing returns, and perhaps trick ourselves into thinking we are more certain of our forecast than is actually possible. Use your judgment based on where your “balance” (refer to Key Concept 1) equilibrium point lies for your modeling purpose.

50

Chapter 3: Balance Sheet and Cash Flow Modeling

Exhibit 31—The Historic Accounts Receivable Turnover Calculation

Step 8b.2—Receivables, Less Allowance (Future Period Equations): The next step is to use the historic A/R turnover ratio to predict the future A/R balance. The only cell that needs to be populated for this forecast is the A/R turnover. The equation in the accounts receivable line of the Balance Sheet will incorporate the turnover ratio and the revenue estimate, which is a function of the assumptions entered in the Income Statement model from Chapter 2. Remember that we calculated the A/R turnover using the two-quarter average receivable balance. To reverse engineer the equation used for the forecast period, start with the historic equation and solve for the accounts receivable balance in the current quarter: A/R Equation—Sub-step 1: A/R turnover = salesQ ÷ [(A/RQ + A/RQ-1)/2]. A/R Equation—Sub-step 2: Multiply both sides of the equation by [(A/RQ + A/RQ-1)/2]. A/R Equation—Sub-step 3: Multiply both sides of the equation by 2. A/R Equation—Sub-step 4: Divide both sides by the A/R turnover ratio. A/R Equation—Sub-step 5: Subtract the A/R Q-1 from both sides of the equation. A/R Equation—Sub-step 6: We are left with: A/RQ = (2 × salesQ ÷ A/R turnover) - A/R Q-1 This equation is entered in cell S245 in Exhibit 32 below. Note that Q = the current Quarter, and Q-1 = the previous quarter.

Exhibit 32—Modeling Future Receivables

Pitfall: Seasonality is critical when determining future period A/R turnover assumptions, particularly for companies with long receivable periods in cyclical industries. For example, a retail company may experience an increase in A/R turnover in the December quarter when sales peak during the holiday season, with lower turnover for the remainder of the year. Consider what would happen if you kept the A/R turnover ratio constant from the December peak throughout the year. The accounts receivable balance forecast for the next year would decrease significantly, and the net cash per share would increase, causing an overstatement in share valuation. Given the potential impact of such assumptions, it is very important to consider the estimates made for the future working capital ratios. For FedEx the turnover tends to be relatively high in the first and second quarters, decreases in the third quarter, and increases again in the fiscal fourth quarter. To capture this effect I set the blue forecast cells equal to the last reported value for the comparable quarter. For example, in Exhibit 32 I set cell S289 equal to N289, set T289 equal to O289, and so on.

51

Step 8c—Spare Parts, Supplies and Fuel, Less Allowances: This account represents the weighted average cost of parts supplies and fuel after deducting an allowance for obsolescence. Since the account is primarily related to aircraft parts, it can be modeled it as a percentage of Property and Equipment (P&E). The percentage has remained relatively constant over time. Therefore, we do not have to incorporate any seasonality effect in the future period forecast. Instead, set the percentage of P&E equal to the average of the last four quarters. Gutenberg Framework Classification (Spare Parts): Type 2, medium level of projection confidence.

Exhibit 33—Modeling Future Spare Parts, Supplies, and Fuel

Step 8d—Prepaid Expenses and Other: This account is inherently difficult to predict because it includes the catch-all “other” classification. If you review the historic results, you will notice that there appears to be some seasonal effects in the busy November quarter; however, this effect can be overshadowed by one-off increases in random quarters. Given the low materiality of this balance relative to the total asset balance, and the difficulty in forecasting this account, I have set the future forecast equal to the average of the previous four quarters. In the future we will need to remain vigilant to identify any one-off items in this account which may require an adjustment to the forecast periods. Gutenberg Framework Classification (Prepaid Expenses): Type 4, low to no projection confidence. Step 8e—Property and Equipment: Changes in the Property and Equipment (P&E) account are driven by new purchases of capital assets and depreciation of existing assets. New P&E purchases are reported in the Cash Flow Statement as capital expenditures, or “capex”. Most companies give guidance for their expected investment in P&E (or fixed assets), which you can incorporate into your model by adjusting the capex growth rate in the ratio section below the Cash Flow Statement. Since FedEx reports accumulated depreciation on a separate line, the equation for P&E is relatively straight forward. First, take the previous reported balance of P&E (Exhibit 34, cell Q250), and add the expected cash outflows for new P&E acquisitions from the Cash Flow Statement (S322), which is based on the ratio of capex to revenue (S354). Note that FedEx operates in a capital-intensive business. Using capex as a percentage of revenue implies that it takes capital investments to produce additional revenue, which is true to some extent. The cyclical nature of revenue, mixed with the relatively long useful life of the company’s assets, means that there will be some seasonality effect in this ratio. Keep this in mind when entering ratio estimates for future quarters. After you have the ratio in place, you can set the future capex on the Cash Flow Statement equal to your projected revenue times the ratio of capex to revenue. Gutenberg Framework Classification (Property and Equipment): Type 2, medium level of projection confidence.

Exhibit 34—Modeling Property and Equipment

Step 8f—P&E Accumulated Depreciation and Amortization: Changes to depreciation should be considered along with P&E estimates. This can be difficult to quantify since depreciation of P&E and amortization of intangible assets are typically reported on the same line within the Cash Flow Statement. 52

Chapter 3: Balance Sheet and Cash Flow Modeling Depreciation and amortization are usually disaggregated within the footnotes of the annual 10-K filings, typically with details of asset classes and useful lives. For example, the details of accumulated depreciation for P&E and amortization of acquired intangibles is included in Footnote 1: “Description of Business and Summary of Significant Accounting Policies” and Footnote 4: “Goodwill and Other Intangible Assets” of the 10-K filing (refer to Exhibits 35 and 36 below).

Exhibit 35—FedEx Corp 10-K Footnote 1

Source: SEC.gov, FedEx Corp 10-K, filing date July 16, 2018, retrieved September 24, 2018.

Exhibit 36—FedEx Corp 10-K Footnote 4

Source: SEC.gov, FedEx Corp 10-K, filing date July 16, 2018, retrieved September 24, 2018.

The annual filings give the in-depth detail required to model P&E as well as depreciation and amortization, however, the quarterly filings do not contain the same level of granularity. Throughout the year, the carrying value of the assets will change with asset acquisitions, sales, and impairments. These events will change the accumulated depreciation or amortization, making it difficult to derive the related expense for an individual quarter. Given the complexity in splitting out amortization of intangibles from all other depreciation and amortization, you will need to make a decision as to whether or not this level of detail is worth the effort (again consider the Balance and Drive concept from the Introduction). In the case of FedEx, annual amortization of intangibles was only $88M of the total $3.1B of depreciation and amortization. Given the immateriality and the fact that intangible assets are not reported separately on the Balance Sheet, it is probably adequate to approximate future period depreciation based solely on a percentage of P&E. To do this first calculate the historic ratio of depreciation and amortization to P&E (refer to Exhibit 37 cell Q297). Next set the future period forecast for the ratio to the average ratio of the last four quarters (cell S297). Now you can use this ratio to calculate the implied future accumulated depreciation and amortization balance, by setting the balance for next quarter (S251), equal to the ending balance of the previous quarter (Q251), plus the depreciation forecast (S307), which is based on the average P&E balance (S250 + Q250 ÷ 2), times the ratio of depreciation and amortization to average P&E (S297). Gutenberg Framework Classification (Depreciation): Type 2, medium level of projection confidence.

53

Exhibit 37—Modeling Accumulated Depreciation and Amortization

If the company you are modeling reports intangible assets separately on the Balance Sheet, and has a more meaningful value of intangible amortization, you could use the following steps to overcome the limitations of the quarterly filings, and estimate future amortization allocated between P&E and intangibles: Depreciation & Amortization—Sub-step 1: Calculate the historic rate of depreciation and amortization expense (reported on the Cash Flow Statement) as a percentage of the total P&E and intangible assets reported on the Balance Sheet. Depreciation & Amortization—Sub-step 2: Set the future rate of depreciation and amortization expense-to-average P&E and acquired intangibles equal to the historic average of the last four quarters. Depreciation & Amortization—Sub-step 3: Set the total depreciation and amortization expense equal to the rate calculated in Step 2, times the prior period balance of P&E and acquired intangibles. Depreciation & Amortization—Sub-step 4: Input the future expected amortization of acquired intangibles based on the last 10-K filing. Note that the estimates given in the 10-K represent a full year of amortization. To get an approximate for each quarter, divide by four. Depreciation & Amortization—Sub-step 5: Take the difference between the total depreciation and amortization expense and the expected amortization. This difference will be applied to P&E. When I follow this approach in my models, I shade the amortization of acquired intangibles blue even in the historic columns. I do this to remind myself that the cell represents an estimate and the company has not disclosed the figure. The only value which is known, is the total depreciation and amortization. Note that this process is an approximation of depreciation and amortization. There are some shortcomings of the approach since it does not reflect new assets acquired with different useful lives, or impairment changes due to the disposal of specific assets. In addition, as the forecast moves further away from the last reported 10-K, the projection may drift further from the reality of the amortization estimate. Despite these limitations, the calculation should provide a reasonable estimate, and when a new 10-K is released, the estimate will be recalibrated to management’s latest forecast. Step 8g—Goodwill: In general, goodwill is not amortized over time. Instead, it is tested periodically for impairment. Given the difficulty in estimating future impairments, I set the balance equal to the prior period balance plus a growth rate scaling factor of 50 basis points per quarter, based on historic results. By applying this scaling factor, the balance will grow with a similar trend as seen in past periods. This approach assumes that the company will make similar acquisitions over time, as it has in the past. Gutenberg Framework Classification (Goodwill): Type 3, low to no projection confidence.

Exhibit 38—Modeling Goodwill

Step 8h—Other Long-Term Assets: This account is somewhat similar to the other current assets account, in that there are multiple assets reported on this line, making it difficult to project. There is one key difference. The long-term nature 54

Chapter 3: Balance Sheet and Cash Flow Modeling of the account makes it slightly less susceptible to large one-off driven fluctuations. While the balance fluctuates overtime it seems to make sense to incorporate some growth into the future balance. We can set the forecast quarters equal to the last reported value, plus a growth scaling factor based on historic results of 1.25% per quarter. Gutenberg Framework Classification (Other Long-Term Assets): Type 3, low to no projection confidence.

Exhibit 39—Modeling Other Long-Term Assets

Step 8i—Asset Modeling Overview: Now that you have completed the asset side of the Balance Sheet, take a minute to review the results of your forecast. Compare the future growth in your asset projections to historic results, in aggregate and at the individual account level. Does the forecast appear reasonable based on the historic trends? If not no need for concern just yet, as we will adjust the blue input cells in the calibration chapter.

Table 4—Summary of Asset Modeling Assumptions Line Item

Driver (Assumptions)

Future Period Equation

Cash and Equivalents

N/A

Derived in the Cash Flow Statement

Accounts receivable

Revenue and receivables turnover ratio

(Revenue ÷ receivables turnover ratio × 2) – prior quarter accounts receivable balance

Spare parts, supplies and fuel, less allowances

Ratio of spare parts-to-gross property and equipment

Property and equipment × projected ratio

Property & Equipment (P&E)

Ratio of capex-to-revenue

Previous quarter P&E + (sales × capex ratio estimate)

Accumulated depreciation and amortization

Historic ratio of depreciation-to-average P&E

Previous quarter Accumulated depreciation + (Average P&E × ratio of depreciation-to-P&E)

Prepaid expenses and other current assets

Average balance of last four quarters

Average balance of last four quarters

Goodwill

Previous quarter balance plus growth factor

Previous quarter balance plus a growth scaling factor.

Other long-term assets

Previous quarter balance plus growth factor

Previous quarter balance plus a growth scaling factor.

Step 9: Balance Sheet Modeling—Liabilities

Step 9a—Short-Term Borrowings, Current Portion of Long-Term Debt, and Long-Term Debt: Modeling a company’s future debt balance is difficult because changes in debt are driven by management’s decisions, which can be unpredictable. In addition, most companies have both fixed and floating rate debt, with varying maturities, denominated in multiple currencies, and hedges to protect against currency and interest rate fluctuations. In some cases, it may make sense to spend additional time and effort to model debt. For example, it would be more important to accurately model future debt balances for companies in the development stage, since cash burn, liquidity, and accessibility to funding would be of greater concern. On the other hand, modeling debt for a well-capitalized, developed company is not as critical. Similarly, if a company’s business is in decline, monitoring the debt balance would be extremely important as creditors could force the company into bankruptcy. Since FedEx is in a relatively favorable financial position, has substantial cash, and multiple sources of liquidity, including a shelf registration for unsecured financing, and a revolving credit facility, spending the addition effort to bifurcate and model the individual components of debt is probably unnecessary. The company is also well ahead of its required debt 55

covenants. The FedEx Balance Sheet is financed with commercial paper (short-term liabilities which mature within 90 days) and longer-term notes with varying maturity dates. The details of the debt balances are disclosed in the financial statement footnotes, so if you choose to, you can model each debt category based on interest rate and maturity. For FedEx I take a simplistic approach to modeling debt. First, I make the assumption that commercial paper will roll continuously in the future (assuming a commercial paper balance was reflected in the short-term borrowing account from the last reported quarter). To do this I set the ratio of commercial paper-to-total debt, equal to the last reported value (refer to Exhibit 40, cell S299). I also assume the current portion of debt-to-equity will equal the historic average (S300). Next, I make the assumption that the company’s principal pay downs will be replaced with new debt over time, but that the company’s debt-to-equity ratio will stay consistent with the historic average (S298, this assumption will be revisited in Chapter 4: Model Calibration & Forecasting). The debt-to-equity assumption is important because it forces the debt balance to grow with the Balance Sheet over time. To understand this concept, consider what would happen to the FedEx future Balance Sheet if the debt balance was held constant. Equity would continue to grow through retained earnings, offset to some extent by share repurchases and dividend distributions. If debt was held constant, total leverage and interest expense would decline. More importantly the Weighted Average Cost of Capital (WACC) would increase since debt would make up a smaller percentage of total capital. The required return on equity is much higher compared to the after-tax cost of debt. This would have a negative impact on share valuation through the higher weighted average cost of total capital. Using the debt-to-equity ratio to forecast future changes in debt solves this problem by allowing the debt balance to grow with the Balance Sheet. Once the debt-to-equity, commercial paper-to-total debt, and current portion of debt assumptions are input into the model, you can link the debt and commercial paper lines to the ratio assumptions, which will project the future balances (S257, S258, and S259). Note that I have added a row to track the status of the company’s revolving credit facility covenant (row 301). This covenant requires the company to maintain a ratio of debt-to-consolidated earnings (excluding non-cash pension markto-market adjustments and non-cash asset impairment charges) before interest, taxes, depreciation and amortization (“adjusted EBITDA”) of not more than 3.5 to 1.0, calculated as of the end of the applicable quarter on a rolling fourquarters basis. Gutenberg Framework Classification (Debt): Type 3, low to no projection confidence.

Exhibit 40—Modeling Debt

Step 9b—Accrued Salaries and Employee Benefits: Accrued salaries and benefits will increase with seasonality due to changes in headcount to support peak shipping periods. We can use a ratio of average accrued salaries- (Balance Sheet) to-salary expense (Income Statement). To approximate the effect of seasonality, set the ratio in the forecast periods equal to the comparable quarter from the previous year. Refer to Exhibit 41 below. Set the ratio in cell S292 equal to the comparable first fiscal quarter in cell N292. After you have set the forecast ratio assumption, you can link the Balance Sheet account (S259) to the ratio (S292), multiplied by salaries and employee benefit expense (S15), plus the previous quarter balance (Q259), divided by two. Gutenberg Framework Classification (Accrued Salaries): Type 2, medium projection confidence. 56

Chapter 3: Balance Sheet and Cash Flow Modeling

Exhibit 41—Modeling Accrued Salaries and Employee Benefits

Step 9c—Accounts Payable: This account is forecasted similar to accounts receivable, using the accounts payable turnover ratio, which is the number of times payables are collected during the period. The first step in the forecast is to calculate the historic payables turnover ratio (refer to Exhibit 42 cell Q293) as other operating expenses (Q23) divided by the average payable balance (P260 and Q260). Next, divide the turnover ratio by the number of days in the quarter. The resulting number of days represents how long the company typically takes to pay its vendors (Q294). Use the payables turnover ratio from the historic columns to predict the future accounts payable balance. The only cell which needs to be populated for this forecast is the payables turnover (S293 through V293). The equation in the Balance Sheet (S260) will incorporate the turnover ratio and the operating expense estimate (S23), which is driven by the revenue forecast, and will populate the future accounts payable balance. Similar to the receivables balance, in the forecast periods we can set the payables turnover ratio equal to the comparable quarter from the previous year. For example, in Exhibit 42 set cell S293 equal to N293. Gutenberg Framework Classification (Accounts Payable): Type 2, medium projection confidence.

Exhibit 42—Modeling Accounts Payable

Step 9d—Accrued Expenses: Accrued expense is similar in nature to accrued salaries, since the balance will change with general business activity. To approximate this effect use the historic average accrued expense to revenue ratio (refer to Exhibit 43 cell Q295). In the forecast section set the ratio (cell S295) equal to the last reported comparable quarter (cell N295). After you have set the forecast ratio assumption, you can link the Balance Sheet account to the forecast ratio (S261), by setting it equal to S13 times S295 plus Q261 divided by 2. Gutenberg Framework Classification (Accrued Expenses): Type 2, medium projection confidence.

57

Exhibit 43—Modeling Accrued Expenses

Step 9e—Deferred Tax Liability Adjustments: Deferred income taxes represent temporary differences between the tax basis of assets and liabilities and the amounts represented in the financial statements. On the Balance Sheet the net tax asset or liability is reported (liability in the case of FedEx). Management discloses their estimate of Deferred Tax Liability (DTL) in Footnote 12: “Income Taxes” within the 10-K annual filing (refer to Exhibit 44 below). When valuing FedEx shares, you may choose to value the net tax liability, separate from the remaining equity value, since this would theoretically be realized if the company were to resolve, and return the net assets to shareholders. This valuation concept was a more important issue prior to recent U.S. corporate tax reform, when the U.S. income tax rates were much higher compared to other countries. As a result U.S. companies faced significant DTL realization when repatriating overseas earnings to their U.S. parent company. The reduction in the tax rates has dramatically reduced the DTL that most companies were carrying on their Balance Sheets.

Exhibit 44—FedEx DTA & DTL 10-K Footnote

Source: SEC.gov, FedEx Corp 10-K, filing date July 16, 2018, retrieved September 24, 2018

Given that the net DTL represents a combination of tax assets and liabilities driven by multiple types of Balance Sheet accounts, with many different factors at play, the net future balance is extremely difficult to predict. As an approximation set the future balance equal to the last reported value plus a growth scaling factor of 3.8% per quarter based on historic growth. Gutenberg Framework Classification (Deferred Taxes): Type 3, low to no projection confidence. Step 9f—Pension, Postretirement Healthcare and Other Benefit Obligations: There are three primary concerns for pension plan forecasting: 1) To incorporate annual pension Mark-to-Market (MtM) adjustments in the projected earnings forecast, which FedEx calculates in the fiscal fourth quarter, 2) ensure that the funded status remains above the U.S. Internal Revenue Service (IRS) requirements of 80%, and 3) to forecast the retirement obligation on the Balance Sheet. Before diving into the details of the pension accounts, we should think about what we are generally trying to accomplish with our forecast. The company must reflect its pension retirement liability on the Balance Sheet. This liability is essentially the net funded status of the retirement and other benefit plans, which is the difference between all retirement plan assets, and all retirement plan obligations. In a perfect world we could take the simplified approach of entering the net funded status of the plans in this account and call it a day; however, it is not quite that simple for two reasons: 1) a portion of the periodic benefit related to the amortization of prior services credit is recognized through Other Comprehensive Income (OCI) in the Equity section of the Balance Sheet not in the Liability section, and 58

Chapter 3: Balance Sheet and Cash Flow Modeling 2) part of the total pension liability represents current plan obligations which are included in the accrued salaries and employee benefits account, not the non-current pension, postretirement healthcare and other benefit obligations account. Once you understand the concepts at a high level you can calculate the components of each fairly simply using the details from the remaining items in Step 9f below. Note: Refer to Chapter 4, Step 15h for important changes to the fiscal year 2019 pension accounting standards. Gutenberg Framework Classification (Pension Obligation): Type 5, little to no projection confidence. Note that we are defaulting to the lowest level of confidence in our forecasting ability of the individual pension components, based primarily on the impact of changes in mortality tables, which we have almost no confidence in our ability to forecast. Step 9f.1—Funded Status of Plans: Forecasting the funding status is more important for companies which are close to the 80% IRS threshold for plan funding, which is the fair value of plan assets as a percentage of the Projected Benefit Obligation (PBO). In 2017 the funded status for the FedEx benefit plan jumped to 88% due to a one-time increase in benefit payments driven by former employees with vested pension plans who were able to make a one-time election to receive their benefit as a lump sum. As a result, the funding status is less critical for FedEx compared to companies who are closer to the 80% threshold. To estimate the funded status, first calculate the PBO and plan assets as shown below: Changes in PBO: PBO increases with services costs, which is driven by employee wages, interest costs, and actuarial losses. To calculate the expected ending PBO balance for the forecast period (refer to Exhibit 45 cell W186), add these increases (W199) to the starting PBO (R186) and subtract benefit payments made during the year (W198). Changes in Plan Assets: First add the actual plan return (approximate as the starting plan assets in R187 × Actual Plan Return W209), to the starting plan assets (R187). Then add the contributions to the plan made during the period (W196), and subtract benefits paid (W198). Funded Status: Subtract the fair value of plan assets (W187) from the PBO (W186) to get the final funded status.

Exhibit 45—Modeling Pension Plan Funding Status

Step 9f.2—OCI Portion: Next we will use the details in Footnote 13 “Retirement Plans” from the 10-K SEC filing, to determine the portion of the pension obligation related to amortization of prior service credits which flow through Accumulated Other Comprehensive Income (AOCI).

59

FedEx has three benefit plans so we must add the gross amount from each plan $118M + $2M + $6M = $126M which we input into the model (Exhibit 46 cell R191). Next, we take the difference between the gross and net amounts to derive the tax amount, $83M + $1M + $5M - $126M = $37M (R192). For the future period forecast we can set these metrics equal to the last reported value as an approximation (W192 and W193).

Exhibit 46—FedEx Plan Impact on OCI

Source: SEC.gov, FedEx Corp 10-K, filing date July 16, 2018, retrieved September 24, 2018

Step 9f.3—Current Liability Portion: Next we can back into the implied historic portion of the liability classified as “current”, by taking the difference between the pension, postretirement healthcare and other benefit obligations account (Exhibit 46 cell R265), and the net funded status after considering the OCI portion (R191 and R192). We can set the future period forecast equal to the last reported value as an approximation (W194). Step 9f.4—Forecast of Pension, Postretirement Healthcare and Other Benefit Obligations: Now we have all the components in place to calculate the non-current portion of the pension obligation for the liability section of the Balance Sheet. Take the sum of the net funded status (refer to Exhibit 47 cell W188), minus the portion which flows through OCI (W191 and W192), minus the non-current portion (W194).

Exhibit 47—FedEx Plan Liability Forecast

Step 9f.5—10-Q vs 10-K Disclosures: The full PBO analysis is performed annually, so the 10-Q disclosures have limited information about the retirement plan. To overcome the limitation of the quarterly filings, we model the non-current liability in Step 9f.4 based on our annual forecast, and then divide the year-end result by the number of periods remaining. As the year progresses and the four quarterly metrics are reported, we can compare our annual forecast to make sure it is still reasonable, and adjust as necessary. As shown in Exhibit 47, we can project the liability for the end

60

Chapter 3: Balance Sheet and Cash Flow Modeling of the year (W265), then divide the change year-over-year by four to establish an estimate for each quarter (W265 – R265 ÷ 4. This amount is then added to each quarter S265, T265, U265, and V265). Step 9f.6a—Mark-to-Market Benefit Plan Adjustment: Now that we have the benefit plan Balance Sheet forecast, we can calculate the Mark-to-Market (MtM) adjustment, and replace the placeholder in the Income Statement from Chapter 2, Step 5f. First, breakout the primary components of the adjustment: 1) the actual versus the expected return on assets, 2) discount rate changes, and 3) demographic assumption experience, as described in the 10-K filing in Exhibit 48 below.

Exhibit 48—Primary Drivers of the MtM Adjustment

Source: SEC.gov, FedEx Corp 10-K, filing date July 16, 2018, retrieved September 24, 2018

Estimating the components of the MtM adjustment is extremely difficult; however, using a few basic inputs, we can develop a somewhat reasonable forecast, which is at least directionally correct based on changes in the underlying assumptions. We will discuss each independently in the next three steps. Keep in mind that the annuity contract FedEx purchased to retire a portion of the pension obligation was a one-time event, so we will not be projecting this item in our future forecast. Step 9f.6b—Mark-to-Market Benefit Plan Adjustment (Return): The company sets their expected return assumptions at the beginning of the year. Investment returns will vary over the course of the year, and can outperform or underperform the expected return. Management provides estimates of a 1 basis point change in expected return on the pension expense and discloses the estimate in the 10-K filing. For example, the estimated impact of 1 a basis point decrease in the actual return on plan assets, from the expected return would result in a $2.2M increase in expense.

Exhibit 49—Impact of Changes in Plan Asset Return

Source: SEC.gov, FedEx Corp 10-K, filing date July 16, 2018, retrieved September 24, 2018

If you would like to attempt to forecast the difference between the expected return on plan assets, which is disclosed in the SEC filing above at 6.75%, and the actual asset return, you could watch how equity and corporate bond returns are performing throughout the year, and multiply your expected basis point outperformance or underperformance by the approximate impact of a 1 basis point difference in return. The result would be the forecasted return impact on the MtM adjustment.

61

In Exhibit 50 below I have entered management’s impact estimate in cell W208, the expected return in cell W207, and my forecasted return of 7% in cell W209 for demonstration purposes. This results in an outperformance of 25 basis points (cell W212), which is multiplied by $2.2M to arrive at an estimated impact of a $55M decrease in pension MtM (V201).

Exhibit 50—Modeling Future Plan Asset Return

Step 9f.6c—Mark-to-Market Benefit Plan Adjustment (Discount Rate): The PBO discount rate is based on a theoretical portfolio of high grade (Aa rated or better) corporate bonds as described in the 10-K footnote below in Exhibit 51. Similar to the return on plan assets, management also discloses the expected impact on the MtM adjustment of a 1 basis point increase in the discount rate, which is estimated at $31M.

Exhibit 51—Impact of Changes in the Plan Discount Rate

Source: SEC.gov, FedEx Corp 10-K, filing date July 16, 2018, retrieved September 24, 2018

As interest rates increase, you could monitor high grade corporate bond spreads, and update the discount rate forecast based on changes in bond yields. Then, you could multiply your discount rate assumption by the impact per a 1 basis point change, to arrive at the forecasted MtM adjustment related to discount rate changes. This methodology is demonstrated below in Exhibit 52. I have entered management’s impact estimate in cell W211, and a forecasted interest rate increase of 5.56 basis points for demonstration in cell W213. This results in a $172M decrease in the MtM adjustment as shown in cell V203 ($31M per basis point × 5.56 basis points = $172M). The direction of the adjustment amount reflects the fact that as the discount rate increases (which is in the denominator of the valuation equation), the liability decreases, hence the MtM adjustment decreases.

62

Chapter 3: Balance Sheet and Cash Flow Modeling

Exhibit 52—Modeling Changes in the Plan Discount Rate

Step 9f.6d—Mark-to-Market Benefit Plan Adjustment (Demographic Assumption): The third factor in the MtM adjustment is the most difficult to project. Each year the IRS publishes new mortality tables which are incorporated into pension plan expectations. In order to properly estimate the impact of the table changes, actuaries with full knowledge of the plan participants, must work through complex actuarial models. This will not be possible in your forecast; however, you can listen for comments from management on expected directional movements, and read research from actuaries after the latest IRS tables are published, to gain knowledge of how the mortality tables may impact companies in general. In a worst-case scenario, if you cannot develop an expectation for this item, hold it flat in your forecast, as I have done in Exhibit 52 cell V204. Step 9f.6e—Mark-to-Market Benefit Plan Adjustment (Link to Income Statement): Now that the pension plan assets, liabilities, and MtM adjustment have been forecasted, we can revisit the placeholder we saved in the operating expense section of the Income Statement from Chapter 2, Step 5f. All we have to do is add the impact from each of the three MtM components in our Pension Fund Analysis section (Exhibit 53 Rows 201 through 205), and insert an equation in the Income Statement (V22) to reference the total MtM adjustment amount (V205).

Exhibit 53—Linking the MtM Adjustment to the Income Statement

FAQ 3—Do I really have to forecast the pension plan MtM Adjustment, PBO, and funding status? No you do not need to perform all of these steps. Given the assumptions we are forced to make, there may not be much value in the forecast anyway. It is important to understand the concepts, and this exercise helps to achieve that. The primary takeaways are: 1) If the company you are covering is near the plan funding limits, the pension plan should be analyzed in detail. 2) If the plan assets outperform the expected return, the MtM expense on the Income Statement will decline. If the assets underperform, the expense will increase. 3) The pension liability is very sensitive to changes in the discount rate. In a rising rate environment the pension obligation declines, and the MtM expense on the Income Statement declines. In a decreasing rate environment the pension obligation increases, and the MtM expense on the Income Statement increases.

63

4) Changes in mortality rates and other demographic measures are difficult to project, and could have a substantial impact on the MtM adjustment. 5) When analyzing the pension plan, consider how you plan on valuing the company. If you use a market multiple-based on non-GAAP earnings, and you add back the impact of the MtM adjustment as part of the non-GAAP adjustments, then changes in the pension plan will not impact your share valuation. For a stable pension plan this may be acceptable, however, if there are significant changes in the plan, you may need to reconsider your valuation methodology. Step 9g—Self-Insurance Accruals: FedEx self-insures certain risk exposures, such as delivery vehicle accidents and workers’ compensation claims. Approximately 40% of theses obligations are accounted for in the current liabilities section of the Balance Sheet. The non-current portion is included in the self-insurance accruals account. Insurance accruals tend to increase slightly during the company’s peak delivery quarters, which is logical given that there are more employees working during peak times (greater risk of worker injury), and more deliveries (greater risk of vehicle accidents). One possible solution for forecasting this account is to use revenue as an approximation for the increase in insurance claims, by taking a ratio of the average self-insurance accruals-to-total revenue. Since this ratio remains relatively stable, this should be a good metric to use in projecting the future accrual balance. By applying the ratio to our future revenue forecast, the balance will grow with an expected increase in delivery accidents and employee worker’s comp claims, if our revenue forecast increases. This method is demonstrated in Exhibit 54 below. In cell Q296 we calculate the average self-insurance accrual-to-revenue, and then use the historic average ratio (S296), and revenue forecast (S13), to project the future accrual balance (S266). Gutenberg Framework Classification (Self-Insurance Accruals): Type 2, medium level of projection confidence.

Exhibit 54—Modeling the Self-Insurance Accrual

Step 9h—All Other Liability Accounts: The remaining liability accounts, including deferred lease obligations, deferred gains, and other liabilities are inherently difficult to forecast, and make up less than five percent of total liabilities. Given the immateriality of the accounts, I use a simplistic scaling factor approach. First set the balance equal to the prior period balance plus the quarter-over-quarter growth from the previous year, which is essentially a growth scaling factor. By applying this scaling factor, the balance will grow with a similar trend seen in past periods. For example, in Exhibit 55 below we can project the other non-current liabilities account in cell S269 by setting the future balance equal to the last reported value in cell Q269, plus the growth rate from the prior fiscal first quarter from L269 to N269 ($534M × $458M ÷ $518M = $472M). Note that this approach will result in the application of the historic growth trend from the previous year, so if the Balance Sheet begins to change course, the liability forecast will need to be adjusted. This scaling factor approach is slightly different from the method used in Step 8h where we applied a historic percentage directly to the previous balance. I am using this approach for the liability accounts simply to demonstrate that there are many different methods for applying a growth scaling factors. Feel free to incorporate your preferred method into your version of the model. Gutenberg Framework Classification (All Other Liabilities): Type 3, little to no projection confidence.

64

Chapter 3: Balance Sheet and Cash Flow Modeling

Exhibit 55—Modeling the Remaining Liability Accounts

Before moving on to the equity section of the Balance Sheet, let’s take a few minutes to review the drivers of the liability forecast in Table 5 below.

Table 5—Summary of Liability Modeling Assumptions Line Item

Driver (Assumptions)

Future Period Equation

Debt-to-equity ratio, short-term debt-to-total debt ratio

Equity × debt to equity (allocated to shortterm and long-term debt)

Accrued salaries

Accrued salaries-to-revenue ratio

Average accrued salaries-to-salary expense ratio × future salary expense

Accounts payable

Operating expenses and payables turnover ratio

(Operating expenses ÷ payables turnover ratio × 2) – previous payable balance

Accrued expenses

Accrued expenses-to-revenue ratio

Average accrued expenses-to-revenue ratio × future revenue forecast

Deferred Tax Liability (DTL)

Past DTL (approximation)

Average DTL of the last four quarters

Pension, postretirement healthcare and other benefit obligations

Cash contributions, benefit payments, service costs, interest costs, actuarial loss, plan asset return, discount rate, change in demographics/ mortality tables

Refer to Step 9 for pension analysis

Self-insurance accruals

Self-insurance accruals-to-total revenue

Self-insurance accruals-to-total revenue × future revenue forecast

Deferred lease obligations

Growth scaling factor

Previous quarter balance × comparable quarter growth rate

Deferred gains

Growth scaling factor

Previous quarter balance × comparable quarter growth rate

Other non-current liabilities

Growth scaling factor

Previous quarter balance × comparable quarter growth rate

• • •

Short-term borrowings Current portion of LT debt Long-term debt

Step 9i—Revisit the Income Statement (Interest Expense): Now that we have completed the liability side of the Balance Sheet, including the forecast of debt, we can go back to the interest expense account in the Income Statement from Chapter 2, Step 5g and update the placeholder we saved for the forecast. First, calculate the historic interest expenseto-average debt ratio (refer to Exhibit 56, cell R223). Then, apply the interest expense ratio to the prior quarter debt balance (S223 applied to the sum of Q257, Q258, and Q26). Once we have the total interest expense, we must divide by four since the ratio is based on the annual interest expense, and we are solving for the quarterly value. Notice that the ratio of interest expense as a percentage of debt is based on the two-quarter average debt balance, yet in the application of the ratio for the forecast, I have only used the prior quarter debt balance. I take this shortcut to 65

avoid a circular equation reference error. The error would occur since interest expense is part of net income, which is the starting point of the Cash Flow Statement, and the latest debt balance is used in the cash flow calculation. If your debt forecast remains relatively stable over time, the shortcut for the interest expense calculation will not have a material impact on your forecast. If you are increasing or decreasing the debt balance significantly in your forecast, then you may want to consider an adjustment to the interest expense ratio (Exhibit 56 cells S223 through V223) to reflect the impact of the change in debt balance and the shortcut approach used for the average debt. Gutenberg Framework Classification (Interest Expense): Type 2, medium level of projection confidence.

Exhibit 56—Modeling Interest Expense

Step 10: Balance Sheet Modeling—Equity

At the start of this program we discussed modeling the “three primary financial statements,” the Income Statement, Balance Sheet, and Cash Flow Statement. The reality of our forecast is that we are actually modeling four statements, including the Statement of Changes in Shareholder Equity, which we model in a summarized format through the equity section of the Balance Sheet. Before we begin refer to the Consolidated Statement of Changes in Stockholder’s Investment in Exhibit 57 below. In this statement you can clearly see the drivers of each of the components of equity which we will be modeling in this step.

Exhibit 57—FedEx Changes in Equity

Source: SEC.gov, FedEx Corp 10-K, filing date July 16, 2018, retrieved September 24, 2018.

Step 10a—Common Stock: If a company reports common stock net of Additional Paid in Capital (APIC) on the Balance Sheet, the balance will fluctuate over time. Since FedEx breaks out common stock from APIC on separate Balance Sheet lines, the common stock account reflects only the par value of the stock issued by the company. If you are not forecasting an issuance of additional equity in the future, you can simply set this account equal to the last reported value. As you can see in the common stock column of Exhibit 57 above, this forecast is consistent with the historic reporting of the last three years, as the balance of common stock has remained constant. Gutenberg Framework Classification (Common Stock): Type 2, medium level of projection confidence. 66

Chapter 3: Balance Sheet and Cash Flow Modeling Step 10b—Additional Paid in Capital: The APIC account represents the equity capital value in excess of the par value. APIC is impacted by stock-based compensation, as show in column two of Exhibit 57 above. The first step in setting up the forecast equations, is to model the expected stock-based compensation expense. Since this account shows up in the Cash Flow Statement, we can jump ahead to setup a placeholder in this section of our model (refer to Exhibit 58, row 309). We model the stock-based compensation expense as a percentage of revenue, and use the historic average ratio (S352) along with our revenue forecast (S13), to project the next quarter (S309). Now, we add our stock-based compensation forecast (S309) to the last reported value of APIC (Q275), which will result in our final projected APIC balance (S275). Gutenberg Framework Classification (APIC): Type 2, medium level of projection confidence.

Exhibit 58—Modeling Stock-Based Compensation and APIC

Step 10c.1—Retained Earnings: This account represents the net accumulated earnings over time which have not been distributed to shareholders. The future period calculation of retained earnings is equal to the beginning balance (refer to Exhibit 59, cell Q276), plus net income for the current period (from either the Income Statement in cell S37 or the Cash Flow Statement in cell S306), minus dividends paid (S331). Note that in the forecast equation the cells referenced appear to be adding back dividends. This is due to the fact that the cells come from the Cash Flow Statement, where dividends are reported as negative values to reflect the fact that the amounts represent cash outflows. As a result the equation is “adding” negative values; therefore, reducing retained earnings as the dividends represent distributions of earnings to shareholders. This approach is consistent with column three of Exhibit 57. Gutenberg Framework Classification (Retained Earnings): Type 2, medium level of projection confidence.

Exhibit 59—Modeling Retained Earnings

Step 10c.2—Dividends: In the retained earnings step above we used the dividend forecast which was covered in Step 5k at the end of the Income Statement section. In Step 5k we introduced the concept of projecting the dividend growth rate based on the historic dividend payments. When we get to the Cash Flow Statement section in Step 11, we will simply multiply the estimated dividend per share from Step 5k (refer to Exhibit 60 below, cell S45), by the number of

67

shares outstanding (S44) to forecast the dividend payment reported in the Cash Flow Statement (S331), which we use to reduce the retained earnings balance in Step 10.c.1. FedEx is in a relatively strong overall financial position. There is no reason to believe the company’s dividend will be cut significantly in the near-term; however, the future dividend policy may be something to consider if you are modeling a company which is facing challenges. If you disagree with my assessment of FedEx, you can enter your dividend forecast in row 46. Gutenberg Framework Classification (Dividends): Type 2, medium level of projection confidence.

Exhibit 60—Modeling Dividends

Step 10d—Accumulated Other Comprehensive Income (AOCI): For most companies, changes in AOCI are typically driven by currency translation, retirement plan adjustments (related to amortization of prior service costs), changes in fair value of derivative contracts classified as cash flow hedges, and unrealized gains and losses on Available For Sale (AFS) securities. FedEx includes two of these categories in the AOCI balance: foreign currency translation, and retirement plan adjustments (refer to the footnote disclosure in Exhibit 61 below). Gutenberg Framework Classification (AOCI): Type 3, little to no projection confidence.

Exhibit 61—FedEx AOCI Footnote

Source: SEC.gov, FedEx Corp 10-K, filing date July 16, 2018, retrieved September 24, 2018.

The effect of foreign currency translation may be easy to consider from a qualitative standpoint. Meaning when the U.S. dollar appreciates relative to foreign currencies we would expect the AOCI impact for FedEx to be negative, and when the U.S. dollar depreciates we would expect the AOCI impact to be positive; however, quantifying the translation projection for future periods is extremely difficult. I take a simplistic shortcut and set the value equal to the last reported value. When deciding how you would like to approach the forecast of AOCI consider the Balance and Drive concept from Chapter 1, and where your analytical equilibrium balance point lies on the spectrum. 68

Chapter 3: Balance Sheet and Cash Flow Modeling In our pension analysis section we covered the portion of the pension exposure recognized through OCI. To forecast the future AOCI balance, add the amount of the pension exposure recognized through OCI net of tax (refer to Exhibit 62 below cell S192), to the beginning balance (R279).

Exhibit 62—Modeling AOCI

Step 10e—Treasury Stock: This account reflects the value of common stock which has been repurchased by the company. The forecast of share repurchases was covered in Chapter 2, Step 5i. For the equity section of the Balance Sheet, add the forecasted repurchase amount (Exhibit 63 cell S228), to the prior period balance of treasury shares (Q282). Gutenberg Framework Classification (Treasury Stock): Type 2, medium level of projection confidence.

Exhibit 63—Modeling Treasury Stock

This completes the equity section of the Balance Sheet. To calculate the total shareholders’ equity balance (Exhibit 63, cell S283), add common stock (S274), APIC (S275), retained earnings (S276), AOCI (S281), and treasury stock (S282). Step 10f—Off-Balance Sheet Commitments: Some companies have substantial obligations reported off-Balance Sheet, which you may choose to bring onto the Balance Sheet using top-side adjustments in your analysis. For FedEx the offBalance Sheet exposure is primarily related to aircraft purchase obligations. One of the fundamental assumptions we make in our attempt to model future financial statements, is the “going concern” notion. This means we expect the company to continue to operate as an entity for the foreseeable future. Given this point, we can assume that FedEx 69

will purchase and utilize the aircraft and equipment for which it has made commitments. As these contracts materialize in the future, the financial statements will reflect the associated costs, assets, and liabilities of the transactions. Therefore, there is no need to adjust the financial statements for these off-Balance Sheet exposures at this time.

Exhibit 64—FedEx Off-Balance Sheet Consideration

Source: SEC.gov, FedEx Corp 10-K, filing date July 16, 2018, retrieved September 24, 2018.

70

Chapter 3: Balance Sheet and Cash Flow Modeling

Table 6—Summary of Equity Modeling Assumptions Line Item

Driver (Assumptions)

Future Period Equation

Common Stock

N/A

Prior quarter balance

Additional Paid in Capital (APIC)

Stock-based compensation

Prior APIC + Stock-based compensation

Retained Earnings

Dividends and net income

Prior quarter balance + net incomes – dividend distributions

Accumulated Other Comprehensive Income (AOCI)

Foreign currency translation, portion of pension expense recognized in OCI

Prior AOCI balance + pension expense recognized through OCI net of tax

Treasury stock

Share repurchases

Prior balance + share repurchase amount

Follow Along in the Spreadsheet: Refer to “File 6–Balance Sheet Forecast Equations (Through Step 10)” for details on what your model should look like after completing the Balance Sheet section of the template. Refer to the "How to Use This Textbook" section of Chapter 1 for instructions on how to access the spreadsheet files.

Step 11: Cash Flow Statement Modeling

Cash Flow Statement modeling is critical for companies facing liquidity issues. Although the FedEx capital expenditures have increased over time, the company’s positive operating cash flow will likely support growth in the future. For companies in the development stage, cash-burn can be a substantial issue. In this case, using a full financial statement model, including the Cash Flow Statement, can be useful to determine if the company will require additional debt or equity to fund future growth. In our FedEx model, the main reason for forecasting the Cash Flow Statement is to arrive at a Free Cash Flow (FCF) estimate to use in a Discounted Cash Flow (DCF) valuation calculation. The Cash Flow Statement is relatively easy to model, since it essentially represents estimates already included in the Income Statement and Balance Sheet. The following steps show the references where each of the Cash Flow Statement lines are calculated. Step 11a—Net Income: The Cash Flow Statement begins with the net income from the Income Statement. Refer to Exhibit 65 below for details. For the future period estimates, set the cells in row 306 equal to row 37. Gutenberg Framework Classification (Net Income): Net income flows to the Cash Flow Statement from the Income Statement, and includes all the assumptions made in Chapter 2. As a result, our projection confidence lies somewhere between our highest and lowest level of confidence for the Income Statement assumptions. There is no additional benefit in classifying net income separately in our framework at the Cash Flow Statement stage. Instead this serves as a reminder of how limited our overall ability to forecast is.

Exhibit 65—Enter Equations for Net Income in the Cash Flow Statement

71

Step 11b—Depreciation and Amortization: The calculation for depreciation and amortization is covered in Step 8f as part of the calculation of Property and Equipment (P&E). The ratio of depreciation and amortization to average P&E (Exhibit 66, cell S297) is applied to the average P&E balance S250 + Q250 ÷ 2. If you have not already done so, you can now link the depreciation expense line from the Income Statement to the depreciation line in the Cash Flow Statement. Gutenberg Framework Classification (Depreciation and Amortization): Type 2, medium level of projection confidence.

Exhibit 66—Enter Equations for Depreciation in the Cash Flow Statement

Step 11c—Provision for Doubtful Accounts: Recall from Step 8b.1 that the receivables balance is reported net of the allowance for doubtful accounts, which means that the method of calculating the receivables turnover includes the impact of changes in the allowance. When it becomes unlikely that a receivable balance will be collected, the company will book a provision for uncollectable accounts, which increases the allowance. The provision is reflected in the Cash Flow Statement since it is a non-cash charge which must be added back to net income. To approximate the provision charge in the forecast section of our Cash Flow Statement model, we can apply a ratio of the provision to the receivable balance (Exhibit 67 cell S351), and use the historic average of this ratio to estimate the future provision charges by multiplying the ratio and previous quarter’s accounts receivable. Note that since the receivables balance is reported net of the allowance (Q245), we must first add the previous quarter’s provision (Q308), to get the gross receivable balance, before applying the ratio for the future forecast. Gutenberg Framework Classification (Provision): Type 3, little to no projection confidence.

Exhibit 67—Enter Equations for Provisions in the Cash Flow Statement

Step 11d—Share-Based Compensation: The forecast for Share-Based Compensation (SBC) was discussed with Additional Paid In Capital (APIC) in Step 10b. The ratio of share-based compensation to revenue (Exhibit 68, cell Q352) is applied to the revenue forecast (S13) for the projection of share-based compensation (S309). Gutenberg Framework Classification (Share-Based Compensation): Type 2, medium level of projection confidence.

72

Chapter 3: Balance Sheet and Cash Flow Modeling

Exhibit 68—Enter Equations for SBC in the Cash Flow Statement

Step 11e—Deferred Income Taxes: We forecasted the Deferred Tax Liability (DTL) reported on the Balance Sheet in Step 9e. In the Cash Flow Statement enter the change in the Balance Sheet account (Exhibit 69, cell S264 minus Q264). Gutenberg Framework Classification (Deferred Income Taxes): Type 3, little to no projection confidence.

Exhibit 69—Enter Equations for Deferred Tax in the Cash Flow Statement

Step 11f—Retirement plan MtM Adjustment: The MtM adjustment is booked in the fiscal fourth quarter. This amount was calculated in Step 9f.6e. Insert an equation to reference the adjustment in the Cash Flow Statement (Exhibit 70 set row 312 = row 205). Gutenberg Framework Classification (Benefit Plan MtM): Type 5, little to no projection confidence. Note that we are defaulting to the lowest level of confidence in our forecasting ability of the individual pension components, based primarily on the impact of changes in mortality tables, which we have nearly no confidence in our ability to forecast.

Exhibit 70—Enter Equations for Retirement MtM in the Cash Flow Statement

Step 11g—Impairments or Gains From Sale of Investments: If you have included any impairments or gains in your forecast, they should be included in the cash flows from operating activities section of the Cash Flow Statement (discussed in Step 5e). As shown in Exhibit 71 below, I have not entered a forecast for these rows. Gutenberg Framework Classification (Impairments/Gains on Sales): Type 3, little to no projection confidence.

Exhibit 71—Enter Impairments and Gains in the Cash Flow Statement

73

Step 11h—Changes in Operating Assets and Liabilities: This section calculates the changes from one quarter to the next, for all of the company’s assets and liabilities, other than those classified as investing or financing activities. For example, the difference between the fiscal first quarter forecasted accounts payable balance and the fiscal fourth quarter (Exhibit 72, cell S260 minus Q260), equals the accounts payable cash flow in the operating activities section of the Cash Flow Statement (S318). Notice that since the first fiscal quarter is greater than the fourth quarter, the cash flow is positive, which indicates that the increase in the liability represents a cash inflow. The majority of the lines on the Balance Sheet have a corresponding line on the Cash Flow Statement. If there is not a specific line on the Cash Flow Statement, then add the amount to one of the “other” categories. Gutenberg Framework Classification (Operating Assets and Liabilities): Type 2, medium level of projection confidence.

Exhibit 72—Enter Operating Assets and Liabilities in the Cash Flow Statement

Step 11i—Changes in Investing Activities: This section of the Cash Flow Statement shows the changes in cash due to capital expenditures (capex) which was discussed in Step 8f. In this example, the ratio of capex-to-revenue was used with the forecast of revenue, to project future capex (refer to Exhibit 73, cell S322). This capex forecast is also used to project the property and equipment account on the Balance Sheet. Goodwill is also included in the Investing Activities section of the Cash Flow Statement. We can use our projected change in goodwill from Step 8g to approximate the cash flow related to dispositions and other investing activities (S253 - Q253 = S324). Note that this is merely an approximation. Goodwill may decrease due to impairments, not just dispositions. In addition, many other items which cannot be projected flow through this line on the Cash Flow Statement. Gutenberg Framework Classification (Investing Activities): Type 2, medium level of projection confidence.

74

Chapter 3: Balance Sheet and Cash Flow Modeling

Exhibit 73—Investing Activities Section of the Cash Flow Statement

Step 11j—Changes in Financing Activities: The financing section of the Cash Flow Statement shows inflows and outflows of cash to fund changes in the Balance Sheet, primarily related to debt and equity issuance. This section includes the following items: 1) Change in Debt: Short-term borrowings and long-term debt were forecasted in Step 9a. The cash inflow or outflow related to these accounts is calculated as the difference from one quarter to the next. For example, the proceeds from issuance of debt in Exhibit 74, cell S329 is calculated as the fiscal first quarter projected debt balance ($16.43B, cell S263), minus the balance from the previous quarter ($15.24B, cell Q263), which equals $1.19B (cell S329). 2) Proceeds From Issuance of Common Stock: In my model, I have not included estimates for future equity issuance (Exhibit 74, cell S330) since I believe it is unlikely that FedEx will need to raise funds through common stock. Keep in mind that even if a company is not raising funds through an equity offering, there can be some immaterial fluctuations, usually driven by the impact of dilutive preferred securities or stock option transactions. If you are modeling a company which will need to raise an equity offering in the future based on your analysis, you should include the expected proceeds in the Cash Flow Statement, adjust the equity on the Balance Sheet, and increase the number of shares used in the calculation of EPS on the Income Statement. 3) Dividend Payments: Dividends are represented on the Balance Sheet as reductions in retained earnings (refer to Step 10c.2). The calculation for dividend payments in the Cash Flow Statement is based on the number of shares outstanding (Exhibit 74, cell S40), times the dividend amount per share (S45). 4) Share Repurchases: Repurchases were calculated in Chapter 2, Step 5i. The total dollar amount of repurchases (Exhibit 74, cell S228) are represented in the financing activities section of the Cash Flow Statement (S332). Gutenberg Framework Classification (Financing Activities): Type 2, medium level of projection confidence.

75

Exhibit 74—Financing Activities Section of the Cash Flow Statement

Step 11k—Calculate the Final Cash and Equivalents Balance: Now that we have completed the Cash Flow Statement, we can add the cash from operating, investing, and financing activities to get the net change in the cash and equivalents balance (refer to Exhibit 75, cell S337). Recall from our forecast of AOCI in Step 10d, we did not project the impact of currency translation, so we leave row 336 blank. Next, add the change in cash to the beginning balance. This is the final forecasted balance of cash and equivalents. Link row 339 from the Cash Flow Statement to row 244 on the Balance Sheet. Note: If the company you are covering has restricted cash, you may see a difference between the ending cash on the Cash Flow Statement and the cash shown on the Balance Sheet. This is a result of the issuance of Accounting Standard Update (ASU) Number 2016-18, which requires companies to include restricted cash in the Cash Flow Statement. If this is the case, simply add a row at the bottom of your Cash Flow Statement to carve out changes in restricted cash, and reconcile the remaining cash balance to the Balance Sheet.

Exhibit 75—Calculate Projected Cash and Equivalents Balance

Step 11l—Revisit the Interest Income Projection in the Income Statement: With the completed forecast for the cash and equivalents balance, we can now return to the other income account from Chapter 2, Step 5g, and calculate the interest income as shown in Exhibit 76 below. Your instinct will probably tell you to set the future period equation equal to the interest income as a percentage of the average cash and equivalents estimate in cell S244 and Q244. Unfortunately, if you setup the equation this way, it will cause a circular reference error in the spreadsheet due to the links between the three financial statements.

76

Chapter 3: Balance Sheet and Cash Flow Modeling Instead of using the average balance for the two quarters, apply the ratio to the last reported value. This shortcut is appropriate given the immateriality of the interest income for FedEx. If you are modeling a company whose interest income makes up a larger percentage of overall earnings, or the Balance Sheet changes significantly in the future, you may need to revisit this simplified approach. Gutenberg Framework Classification (Interest Income): Type 2, medium level of projection confidence.

Exhibit 76—Modeling Investment Income

Follow Along in the Spreadsheet: Refer to “File 7–Cash Flow Forecast Equations (Step 12)” for details on what your model should look like after completing the Cash Flow Statement section of the template. Refer to the "How to Use This Textbook" section of Chapter 1 for instructions on how to access the spreadsheet files.

Step 12: Primary Financial Statement Links

After the Income Statement, Balance Sheet, and Cash Flow Statement are complete, take a minute to review how the three statements are interconnected. The following is a recap of some of the primary links between the various accounts. The cell references correspond to the model example at the end of this section. Earnings: GAAP net income (cell S37) flows through retained earnings in the Balance Sheet (cell S282 minus Q282, minus), after removing dividend distributions (cell S331). Share repurchases reduce the share count used to calculate net income (cells S40 and S41), are reflected in the equity section of the Balance Sheet (cell S282), and flow through the cash flows from financing activities in the Cash Flow Statement (cell S332). Income Statement Links to Balance Sheet: Depreciation expense (cell 18) flows through the net property and equipment account on the Balance Sheet (S251 minus Q251). Interest expense (cell S29), is calculated based on the debt accounts from the Balance Sheet (cells S257, S258, and S257). Income Statement Links to the Cash Flow Statement: GAAP net income (cell S37) represents the starting point for the Cash Flow Statement (cell S306). This means that all of the assumptions made to the future earnings forecast will impact the projected cash flows. Since depreciation expense (cell S18) is a non-cash charge, it is added back to net income on the Cash Flow Statement (S307). Balance Sheet Links to the Cash Flow Statement: All of the changes in the Balance Sheet accounts are reflected throughout the Cash Flow Statement. For example, in F1Q2019E the accounts payable from the Balance Sheet increase by $245M (cell S260 minus Q260). This increase in the liability account will produce a cash inflow, which is reflected in the Cash Flow Statement (cell S318). Cash Flow Links to the Balance Sheet: The Cash Flow Statement will add all the inflows and outflows of cash (cell S337) to the beginning balance, resulting in the ending cash balance (cell S339), which flows to the Balance Sheet (cell S244). This is the final link which ensures that the assets reported on the Balance Sheet equal the liabilities plus equity.

77

Primary Financial Statement Links

78

CHAPTER 3 WRAP-UP Takeaways • •

• •

•

•

•

•

Modeling the future period cash balance should be performed within the Cash Flow Statement, and then linked to the Balance Sheet, not directly forecasted in the Balance Sheet. Many asset line items have associated contra-asset balances which must be considered in the modeling process. Some are reported on a gross basis, while others are presented on a net basis. For example, Property & Equipment (asset) is typically shown gross of accumulated depreciation (contra-asset). On the other hand, accounts receivable (asset) is typically presented net of the allowance for doubtful accounts (contra-asset). Depending on the purpose of your model you may choose to forecast each line separately, or you may take a simplistic approach and model each line on a net basis. The forecasted Balance Sheet must balance (total assets must equal liabilities plus shareholders’ equity). Equity includes common stock, Additional Paid in Capital (APIC), retained earnings, treasury shares, and Accumulated Other Comprehensive Income (AOCI). o APIC represents equity capital in excess of par value, and is impacted by stock-based compensation. o AOCI includes currency translation, retirement plan amortization of prior period service costs, changes in fair value of derivatives contracts classified as cash flow hedges, and unrealized gains and losses of Available For Sale (AFS) securities. Common links between the Income Statement and Balance Sheet include: o Depreciation expense is linked to fixed assets. o Net income flows through retained earnings in equity. o Changes in treasury shares from the equity section of the Balance Sheet reduce the share count on the Income Statement. Common links between the Income Statement and the Cash Flow Statement include: o Net Income represents the starting point of the Cash Flow Statement. o Depreciation is removed from the Cash Flow Statement since it is a non-cash charge. Common links between the Balance Sheet and the Cash Flow Statement include: o Changes in Balance Sheet accounts are reflected in the Cash Flow Statement. o Cash flows for purchases/(sales) of fixed assets will increase/(decrease) the fixed asset (or P&E) account on the Balance Sheet. o Cash outflows for dividends paid will decrease the retained earnings balance in the equity section of the Balance Sheet. Key Concept 3—An Approach to Model Any Account, for Any Company: There are many different approaches to creating a financial model. The Gutenberg Modeling Framework provides a basic blueprint of tasks to perform during the modeling process including: o Develop an understanding of the accounts you are modeling. o Review the historic results. o Classify each account based on the future period modeling logic using the following questions to guide the classification:  Is there a metric available which can be used to develop a reasonable future forecast?  Is there any data available before the company reports results which would give a high degree of comfort in the forecast?  Does seasonality or business activity impact the account?  If no reasonable forecast approach exists for the account, would an average of the last few historic periods, or the last reported value be appropriate? 79

Concept Quiz Instructions: Answer each of the following questions as “true” or “false” 1) Modeling assets is much easier then modeling liabilities. 2) All assumptions used in modeling the Income Statement will have an impact on the Cash Flow Statement. 3) If an equity research team has the resources available, each financial statement for a particular company should be modeled by separate analysts. 4) Changes in inventory balances are reflected on the Income Statement. Instructions: Select the best answer for each of the following questions. 5) If a company’s accounts receivable balance increases quarter-over-quarter, what is the most likely impact on the Cash Flow Statement? A. The cash flow from operations will decrease. B. The cash flow from operations will increase. C. There is not enough information to determine the correct answer. 6) If a company’s accounts payable balance increases quarter-over-quarter, what is the most likely impact on the Cash Flow Statement? A. The cash flow from investments will increase. B. The cash flow from operations will increase. C. There is not enough information to determine the correct answer. 7) ABC Company’s revenue has increased this quarter, with no change in operating margin or effective tax rate. What is the most likely impact on cash flow? A. Cash flow from financing activities has likely decreased. B. Cash flow from investing activities has likely increased. C. Cash flow from operations has likely increased. 8) Jack purchases 100 pounds of coffee from the Washington Coffee Wholesale Corporation (WCWC). What is the likely impact of this transaction on WCWC’s financial statements? A. Revenue has increased on the Income Statement, Inventory has decreased on the Balance Sheet, and Cash Flow from operations has increased. B. Revenue and cost of goods sold have increased on the Income Statement, Inventory has decreased on the Balance Sheet, and Cash Flow has not changed. C. Revenue and cost of goods sold have increased on the Income Statement, there is no impact on the Balance Sheet, and cash flow has increased. 9) Acme Automotive Group Incorporated (AAG) repurchased common shares on a public exchange. What was the likely impact of this transaction on AAG’s financial statements? A. Cash flow from operations has increased, diluted share count has decreased, Earnings Per Share (EPS) has increased, and equity has increased on the Balance Sheet. B. Revenue has increased, Cash flow from financing activities has decreased, diluted share count has decreased, Earnings Per Share (EPS) has increased, and equity has increased on the Balance Sheet. C. Cash flow from financing activities has decreased, diluted share count has decreased, Earnings Per Share (EPS) has increased, and equity has increased on the Balance Sheet. 10) Alex Smith is modeling the future accounts receivables balance for Social Network Media Corporation. What is the most reasonable approach for Alex’ forecast?

80

Chapter 3 Wrap-Up A. A forecast based on the future revenue estimate and a ratio of receivables to revenue. B. A forecast based on the future forecasted cash balance and a ratio of historic cash to receivables. C. A forecast based on historic cash flow from operations.

Concept Quiz Answers 1) False. The process for modeling Assets and Liabilities is very similar. 2) True. Not all Income Statement assumptions will have a cash flow impact, for example depreciation is a noncash charge; however, since net income is the starting point for the Cash Flow Statement, all assumptions entered into the Income Statement model will impact the Cash Flow Statement. 3) False. Financial statements are interconnected. Therefore, modeling each financial statement in a vacuum by separate analysts would not be a practical approach. 4) False. Changes in operating assets, such as inventory balances, are presented in the Cash Flow Statement. Note: If your answer to this question considered the fact that changes in inventory balances will impact cost of goods sold which is shown on the Income Statement, than your logic is correct and the answer for this question would be “true”. 5) A. Accounts receivable is an asset. An increase in an asset represents a cash outflow which will decrease cash, and must be funded through a liability or equity. 6) B. Accounts payable is a liability. An increase in a liability represents a source of cash which will increase the cash from operations. Note the distinction between cash from operations and financing or investment activities is critical for the calculation of free cash flow, which is the basis of many valuation methodologies. 7) C. Under these circumstances cash flow from operations has likely increased. 8) A. Revenue has increased on the Income Statement, inventory has decreased on the Balance Sheet, and cash flow from operations has increased. 9) C. This represents a use of cash, therefore, cash flow from financing activities has decreased, diluted share count has decreased, Earnings Per Share (EPS) has increased with the decrease in share count, and equity has increased on the Balance Sheet with the increase in treasury shares. 10) A. The forecast based on the future revenue estimate and a ratio of receivables to revenue. The concept of drive is the key to this question. Revenue is the primary driver of accounts receivable, so the most appropriate modeling technique should incorporate revenue.

81

CHAPTER 4: MODEL CALIBRATION & FORECASTING Step 13: Incorporate Historic Trends

Step 16: Consider Management’s Guidance

Step 14: Adjust for Seasonality

Step 17: Review the Consensus Estimates

Step 15: Adjust for Changes in Circumstances

Step 18: Incorporate Opinions and Monitor

Chapter 4 Overview: In the last chapter we introduced the modeling framework to decide on the metrics we would use to create equations for the forecast of each account. Now we must determine what values to use for the inputs in the equations (the blue cells in our model). This chapter will cover how to develop a forecast for your model, and incorporate future period estimates for the assumption.

Step 13: Incorporate Historic Trends

The historic trends should be considered first, prior to incorporating management’s guidance, or any other factors. This will prevent us from developing a bias in our analysis. Or at least to some extent, other than our general biases about the company, which have been developed through past experience. By extending the historic trends into the future, you will force yourself to question deviations from recent history. If you skip this step, and jump directly to developing your own forecast, you may have a difficult time recognizing changes from historic trends in your results. As you perform this step, remember that history does not always repeat itself. There could be very good reasons for why future results differ from historic trends. If this is the case, you should be prepared to articulate the reasons. Starting with an analysis of historic trends will help you accomplish this. In this section, I will demonstrate how to incorporate historic trends into the model using the FedEx U.S. Overnight Box sub-segment as an example. As discussed in Chapter 2, Step 4, we use forecasts of year-over-year changes in Average Daily Volume (ADV) and yield (revenue per package) to forecast the revenue for this segment. Let’s start by reviewing the historic trend for ADV. In Exhibit 77 there are three primary observations to take away from the historic results: 1) Overall the ADV for the U.S. Overnight Box sub-segment has been decreasing slightly over time. This point is apparent if you sum the four quarters of each year (refer to cells M74 and R74). 2) In the last reported quarter, the slightly decreasing ADV was interrupted by the strongest year-over-year growth rate seen in over three years (refer to cell Q74). 3) ADV tends to peak in the fiscal third quarter which includes the holiday shipping season (refer to cells F73, K73, and P73). We will discuss the effect of seasonality in the next step.

82

Chapter 4: Model Calibration & Forecasting

Exhibit 77—Historic Trend for the U.S. Overnight Box Segment

Without applying any judgment about competitive pressure, general business position, or macroeconomic environment, we have a few options for our forecast of the ADV year-over-year growth rate. We could set it to zero in all quarters, but this would not incorporate the slight downward trend from the recent historic results. We could set it to the last reported growth rate, but this would over estimate the benefit of the performance from the last quarter. Instead let’s set the growth rate to the moving average of the last four quarters.

Exhibit 78—Setting the Forecast for U.S. Overnight Box ADV

The impact of this moving average is that the first quarter in our forecast experiences a decline (average of cells N74, O74, P74, and Q74 is negative one percent) since it picks up the trough growth rate of the fiscal first quarter in 2018 (cell N74). After the first quarter, the growth rate stabilizes and ends 2019 with a forecasted increase of 0.27%. Given the strength in the last reported quarter, offset by the downward trend overtime, this is probably a reasonable result. After the next fiscal quarter is released, if the strength seen in the fiscal fourth quarter of 2018 does not continue, then the average growth rate approach will result in the growth projection returning to the slightly downward trend over time.

Step 14: Adjust for Seasonality

Since we used year-over-year growth rates, the forecasting approach from Step 13 will automatically include the effect of seasonality, assuming the previous year did not have any non-recurring items included in the reported results. To demonstrate this point consider what would happen in Exhibit 78 if we applied a 1% growth rate to each of the forecast quarters in cells S74, T74, U74, and V74. For the F1Q2019E forecast, the 1% growth would be applied to the F1Q2018 ADV. Similarly the F2Q2019E, F3Q2019E, and F4Q2019E forecasts would be based on applying the growth rate to F2Q2018, F3Q2018, and F4Q2018 respectively. As a result the fiscal third quarter peak would be correctly reflected in our forecast since the comparable fiscal third quarter of the previous year was the basis of the forecast. After we have entered our forecast, we should review the general trend of the projections, and compare it to the past few years to confirm that the seasonal trend is reasonable. If not, it could be an indication of an error in our forecast, or evidence that the previous year included a one-off event which may require an adjustment to the future period assumptions.

Step 15: Adjust Trends for Changes in Circumstances

Many different factors could arise which may cause future results to breakaway from the historic trends. Some examples include: changes in strategic direction, new product launches, acquisition of a new business, disposition of an old business, changes in accounting policies, changes in competitive pressure, changes in input prices, or macroeconomic events. If any such changes occur, then using historic trends and seasonality alone would not produce an accurate forecast of future results. These events typically fall into one of three categories: 1) “Big bang” events that instantly change the company’s earnings profile indefinitely, such as the announcement of a divestiture or acquisition.

83

2) Gradual structural changes which continue for long periods of time, such as a shift in sales from one region to another, or a change in the type of business booked over time (i.e. shift from service-based to product-based). 3) Non-recurring items such as legal settlements or large one-time orders, which could impact year-over-year comparisons for the period in which they are booked. If the company you are modeling is impacted by one, or more of these events, then consider the ramifications in your future projections. To find changes which may require forecast adjustments, read the last few earnings transcripts, the MD&A section of the 10-K, details of accounting changes in the 10-K, any investor presentations the company has given during the year, competitor disclosures, and news articles. As you research the company you will begin to form your opinions about the company’s future prospects. Focus your efforts on finding the material items which could have an impact on the company’s financial statement forecast. To demonstrate this point I will highlight a number of items from various sources, and show how to adjust the model to incorporate each in the FedEx forecast (Steps 15a through 15k). We will start with a relatively straightforward pricing example (Step 15a). After that we will go through a number of data points found during the research process, which impact various components of the FedEx model. Step 15a—Pricing Forecast Example: Exhibits 79 and 80 are excerpts from the Business Segment section of the 10-K which describe changes in the pricing (yield) environment for FedEx. In the case of the Express Segment (Exhibit 79) there were pricing increases effective January 1, 2018, therefore, the last two reported quarters would likely be more representative of the correct future pricing for this segment compared to any other historic period. Using the historic yield prior to these periods to inform the forecast of future results, would likely lead to an underestimation of the yield in the projection period. In addition to explicit pricing changes you may also want to consider how changes in shipping trends will impact price and volume. For example, an increase in e-commerce shipping could increase the volume and decrease the yield overtime.

Exhibit 79—Description of Pricing Changes (Express Segment)

Source: SEC.gov, FedEx Corp 10-K, filing date July 16, 2018, retrieved September 24, 2018.

The Ground Segment (Exhibit 80) implemented a pricing increase in January of 2017. Therefore, the last reported year is probably a fair predictor of what prices will look like in the future.

Exhibit 80—Description of Pricing Changes (Ground Segment)

Source: SEC.gov, FedEx Corp 10-K, filing date July 16, 2018, retrieved September 24, 2018.

Before we adjust the historic trends in our model for the impact of price changes, we will need to check a few additional sources. Start with the company’s investor relations page. A press release dated August 3, 2018 announced that FedEx 84

Chapter 4: Model Calibration & Forecasting will not impose an additional residential surcharge for the holiday season. UPS (a primary competitor) on the other hand, will have a peak season surcharge. The interesting part of the release is the fact that this is the second consecutive peak season without a surcharge. From a modeling standpoint, this supports using past year’s pricing trends for the Ground Segment, and the increased pricing from the last quarter for Express. Next turn to the company’s fiscal fourth quarter earnings call. A few comments were made regarding pricing and other related metrics. The comments echo those written in the FedEx Chairman’s Letter to Investors, and previous earnings calls, that the company has been investing heavily in modernizing, and automating key shipping hubs: In terms of adding volume, I would say our investments in automation and technology have led us to have the most automated package sortation network in North America, if not the world. And that’s enabled us to be able to flex our capacity as needed based on market dynamics, and probably the best example of that is how we handle record volumes each year at peak. Concerning margins, we expect to continue to deliver strong volume and revenue growth driven by the trends that everybody here today so far has talked about by e-commerce. We’re intently focused on modernizing and optimizing the Ground network to drive profitability, and we continue to look for ways to maximize asset utilization by leveraging existing capacity and automation. -Henry J. Maier, FedEx Corp, President and CEO of FedEx Ground, Fiscal fourth quarter 2018 earnings conference call, June 19, 2018. Overtime this investment should improve the package yield and volume at peak times. There are two ways to approach this from a modeling standpoint: The aggressive approach would be to incorporate significant pricing improvements in the near-term. The conservative approach would be to continue using average pricing over the last few quarters, and as automation improves the package yield, it will increase the future pricing forecast. Before we decide which approach to take in our model, we can continue to research by turning to recent news headlines. There are articles covering President Trump’s twitter posts which focus on Amazon.com’s shipping costs which have been effectively subsidized by U.S. taxpayers through the U.S. Postal Service (USPS). Generally speaking the USPS is in an unfavorable financial position, and it is likely that price increases are in the works. Perhaps with pressure from Washington, these increases could be accelerated. This should ultimately result in a more favorable pricing environment for FedEx. The point was also discussed on the FedEx earnings call: Analyst Question: When do you think the Post Office will meaningfully raise rates? Is this a risk for you? Management’s Answer: I don’t think we can speculate on what USPS might or might not do. However, we do believe that the cost of last-mile delivery will continue to increase in the years to come, which will be an opportunity for FedEx.-Rajesh Subramaniam, FedEx Corp, Chief Marketing and Communications Officer and EVP, Fiscal fourth quarter 2018 earnings conference call, June 19, 2018. Next review news articles about Amazon’s plans to establish its own shipping service. This introduces much larger structural factors which will likely have effects that are long-term in nature, given that it will take time for this new business to develop. In the near-term the fact that the USPS delivers the majority of Amazon’s residential packages, the costs associated with developing a full logistics network, and the problems associated with sparsely populated areas outside of the major urban hubs, are more likely going to shelter FedEx’s pricing in the near-term. If anything this development from Amazon may offset any USPS pricing pressure from Washington, rather then impact overall shipping pricing, which is probably artificially low at this point. Based on all of these factors, I have decided that the most appropriate pricing forecast would be to apply the average year-over-year yield growth of the last fourth quarters for all businesses except for the U.S. Express sub-segments, for which I will apply the average growth for the last two quarters since these periods include the latest pricing increase. Of course if you disagree with my assessment, feel free to enter a different forecast in your version of the model. For the international business sub-segments we can reassess where foreign currency rates against the U.S. dollar have been over the quarter, and adjust the yield slightly up or down if necessary, as we approach the next quarterly release. Exhibits 81 and 82 demonstrate how to enter these forecasts into the model. Exhibit 81 represents the U.S. Overnight Box sub-segment within the overall Express Segment. I have set the forecast in cell S76 equal to the average yield growth rate for the last two reported quarters in cells P76 and Q76. In cell T76 I have expanded this estimate to the

85

average of the last three quarters, cells S76, Q76, and P76. By the time I reach cell U76, I have reverted back to a fourquarter moving average.

Exhibit 81—Express U.S. Overnight Box Segment Pricing Forecast

Exhibit 82 shows the forecast for the Ground Segment. I have set the forecast in cell S138 equal to the average yield growth rate for the last four reported quarters in cells N138, O138, P138, and Q138.

Exhibit 82—Ground Segment Pricing Forecast

Step 15b—Impact of TNT Express Acquisition and TNT Cyber Attack: Through our research in the previous steps, listening to earnings calls, and reading the 10-K and news articles, we have learned that FedEx acquired a large European competitor, TNT Express, in 2016. The TNT acquisition, has three primary implications on our forecast. First, since it was the largest acquisition in FedEx’s history, it will inevitability impact the comparability of historic results pre-acquisition to future results. Second, in June of 2018 TNT experienced an information technology cyber attack, which impacted the Express business and must be considered in our package volume forecast. Third, integration costs will need to be estimated in the non-GAAP adjustment section of our model since integration costs are non-recurring in nature. As a result we will need to strip out the costs from operating expenses. In the remainder of Step 15b we will address how to approach each of these items. Step 15b.1—TNT Acquisition and Comparability with Past Results: The 10-K includes substantial detail concerning the acquisition and integration of TNT. Exhibit 83 below explains the basics at a high level. The key takeaways are that the acquisition will have a meaningful impact, primarily on the International Express segments, although due to the nature of overlapping businesses within the FedEx ecosystem, there will likely be some immaterial effect on the other segments as well. In addition, we cannot use the periods prior to the acquisition, meaning the quarters before the fiscal first quarter ending August 31, 2016, to make projections for the International Express sub-segments.

Exhibit 83—Description of TNT Merger

Source: SEC.gov, FedEx Corp 10-K, filing date July 16, 2018, retrieved September 24, 2018.

To demonstrate this point, refer to the historic Average Daily Volume (ADV) year-over-year growth rates for the International sub-segments in Exhibit 84 below. Notice that after the TNT acquisition in May of 2016, the fiscal year 2017 quarterly ADV growth rates for the International Priority, Economy, and Domestic sub-segments skyrocket to an average of 34%, 40%, and 170% respectively. This is due to the fact that TNT results were not included in the fiscal year 2016 results, but were included in fiscal 2017. 86

Chapter 4: Model Calibration & Forecasting In fiscal year 2018 the growth rates stabilize since TNT results have been included for more than a year. Therefore we would not want to use the trends before fiscal year 2017 to shape our future period assumptions.

Exhibit 84—International Express Sub-Segments Historic ADV Growth Rates

Step 15b.2—TNT Cyber Attack Impact: There are many sources of information available related to the cyber attack. For forecasting purposes, we will focus on the disclosure from the “Risk Factors” section of the 10-K (refer to Exhibit 85 below), as well as comments made by management on the last few earnings calls: Turning to the segments and starting with Express. Adjusted operating income increased 11% to $813M driven by revenue growth, a positive net impact from fuel and continued cost efficiencies. Adjusted operating margin improved 20 basis points to 8.7%. The cyberattack at TNT impacted our as-reported and adjusted results by an estimated $100M for the Express segment, primarily from loss of revenue due to decreased shipments in the TNT network. As we noted last quarter, we are accelerating portions of our TNT integration as a result of the cyberattack. TNT integration expenses for the second quarter were $96M for Express and are included in the GAAP results. Despite the challenges from the cyberattack, total international average package volume increased 5%. - Alan B. Graf, Jr., CFO & Executive Vice President, FedEx Corp, Fiscal second quarter 2018 earnings conference call, December 19, 2017. I think, first thing you got to remember is the effects in Q3 were mostly one-off type of effects. Q4 ends up being a seasonally strong quarter and we’ve already told you what that’s going to be. Our TNT network was fully restored and back to business as usual as of the end of 2017. The recovery of the business over the last five months has been remarkable. And given the value proposition of the TNT road networks, our freight volumes have been strong, and we are experiencing solid growth in these products. The cyberattack continues to have a lingering effect in the third quarter, and our existing customer base has not been fully restored – has not fully restored all volumes as they continue to gain confidence in our ability to provide service and recovery of our business. -David L. Cunningham, Jr., President & Chief Executive Officer-FedEx Express, FedEx Corp, Fiscal third quarter 2018 earnings conference call, March 20, 2018. Analyst Question: How much of the $400M impact from the cyberattack should we assume comes back in fiscal year 2019? Have you fully recovered all TNT share losses? Management’s Answer: You can see in the results that we experienced year-over-year double-digit revenue growth in our international package and freight services this past quarter. While higher rates were certainly a major contributor, we’re also seeing solid year-over-year growth in freight traffic, a piece of our product portfolio that expanded significantly through the addition of TNT. Again, the recovery of the business over the past several months has been remarkable and we certainly owe major thanks to our sales, customer service, and IT professionals who’ve done an outstanding job of recovering from this attack. -David L. Cunningham, Jr., President & Chief Executive Officer-FedEx Express, FedEx Corp, Fiscal fourth quarter 2018 earnings conference call, June 19, 2018.

87

Exhibit 85—TNT Cyber Attack Disclosure

Source: SEC.gov, FedEx Corp 10-K, filing date July 16, 2018, retrieved September 24, 2018.

There is not enough information available to estimate exactly how much of an impact the cyber attack had on results. The most concrete data point is the estimated $100M impact on the fiscal second quarter. We will use this as the basis for adjusting the historic International Express sub-segments. Based on Step 15b.1 we know that the correct starting point for the forecast is the ADV growth rates from fiscal year 2017 (not 2016 before the TNT acquisition), so start by calculating the four quarter average ADV growth rate. For example, in Exhibit 86 the growth rate estimate in cell S92 of 0.1% is an average of cells N92, O92, P92, and Q92. The four-quarter period in our average calculation included an impact of approximately $100M in the fiscal third quarter, and some additional amount in the other quarters, which we can roughly estimate at $200M overall (note the $400M in the analyst’s question from the earnings call likely included some of the integration costs). Now let’s attempt to adjust the growth rate in the fiscal first quarter forecast (cells S92, S98, and S104) to get to a revenue impact of approximately $200M.

Exhibit 86—International Express Sub-Segments ADV Growth Rate Forecast - Average

88

Chapter 4: Model Calibration & Forecasting Start by summing the revenue lines for the International Priority $1,845M, Economy $871M, and Domestic $1,134M sub-segments (refer to Exhibit 86 cells S95, S101, and S107). Before adjusting the growth rates, the total revenue for these three International Express sub-segments is $3,977M. Now adjust the growth rates until the sum of the subsegment revenue is approximately $200M higher. You can accomplish this by adding 5% to each subsegment as shown in Exhibit 87 below. Then for the F2Q2019 forecast I have cut the 5% adjustment down to just 1%, and made no adjustment for the TNT cyber attack in F3Q2019.

Exhibit 87—International Express Sub-Segments ADV Growth Rate Forecast - Adjusted

Step 15b.3—TNT Acquisition Integration Costs: Next we must consider the TNT Express merger integration expenses in our forecast. We will enter this in the non-GAAP adjustment section of our model using the following comments from management on the last earnings call: Analyst Question: Can you review the total cost of integration since acquisition? And what’s left after 2019? Do the benefits outweigh the cost yet? Analyst Question: After the $450M in TNT integration expenses in 2019, will you still have $150M of integration expenses in 2020? Management’s Answer: So we’re now a little bit past two years of day one with TNT, and we’ve earned a tremendous amount. Recall we also suffered a very nasty cyberattack a year ago almost coming up on a year anniversary. And so our integration expenses are going to be a little bit higher than we had originally thought, but that’s not bad at all because we’ll be tightening our defenses from a cyber-standpoint, and we’re going to invest more than that than we had anticipated. But the results will be a much more productive and much more flexible IT network. Additionally, we have found additional productivity enhancers as we go through the integration that are going to really increase our returns substantially. I have to say fiscal year 2019, when we’re going to step up to about $450M worth of integration, is a very big year. We have great plans in place. We’re off to a good start. And this is a key year for us in terms of integration after 2019. We’ll have somewhere around $250M to $300M left in fiscal year 2020, but that’ll be around the edges. I think the meat on the chicken is going to be the current fiscal year that we’re in.Alan B. Graf, Jr., Executive Vice President & CFO, FedEx Corp, Fiscal fourth quarter 2018 earnings conference call, June 19, 2018. Analyst Question: So Alan just mentioned that integration costs will be higher than initially expected. You’re saying higher than the $1.4B guidance from last year. How much? Management’s Answer: I’d say right now based from what we’re seeing, it’s more like $1.5B, about a $100M more. If we find additional opportunities, we’ll keep you posted on that, but those are opportunistic as opposed to it’s costing us more than we thought.-Alan B. Graf, Jr., Executive Vice President & CFO, FedEx Corp, Fiscal fourth quarter 2018 earnings conference call, June 19, 2018. Based on these comments we will add $450M of integration costs to our model for fiscal year 2019, divided evenly across the four quarters, and $275M for fiscal year 2020. Refer to Exhibit 88 below for details. Notice that with these estimates, the total expected TNT integration costs are in-line with management’s comments of $1.5B ($327M + $477M + $450M + $275M = $1.5B).

89

Exhibit 88—TNT Express Integration Cost Forecast

Step 15c—Strategic Alliance with Walmart and Walgreens: In March of 2018 FedEx announced an agreement with Walmart to have FedEx “OnSite” package pickup location within Walmart stores. The company has a similar arrangement with Walgreens as well. The OnSite locations were discussed on the fourth quarter earnings call: Analyst Question: What’s needed to facilitate a higher utilization rate of your retail and OnSite network for customer pickups and returns versus a residential stop? Management’s Answer: Well, as I mentioned to you earlier, we are thrilled to have almost 11,000 convenient locations in the U.S. for customers to pick up their packages today. The three factors that drive it, first obviously is convenience. And we are right there. Locations are open later in the evening. 700 of them are open, in fact, 24 hours. The second issue is around porch piracy. The recent surveys show that 75% of online shoppers were concerned about porch piracy and 45% have a package either stolen or known someone that has had their package stolen. So those are two very important factors. The third one probably over the medium to long term is that, as I mentioned to you, the cost of last-mile continues to increase. These locations will not only become a convenient location but it’ll also become economic value for our customers. - Rajesh Subramaniam, Chief Marketing and Communications Officer & Executive VP, FedEx Corp, Fiscal fourth quarter 2018 earnings conference call, June 19, 2018. Overall the partnership appears to be a favorable strategic development, particularly to establish additional locations ahead of Amazon’s delivery service, and to align with Walmart who has become an important e-commerce player after the acquisition of Jet.com; however, the announcement probably does not change the modeling forecast in the nearterm, so I will not make any adjustments for this point. Step 15d—Saturday Delivery: In September of 2018 FedEx announced that it will expand its delivery service to Saturdays year round, instead of only during peak periods. This will likely increase the volume for FedEx Ground. For the time being I will increase the average year-over-year ADV for the Ground Segment by one percent for each quarter. This adjustment will be reassessed over the next few quarters to determine the true impact of Saturday delivery. The adjustment is shown in Exhibit 89, cell S136 which is set equal to the average of cells N136, O136, P136, and Q136 plus one percent. The increase will create a higher base for comparison in fiscal year 2020 (typically described as facing a “tough comp”). To reflect this I will decrease the 2020 average growth rates over the year beginning with a 2% decrease in F1Q2020 and trailing that adjustment factor down to zero by F4Q2019.

Exhibit 89—FedEx Ground Segment Forecast

90

Chapter 4: Model Calibration & Forecasting Step 15e—Air Fleet and Hub Modernization and Automation Strategy: There are a number of capital expenditure programs underway which need to be incorporated into our model. One of which is the fleet modernization program which will include 24 new aircraft expected to be delivered between 2020 and 2025. In addition there are ongoing automation efforts at many of the company’s shipping hubs. The two primary considerations of these efforts are in the estimates for: 1) capital expenditures, and 2) improved efficiency through reductions in operating expenses. To get a handle on how we should think about our forecast for these items we can review management’s comments from the last few earnings calls: Our capital spending forecast for fiscal year 2018 is now $5.8B, down $100M from the prior forecast due to lower projected capital spending at Ground. As we mentioned last quarter, we are optimizing Capex to capture the benefits of 100% expensing to further grow and improve the business. Plans are underway to modernize our Indianapolis hub for Express, which is expected to cost $1.5B. We are also planning to modernize our Memphis hub for Express, which is estimated to exceed $1B. Both of these projects will span multiple years. These hub modernizations will bring substantial improvements in operational efficiency and reliability. For example, the Memphis project includes construction of the large new sort facility with state-of-the-art sort systems, construction of a bulk truckload building and a new area to improve handling of oversized shipments that continue to increase with the growth in e-commerce. During the quarter, Express entered into an agreement with Boeing to accelerate the delivery of one 777 Freighter aircraft to fiscal year 2019, and three to fiscal year 2020. We will announce our fiscal year 2019 Capex forecast in June.-Alan B. Graf, Jr., Executive Vice President & CFO, FedEx Corp, Fiscal third quarter 2018 earnings conference call, March 20, 2018. The investments in our Memphis and Indy hubs will modernize and automate these key facilities. Big data and our real-time ability to mine and improve efficiency and productivity of these facilities by directing packages most efficiently through the hubs. As Alan mentioned, at Memphis, where we will have a new bulk truckload facility, an oversized shipment handling capability, plus automated sorting and secondary sorting capability. At Indy, we’re increasing the box sort capacity from 111,000 packages to 147,000 packages per hour. We’re putting in a small package sort system of 150,000 packages per hour. And we will have increased international sort capacity as well. These facilities will improve the reliability of our networks, lower costs, improve safety for a better place to work for the thousands of team members who work in these operations. -David L. Cunningham, President & CEO, FedEx Express, Fiscal third quarter 2018 earnings conference call, March 20, 2018. Analyst Question: Given the new planes and the ongoing hub expenses, is it fair to assume that capex will increase in fiscal year 2020 and beyond from fiscal year 2019 levels? Management’s Answer: We’re going to continue to grow our revenues and continue to grow our cash flows. And as we see opportunities to invest whether it’s expense, capital, or acquisitions that is going to increase the long-term value of our company, we’re going to do that. Obviously, we benefited greatly from the TCJA tax rate and 100% expensing for the next five years. That obviously makes capex investing less risky. Although in the case of the aircraft, that was certainly not the driver when you’re presented with opportunities to make asset acquisitions at a really good price and those that are driving tremendous productivity improvements and operating expense reductions. We need to make those moves. So our capex is going to spike up in fiscal year 2020 and fiscal year 2021 mostly as a result of Express. We are significantly reducing our outlook for capex at Ground. We won’t build a hub for a long time. Ground has many innovative and in some ways revolutionary cost reduction programs under way right now, all of which involves sweating assets, improving productivity, and lowering costs. And the same thing at Freight. So it will be easily managed. I won’t say that we’ll be able to stay at the 8% level but it will be very easily managed and covered with our cash flows, and we’ll still have excess cash flow. -Alan B. Graf, Jr., Executive Vice President & CFO, FedEx Corp, Fiscal fourth quarter 2018 earnings conference call, June 19, 2018. For fiscal year 2019, depreciation and amortization is expected to be approximately $3B, so FedEx will generate very strong cash flows. Capital spending is expected to decline slightly to $5.6B, or about 8% of projected revenues. -Alan B. Graf, Jr., Executive Vice President & CFO, FedEx Corp, Fiscal fourth quarter 2018 earnings conference call, June 19, 2018. 91

There is a substantial amount of information in these comments. Let’s start with the capex forecast. Given all of the moving pieces, we will use the estimate provided by management which is $5.6B in capital expenditures for 2019. We can also increase this to $6B for fiscal year 2020 and 2021. Before we input these estimates into our model, keep in mind that we must finalize our revenue forecast since our capex estimates are based on a ratio of capex-to-revenue. For now save these amounts and we will revisit this point after the revenue forecast is complete. Similar to capital expenditures, we also must reflect the improvement in hub efficiency through reductions in operating expenses; however, we have setup the operating expense forecast to be driven by a percentage of our revenue projection. Therefore, we wait for the finalized revenue estimate to incorporate this item as well. Step 15f—Dividend Forecast: In June 2019 management announced a 30% increase in the dividend beginning in fiscal year 2019. This is a relatively easy input into the model in row 46 (refer to Exhibit 90, cells S46 through V36). After inputting the dividend growth rate, the dividend payment will automatically calculate through the equations setup in Chapter 2, Step 5k based on the number of shares outstanding from Step 5i. The dividend outflow will calculate in the Cash Flow Statement, and this amount will be removed from the retained earnings section of the Balance Sheet.

Exhibit 90—Dividend Forecast

Step 15g—Tesla Semi-Trucks and Other Emerging Technological Changes: On March 26, 2018 FedEx announced that the company had reserved 20 electric trucks from Tesla. Production of the trucks will begin in 2019. While fully electric trucks, and potentially autonomous vehicles could be game changers in the future, I have not made any adjustments to the earnings model for these innovations yet. Management discussed some of the emerging technologies on the last earnings call: Well, the impact of innovations in the connected world, whether it’s the Internet of Things and sensorbased logistics, mobility, blockchain, autonomy, all of those things are accelerating across the enterprise. If you look at sensor based logistics and the announcements and capabilities that we have shown in our new Bluetooth low-energy sensors, that’s resulted in a huge number of patent filings and really is a viable technology in and of itself today. What takes some time to roll though is the infrastructure and Wi-Fi networks and rolling stock and an aircraft in order to marshal all those sensors as they operate in the world. And in the back-end systems, they’re needed to handle that volume of data. With regard to blockchain, we’re quickly seeing blockchain capabilities moving towards production primarily through our involvement in BITA, the Blockchain In Transport Alliance, with now hundreds of participating companies. FedEx Freight is leading the charge there and are chairing the committee on standards for BITA, and we have several applications that are working their way forward. And then lastly with regard to drones, drones are still really specializing in observation and inspection with sophisticated optics and their ability to look at aircraft and airframes. That’s a very advanced capability that we’re already using today. But things like lift and range are limiting their use in transport although we are testing them in some of our larger ramp facilities to deliver parts to mechanics and things of that nature. - Robert B. Carter, EVP FedEx Information Services & CIO, FedEx Corp, Fiscal fourth quarter 2018 earnings conference call, June 19, 2018. There are a number of important technologies that could dramatically reshape the future of the shipping industry, however, these changes are long-term in nature. It is probably too early to incorporate meaningful estimates into our earnings model at this point.

92

Chapter 4: Model Calibration & Forecasting Step 15h—Review the “Recent Accounting Guidance Section” of the 10-K for Upcoming Accounting Changes: The accounting disclosure in the financial statement footnotes is important since changes in accounting treatment could have a significant impact on results, and potentially change the way we model a certain line item in the future. The first step is to read through the “Recent Accounting Guidance” section of the 10-K filing. Next you will need to develop an understanding of the accounting changes. If there is not enough information for you to get comfortable with the accounting in the company’s footnote disclosure, then you can look to third party sources for additional details. The Financial Accounting Standards Board (FASB) issues Accounting Standards Updates (ASUs) which are available to the public free of charge at FASB.org, and include clear descriptions of new standards. In addition, public accounting firms (Deloitte, PricewaterhouseCoopers, KPMG, and Ernst & Young) typically write summary reports on new accounting standards, which you can find for free online. If you are still unclear of how new standards will impact the company you are modeling, listen to the last few earnings conference calls to see if management has made any comments about the anticipated effects of the new standard. The final step would be to look for early ASU adopters. Most ASUs have a date when filers can early adopt the new accounting standard, and a later date by which all companies must comply with the new standard. If the company you are covering is not an early adopter, than you can find a company who has already filed under the new standard for an example of how the company’s financials may change. In the case of the FedEx accounting changes, the disclosure is relatively straight-forward, so we will not have to consult with other sources to understand the impact. Read through the expected accounting changes in Exhibit 91. The first two points on debt issuance and share-based compensation do not have a material impact on results. The point on revenue recognition changes would likely be material for most companies, however, the shipping business is more straightforward from a revenue recognition standpoint, and the accounting change effective in fiscal 2019 will not likely have a material impact for FedEx.

93

Exhibit 91—Recent Accounting Guidance Footnote of the 10-K

Source: SEC.gov, FedEx Corp 10-K, filing date July 16, 2018, retrieved September 24, 2018.

The lease accounting change will have a material impact on the Balance Sheet; however, the impact on the Income Statement will be immaterial, and the change will not come until fiscal 2020. Since there is no material impact on earnings or free cash flow, there is no need for us to attempt to adjust our model for this item ahead of the reporting change. The last item related to pension accounting is one of the more important items for our model. It will result in the inclusion of service costs in operating expenses, but all other pension expenses will go to the other expense line on the Income Statement instead of operating expenses. Management discussed the point on the last earnings call: Regarding pensions, I would like to take a few minutes to explain the effect the new pension accounting rules will have on our financial results for fiscal year 2019. Going forward, only the pension service cost will be included in operating expenses. All the other elements that make up total pension expense will now be classified as other non-operating expenses including the year-end mark-to-market adjustment. For example, in fiscal year 2018, our total pension expense excluding mark-to-market and other pension adjustments was $224M while our service cost was $812M. So while there is no impact to net income from these new rules, they will negatively impact our operating margin by about 90 basis points. Of course, prior years will be recast to conform to these new rules, so there will be no year-over-year impact once we have an apples-to-apples comparison starting in the first quarter of fiscal year 2019. -Alan B. Graf, Jr., Executive Vice President & CFO, FedEx Corp, Fiscal fourth quarter 2018 earnings conference call, June 19, 2018. The correct modeling approach would be to reformat our historic results and forecast ahead of the fiscal first quarter 2019 release to strip out the service costs from the other pension expenses, so we could calculate a new operating margin including the latest accounting treatment; however, we are not going to perform this step since the net earnings and cash flows will not be impacted by this accounting change. Instead we will reformat the model after the fiscal first quarter 2019 release, which will include the new accounting standard. 94

Chapter 4: Model Calibration & Forecasting Step 15i—Have Input Prices Changed? Input prices will have a direct impact on profitability. Any expected future changes should be incorporated into your model through changes in the operating expense assumptions. If the company you are modeling uses commodities traded on an exchange as inputs, it may be easier to incorporate future raw material prices into your forecast. For example, FedEx uses jet fuel and gasoline for its fleet of airplanes and delivery trucks. You can track the changes of fuel costs on the exchange, and change the forecast of these lines within the segment section of your model, up until the day of the company reports results. Just remember that there is a fuel surcharge which passes some of the increases in fuel costs onto the customer. We will discuss this point later in the forecasting process. In addition to being capital intensive, shipping companies also incur substantial labor costs. You can track changes in minimum wage rates and other labor developments (i.e. progress on automation initiatives), and incorporate your assumptions on the labor input price changes in your salary expense line forecast. If you are modeling a company which manufactures products, it is likely that many of the input prices of those products will not be published. This will make it difficult to calculate a gross margin for the product. Many research firms publish cost estimates for popular products in reports called “product teardowns”. In a teardown report, the company will physically pull apart the product, identify each input, and estimate its cost. Teardown reports for newly released products are typically expensive. As the product ages the report prices fall, and are eventually available for free. To see examples of such a report, perform a internet search on “teardown” for the company’s product. You can also watch teardown videos on YouTube. Remember to go to multiple sources to validate the information. Some reports come with pricing estimates of each part. Only insiders know exactly how much the company pays for each component. To minimize the risk of overreliance on a poor estimate, consider using multiple teardown reports and averaging the cost of each input. Step 15j—Have Customer Preferences Changed? This step is important to consider in the context of the competitive landscape. For shipping companies, who is the ultimate customer of the shipping service? In many cases it is a consumer who is getting more and more demanding, and looking for overnight delivery to their doorstep for nearly all products. This service does of course come with a price. Preferences can shift over time to be more, or less price elastic depending on many factors including the availability of a similar service from a competitor, or simply changes in the macroeconomic environment. You should also consider how preferences may change in the long-term, and incorporate some of these trends into your forecast. Step 15k—Has the Macroeconomic Environment Changed: If the company you are modeling is in a cyclical industry, consider how current economic conditions, and your expectations for future macroeconomic developments, may impact financial results. Economic impact can be very difficult to quantify, particularly the effects of changes in foreign exchange rates, since the company’s hedging strategies change over time. At the very least you should consider the financial statement lines which would be impacted, and adjust the historic ratios up or down to reflect a reasonable estimate. FedEx operates in a cyclical industry. The company’s revenue tends to have a relatively high correlation with global industrial production, GDP growth, and consumer spending. Management tracks global macroeconomic activity closely, and uses expectations of economic growth for strategic planning and forecasting purposes. The company publishes economic projections on the investor relations website, and discusses updates to the forecast on quarterly earrings calls. For example, on the fiscal fourth quarter conference call management gave the following forecasts (as of June 4, 2018): •

U.S. GDP growth of 2.3%, 2.9%, and 2.7% for calendar years 2017, 2018, and 2019 respectively.

•

Global GDP growth of 3.1%, 3.2%, and 3.1% for calendar years ‘17, ‘18, and ‘19 respectively.

•

U.S. industrial production growth of 1.6%, 3.8%, and 2.8% for calendar years ‘17, ‘18, and ‘19 respectively.

•

U.S. consumer spending growth of 2.8%, 2.7%, and 2.5% for calendar years ‘17, ‘18, and ‘19 respectively.

95

On September 28, 2018 management updated this forecast to: •

U.S. GDP growth of 2.2%, 2.9%, and 2.6% for calendar years ‘17, ‘18, and ‘19 respectively.

•

Global GDP growth of 3.2%, 3.2%, and 3.0% for calendar years ‘17, ‘18, and ‘19 respectively.

•

U.S. industrial production growth of 1.6%, 3.7%, and 2.7% for calendar years ‘17, ‘18, and ‘19 respectively.

In the press release management cited signs of slowing growth in the Eurozone and China. This coupled with growing concerns over a potential global trade war with the U.S. could impact global shipping. To incorporate the potential impact of slowing foreign production on shipping, I have adjusted my average growth trends of volume and price starting in the fiscal third quarter of 2019, reducing the quarters by 4%, than trailing down to 1% over four quarters in the international segments. Notice I have started this adjustment in the third fiscal quarter, since I expect the first two fiscal quarters of macro headwind will be offset by the post-cyber attack tailwinds, discussed in Step 15b.2. As the macro environment changes over the next few months, we can reassess these adjustments.

Step 16: Consider Management’s Guidance

Most companies offer earnings guidance for the next fiscal period in their quarterly earnings press release, or on their quarterly earnings conference calls. FedEx gives guidance on many key metrics which we can incorporate into our model. Refer to Exhibit 92 for the details of the company’s typical quarterly guidance, Exhibit 93 for guidance on the consolidated operating margin including the effects of the pension plan accounting changes, and the following comments from management made on the fiscal fourth quarter earnings call: We are confident that we will achieve the operating income improvement at Express of $1.2B to $1.5B in fiscal year 2020 versus fiscal year 2017 assuming moderate economic growth and current accounting and tax rules. -Alan B. Graf, Jr., Executive Vice President & CFO, FedEx Corp, Fiscal fourth quarter 2018 earnings conference call, June 19, 2018. While no cash contributions are required in our primary U.S. pension plans, we will make voluntary contributions during fiscal year 2019 but they will be at a much lower level than the $2.5B we contributed in fiscal year 2018. -Alan B. Graf, Jr., Executive Vice President & CFO, FedEx Corp, Fiscal fourth quarter 2018 earnings conference call, June 19, 2018. For fiscal year 2019, FedEx is targeting revenue growth of approximately 9%. We are also targeting the following before year-end mark-to-market retirement plan accounting adjustments and excluding TNT Express integration expenses: an operating margin of approximately 8.5%, remember that’s being reduced by the service cost; an earnings of $17 to $17.60 per diluted share; additionally, we are anticipating an effective tax rate of approximately 25% prior to year-end mark-to-market retirement plan accounting adjustments, which is higher than our fiscal year 2018 effective tax rate due to tax benefits from transactions and TCJA impacts that will not reoccur during fiscal year 2019. We expect our fiscal year 2019 cash tax rate to be lower than the U.S. statutory rate of 21% due to the favorable capital expensing provision created by the TCJA. Cash flows will improve as earnings grow. For fiscal year 2019, depreciation and amortization is expected to be approximately $3B, so FedEx will generate very strong cash flows. Capital spending is expected to decline slightly to $5.6B, or about 8% of projected revenues. We are confident that we will achieve the operating income improvement at Express of $1.2B to $1.5B in fiscal year 2020 versus fiscal year 2017 assuming moderate economic growth and current accounting and tax rules. -Alan B. Graf, Jr., Executive Vice President & CFO, FedEx Corp, Fiscal fourth quarter 2018 earnings conference call, June 19, 2018. Management also disclosed additional fiscal year 2019 financial targets within the fiscal fourth quarter 2018 earnings presentation slides which include the following points: •

Depreciation and amortization is expected to be approximately $3B.

•

Capital expenditures are expected to be $5.6B or approximately 8% of revenue.

•

TNT integration expenses are projected to be $450M.

•

Voluntary pension contributions are expected to be significantly lower than fiscal year 2018.

96

Chapter 4: Model Calibration & Forecasting

Exhibit 92—Guidance From Fiscal Fourth Quarter 2018 Earnings Release (8-K)

Source: SEC.gov, FedEx Corp 8-K Current Report, filing date June 19, 2018, retrieved September 24, 2018.

97

Exhibit 93—Operating Margin Guidance (8-K)

Source: SEC.gov, FedEx Corp 8-K Current Report, filing date June 19, 2018, retrieved September 24, 2018.

Guidance can come in a direct form, such as an explicit estimate written in the press release, or in an indirect form, such as a discussion of broad potential outcomes at an investor conference. Indirect guidance tends to be focused on long-term strategic direction rather than short-term financial ranges. FedEx provided substantial guidance for the fiscal year of 2019; However, the level of detail has fluctuated over time. In the past, management has provided additional metrics. At one point management issued forecasts for the individual segments. In some years guidance was only provided for capital expenditures and non-GAAP diluted EPS. Most analysts place a high reliance on guidance in their earnings models since management has access to non-public information about the company. In addition, it typically takes about a full month for a company’s financial reporting team to consolidate the financial statements before releasing the details to the public. This means that one-third of the next reporting period has already passed by the time the guidance is released. The timing gives management relatively good visibility into how the next quarter is developing. Step 16a—Can We Trust Management’s Guidance: If management gives misleading guidance, they will have to answer to analysts and investors at the end of the quarter. Of course there could be short-term events which may give management an incentive to be more optimistic in their guidance, such as a product release or debt/equity offerings. In general, we should expect management to act in good faith. This does not mean we should accept their guidance blindly. Before I rely on management’s guidance, I review their track record. Table 7 below shows a breakdown of the FedEx guidance for capex and EPS over the last four years, and the actual reported results for the same periods. In general, the reported results have been reasonably close to the guidance provided by management. There were two periods where guidance was impacted by unanticipated events. Once in 2016 as a result of the TNT cyber attack, and again in 2018 due to the impact of corporate tax reform. Overall management has historically done a remarkable job of forecasting results, and has been diligent in updating guidance when unexpected events occur. Therefore we should incorporate the latest guidance from management into our model.

98

Chapter 4: Model Calibration & Forecasting

Table 7—Historic Guidance From Management Guidance For FY2015 FY2015 FY2015 FY2015 FY2016 FY2016 FY2016 FY2016 FY2017 FY2017 FY2017 FY2017 FY2018 FY2018 FY2018 FY2018

Date Released 6/18/2014

From the F4Q14 release

9/17/2014

From the F1Q15 release

12/17/2014

From the F2Q15 release

3/18/2015

From the F3Q15 release

6/17/2015

From the F4Q15 release

9/16/2015

From the F1Q16 release

12/16/2015

From the F2Q16 release

3/16/2016

From the F3Q16 release

6/21/2016

From the F4Q16 release

9/20/2016

From the F1Q17 release

12/20/2016

From the F2Q17 release

3/21/2017

From the F3Q17 release

6/20/2018

From the F4Q17 release

9/17/2018

From the F1Q18 release

12/19/2017

From the F2Q18 release

3/20/2018

From the F3Q18 release

Guidance Capex

EPS

EPS

(Non-GAAP exMtM, incTNT)

(Non-GAAP exMtM, exTNT)

$4.2B

N/A

$8.50 to $9.00

$4.2B

N/A

$8.50 to $9.00

$4.2B

N/A

$8.50 to $9.00

$4.2B

N/A

$8.80 to $8.95

$4.6B

N/A

$10.60 to $11.10

$4.6B

N/A

$10.40 to $10.90

$4.6B

N/A

$10.40 to $10.90

$4.8B

N/A

$10.70 to $10.90

$5.1B

N/A

$11.75 to $12.25

$5.6B

$10.85 to $11.35

$11.85 to $12.35

$5.6B

$10.95 to $11.45

$11.85 to $12.35

$5.3B

$10.80 to $11.30

$11.85 to $12.35

$5.9B

$12.45 to $13.25

$13.20 to $14.00

$5.9B

$11.05 to $11.85

$12.00 to $12.80

$5.9B

$11.45 to $12.05

$12.70 to $13.30

$5.8B

$17.90 to $18.30

$15.00 to $15.40

99

Date Released

Actual Reported Results Capex EPS

EPS

(Non-GAAP exMtM, incTNT)

(Non-GAAP exMtM/TNT)

6/17/2015

$4.3B

N/A

$8.95

6/21/2016

$4.8B

N/A

$10.80

6/20/2017

$5.1B

$11.18

$12.30

6/19/2018

$5.7B

$16.67

$15.31

Step 16b.1—Calibrate the Model to Include Management’s Guidance (Revenue): Our next step is to set our upcoming fiscal period estimates to meet management’s guidance. If we are not currently falling into the guided range, then we have probably over or under estimated the effects of the items in Step 15. Typically I set my results equal to the midpoint of management’s guidance. If you believe management tends to be conservative or aggressive with their estimates, then you may choose to set your forecast at the high-end or low-end of the guided range. Notice that FedEx does not give specific guidance for each sub-segment. This means that in order to calibrate the model to meet the guided EPS estimate, we will need to make judgement calls as to which service segments we will adjust, and which metrics within the segments. If I am unsure of which to adjust, I tend to select the segment which has the largest contribution to revenue and operating margin. To keep track of which metrics I have calibrated to meet management’s guidance, I shade the cells purple, and insert a comment with the full details of the range given by management. Exhibit 94 shows the revenue results for all of the inputs we entered in Steps 15a through 15k. Since the resulting yearover-year revenue growth in the model of 9.2% is inline with management’s guidance of approximately 9%, we do not need to make any adjustments to our revenue forecast.

Exhibit 94—Calibrate the Model to Meet Management’s Guidance (Revenue)

Step 16b.2—Calibrate the Model to Include Management’s Guidance (Capex): Recall from Step 15e we intended to calibrate our FY2019 capital expenditure estimate to equal management’s guidance of $5.6B, however, we had to complete the revenue forecast before continuing the process. Since we have completed our revenue estimate in Step 16b.1, we are now able to enter our capex forecast. Refer to Exhibit 95 below. Note that we change the input cells for the ratio of capex-to-revenue in the assumptions section of the Cash Flow Statement. This will recalculate the capex forecast on row 322 based on the revenue projection in the Income Statement on row 13.

Exhibit 95—Calibrate the Model to Meet Management’s Guidance (Capex)

Step 16b.3—Calibrate the Model to Include Management’s Guidance (Depreciation): Now that we have the capex forecast complete, we can enter our assumptions for depreciation and amortization. Refer to Exhibit 96 below for details. Enter the ratio of depreciation and amortization-to-average Property and Equipment (P&E) on row 297. The average for fiscal year 2018 was 1.45%, however, in order to get back to management’s guidance for depreciation of approximately $3B, we must enter a ratio of 1.3%. After we enter this assumption, the P&E balance will recalculate on row 250 based on the capex forecast entered in Step 16b.2. Depreciation will calculate using the new asset base and the ratio on row 297. The forecasted amount will be added to the accumulated depreciation balance on row 251, to the depreciation expense line on the Income Statement on row 18, and to the depreciation adjustment on the Cash Flow Statement on row 307. 100

Chapter 4: Model Calibration & Forecasting

Exhibit 96—Calibrate the Model to Meet Management’s Guidance (Depreciation)

Step 16b.4—Calibrate the Model to Include Management’s Guidance (FedEx Express): Although management does not provide guidance for each segment, they have provided a target for the FedEx Express Segment operating margin, which they anticipate will increase by $1.2B to $1.5B in fiscal year 2020 (compared to fiscal year 2017). The assumptions from Step 15 resulted in an Express Segment operating margin below this guided range. Based on management’s guidance, our original forecast likely underestimated the impact of shipping hub automation and TNT integration synergies on operating expenses. Either that or management’s guided margin is too aggressive. You can decide if you would like to leave the margin assumption as is in your version of the model. For the sake of getting back to management’s guidance in this version of the model, I will adjust the margin upward. I have set the “all other operating expenses as a percentage of revenue” ratio in row 68 (refer to Exhibit 97 below) equal to the comparable quarter from the previous year, minus 50 basis points in fiscal year 2019, and 150 basis points in 2020. After entering these adjustments into the model, the fiscal year 2020 FedEx Express Segment operating margin comes to just over $4B (refer to Exhibit 97, cell AB69), which represents an increase of approximately $1.3B (cell AB70) from the 2017 operating income. This estimate is now within management’s guided range.

Exhibit 97—Calibrate the Model to Meet Management’s Guidance (FedEx Express)

Step 16b.5—Calibrate the Model to Include Management’s Guidance (Operating Margin): Based on the assumptions we have entered into our model, our consolidated operating margin forecast seems reasonable relative to management’s guidance. Refer to Exhibit 98, which shows the consolidated operating margin. The non-GAAP margin is 9.4%, which excludes the forecast for TNT integration expenses and pension MTM adjustment. After removing approximately 90 basis points to account for the impact of the pension expense accounting change (refer to Step 15h for additional details), the net adjusted operating margin would be approximately 8.3%, generally inline with management’s guidance.

101

Exhibit 98—Calibrate the Model to Meet Management’s Guidance (Operating Margin)

Step 16b.6—Calibrate the Model to Include Management’s Guidance (Effective Tax Rate): Management guided the effective tax rate for fiscal year 2019 to 25%. Enter this forecast in row 221, which will be applied against the income before taxes on row 34, to populate the provisions for income tax on row 36 of the Income Statement.

Exhibit 99—Calibrate the Model to Meet Management’s Guidance (Tax Rate)

Step 16b.7—Calibrate the Model to Include Management’s Guidance (Share Count): Management has not explicitly forecasted the fiscal year 2019 share count in their outlook disclosure; however, the diluted share count can be derived using management’s guidance of the TNT integration expenses, and the integration expense per share. Refer to Exhibit 92 which demonstrates the calculation. The guidance given for integration expenses is $450M, less the estimated tax impact of $85M, or a net impact of $365M. In order to get back to the per share impact of $1.35 (refer to Exhibit 92), the share count for fiscal year 2019 must be 270M shares. We can back into the expected share count (refer to Exhibit 100, cell W41) by changing the repurchase assumptions in cells S228 through V228.

Exhibit 100—Calibrate the Model to Meet Management’s Guidance (Share Count)

Step 16b.8—Calibrate the Model to Include Management’s Guidance (EPS): Now that all of the assumptions have been entered into our model, the Earnings Per Share (EPS) will calculate consistent with the historic calculation described in Chapter 2, Step 5j. The modeled non-GAAP diluted EPS for fiscal year 2019 of $17.21, is in the range provided by management of $17.00 to $17.60, therefore no additional adjustments are required.

102

Chapter 4: Model Calibration & Forecasting

Exhibit 101—Calibrate the Model to Meet Management’s Guidance (EPS)

Step 17: Review the Consensus Estimates

The consensus, or average estimate of sell-side analysts, for revenue and EPS are published for free by many different sources (i.e. Nasdaq, Google Finance, Yahoo Finance). Additional details, such as gross margin, operating margin, and EBIDTA estimates are available with premium subscriptions (i.e. Bloomberg, FactSet, Reuters). The consensus estimate provides a reasonableness check for your model. There are two approaches with regards to consensus estimates which you may want to consider in your analysis: 1) After you complete your model forecast compare the results back to the consensus to see if there is something you may have missed, or 2) Before entering your own assumptions, calibrate your model to meet the consensus estimate as a baseline view, then enter your opinions of where you think the consensus estimate may be incorrect. When considering using consensus estimates in an earnings model it is important to understand the inherent limitations of the data. For example, not all analysts enter estimates for each published metric. As a result, some individual financial statement line items may look distorted relative to the consensus top-line revenue, or bottom-line Earnings Per Share (EPS) estimates. To demonstrate this concept consider Table 8 below which shows the FedEx consensus estimates for various Income Statement line items prior to the fiscal first quarter 2019 earnings release. Table 9 shows the number of analysts who have submitted estimates to be included in the consensus. Notice that only 10 analysts have enter fiscal first quarter 2019 estimates for operating profit, which is far fewer than the 18 analysts who have entered estimates for revenue, and 23 analysts who have entered estimates for EPS. Consider what would happen if you attempt to model the FedEx fiscal first quarter based on the consensus revenue estimate, then use the consensus operating margin estimate, and reasonable assumptions for other income and taxes. You may not get back to the consensus EPS estimate. This is due to the fact that the consensus estimates do not reflect the full earnings model of each analyst who participated in the earnings survey.

103

Table 8—Consensus Analyst Estimates for FedEx (Quarterly) Account Revenue ($M)

F1Q2018E Consensus

F2Q2018E Consensus

F3Q2018E Consensus

F4Q2018E Consensus

$16,871M

$17,782M

$17,880M

$18,674M

Operating profit (non-GAAP, $M)

$1,437M

$1,454M

$1,398M

$2,166M

Operating margin (non-GAAP, %)

8.5%

8.2%

7.8%

11.6%

EBITDA ($M)

$2,207M

$2,240M

$2,175M

$2,978M

Pre-tax profit ($M)

$1,371M

$1,439M

$1,346M

$2,155M

Net income ($M)

$1,027M

$1,072M

$1,026M

$1,610M

$3.80

$4.02

$3.84

$6.03

$220M

$534M

$577M

$970M

$1,383M

$1,437M

$1,493M

$1,493M

Earnings Per Share (EPS, $) Free cash flow ($M) Capital Expenditure ($M)

Table 9—Number of Analysts Included in the Consensus (Quarterly) F1Q2018E # of Analysts 18

F2Q2018E # of Analysts 18

F3Q2018E # of Analysts 18

F4Q2018E # of Analysts 18

Operating profit

10

12

10

10

Operating margin

N/A

N/A

N/A

N/A

EBITDA

16

16

16

16

Pre-tax profit

13

13

13

13

Net income

14

14

14

14

Earnings Per Share

23

23

23

23

Free cash flow

3

3

3

3

Capital Expenditure

3

2

3

3

Account Revenue

The consensus limitation is also apparent in the differing number of analysts who enter both quarterly and annual estimates. As you can see in Table 9, only 18 analysts have submitted revenue estimates for the fiscal first quarter of 2019, while 25 have entered estimates for the 2019 fiscal year (Table 11). In fact, if you sum the revenue estimates for the forecasted quarters, the result does not equal the consensus estimate for the full year of 2019 ($17,207M vs $17,267M). The difference is due to the fact that not all analysts submit quarterly estimates. Even fewer enter estimates over a four year horizon, which makes the future consensus estimates less reliable. Given the variability in the number of analyst estimates included in different consensus metrics, it is important to consider how many analysts have entered estimates before deciding whether or not to incorporate the consensus projections into your analysis.

104

Chapter 4: Model Calibration & Forecasting

Table 10—Consensus Analyst Estimates for FedEx (Annual) Account Revenue ($M)

FY2019E Consensus

FY2020E Consensus

FY2021E Consensus

FY2022E Consensus

$71,267M

$75,123M

$78,968M

$83,791M

Operating profit (non-GAAP, $M)

$6,110M

$7,129M

$7,750M

$7,852M

Operating margin (non-GAAP, %)

8.6%

9.5%

9.8%

9.4%

EBITDA ($M)

$9,377M

$10,559M

$11,323M

$11,256M

Pre-tax profit ($M)

$6,120M

$7,102M

$7,842M

N/A

Net income ($M)

$4,637M

$5,364M

$5,858M

$6,028M

$17.52

$20.19

$22.70

$24.56

Free cash flow ($M)

$1,859M

$2,280M

$3,476M

$4,566M

Capital Expenditure ($M)

$5,607M

$6,413M

$6,790M

N/A

Earnings Per Share (EPS, $)

Table 11—Number of Analysts Included in the Consensus (Annual) FY2019E # of Analysts 25

FY2020E # of Analysts 23

FY2021E # of Analysts 9

FY2022E # of Analysts 1

Operating profit

17

16

7

1

Operating margin

N/A

N/A

N/A

N/A

EBITDA

21

20

9

1

Pre-tax profit

19

18

7

0

Net income

21

20

8

1

Earnings Per Share

30

28

9

1

Free cash flow

11

12

5

1

Capital Expenditure

11

10

3

0

Account Revenue

If you choose to calibrate your model to meet the consensus estimates, you should understand that the exercise is not simply a mathematical exercise, as there is some judgment involved. Differences in lower-level model estimates add to the complexity. For example, if you review the models from several different sell-side analysts, each will have very different estimates of the individual components (in the case of FedEx we are referring to the sub-segment volume and yield estimates). Despite the difference in segment estimates, like magic, most analysts will get back to a point within management’s guided range. This is no coincidence. Analysts know that management has access to the best information about their respective companies. Unless the company has a trend of consistently beating or missing guidance, analysts will typically set their estimates within management’s range. Analysts will also consider their investment recommendations in their estimates, meaning if they have a “buy” rating they will make sure they are at the top-end of management’s range, or the low-end for a “sell” recommendation. Similar to our calibration of guidance discussed in the previous section, if we are calibrating our model to meet the consensus estimate, we must choose one metric to back into the final EPS result. I typically choose the estimate which is the most significant driver of earnings. We can use the Goal Seek function in Excel to reach the result. This function 105

will essentially back into a number, based on the equations setup in the Excel file. To run the function, click on “Data”, “What-If Analysis”, and “Goal Seek.” Once the Goal Seek window appears, you can enter the cell you want to change in the “set cell” section, and set the value you want to reach (such as the consensus EPS estimate), by changing a given metric in your model. Excel will automatically change the metric to the value required to meet the consensus estimate (or to reach the “goal you seek”). If you want to find the cells I use to reach the consensus estimate in my consensus-based models, look for the values which are carried out to several decimal places (usually 10 places or more). By looking at the number of decimal places, you can tell that the cell represents the result of a goal seek function (or alternatively an equations which has been saved as a value). Keep this in mind if you are reviewing the model of a sell-side analyst, and want to find out if they backed into a number. If they have, the decimal place will likely be carried out, since it is unlikely that he or she manually typed in that many digits. Pitfall: By now you should be recognizing the flow of assumptions in the forecast columns. New research associates tend to waste hours of time by not following this simple financial modeling rule: When entering future forecasts always move from left-to-right and top-to-down! This means always start by forecasting revenue across the top of the Income Statement. We do this because many of the ratios we use are based on a percentage of revenue. For example, Certain operating expenses are typically projected using a ratio compared to revenue. If you are trying to reach a certain operating expense target for the next quarter, and you start by calibrating the operating expense line, then move up to the revenue line, your operating expense will change again after your revenue estimate changes. As a result you will end up going in repetitive circles throughout your forecast like a dog chasing his tail. The same logic applies in moving from left to right. Always start with the earliest forecast periods (the left) and move across to the later periods (to the right). This is important because many of the future periods are based on growth rates applied against the earlier periods. If you reach your targets for the end of a particular year (a few quarters from now), then move right-to-left to adjust your first forecast quarter, you will need to re-adjust your full year data again since it will change when the earlier periods are adjusted, and the growth rates are applied in the equations.

Step 18: Incorporate Your Opinions and Monitor Changes Overtime

It is time to reflect on the opinions of the company’s future prospects that you have formed beginning in Step 15. Take this opportunity to consider what the consensus may be missing, what management may be over/underestimating, or how the financials my be impacted by developments in the long-term. Modeling is like writing a story with numbers. Some analysts are better story tellers than others. This is the point in the modeling process for you to write your own story about the direction you think the company is headed, using your own assumptions, breaking away from the consensus and management’s guidance, particularly in the out years. Also be sure to monitor the changes in the market as the quarter progresses, as management’s guidance could be stale even before the company releases the next quarterly earnings report. The remainder of Step 18 will highlight some additional questions you may want to consider in your forecast, and things to consider as you approach the next earnings release date. As you go through each item, think about what lines would be impacted in your model: revenue, operating margin, operating expenses, etc. Then, change your future period assumptions based on your analysis. Pitfall: Many new research associates feel uncomfortable entering forecasts into a model which deviate from the consensus. Try to remember that the consensus is often incorrect. No one knows for sure what the future holds. The best we can do is use the available information to make an informed guess at what the results may look like. Also, keep in mind that the consensus estimate represents an average of a broad range of possible outcomes. Even if your estimates look a bit high or low relative to the consensus average, it still may be within the range. Step 18a—Has the Competitive Landscape Changed? While all industries have differing degrees of competitive pressure, this is an important consideration in the modeling process. FedEx is no exception. Toward the end of the fourth fiscal quarter the workers’ union at the United Parcel Service (UPS) disclosed that they are preparing for a potential strike which could come during the busy holiday season. If this occurs we will likely have to revise our FedEx fiscal fourth quarter projections upward. For now it is simply a point to watch as the situation develops.

106

Chapter 4: Model Calibration & Forecasting Remember to search for trade reports which may be relevant for the company you are covering. If your company has a product-based business model, watch for new patent filings to help shape your view on the company’s future earnings. If you believe a competitor’s service or product will have a detrimental impact on your company’s market share, then you should reduce your volume assumptions. On the other hand, if you believe your company will gain market share, you should increase your growth rate assumptions. Step 18b—Supply Chain Analysis: Suppliers can offer substantial insight into a company’s performance, particularly if the supplier reports earnings results ahead of the company you are covering. Teardown reports are useful tools to identify the specific suppliers for a company’s inputs. You can also review the 10-K filings of companies within the industry you are covering, as any significant customers must be disclosed to investors. Step 18c—Channel Checks: You may have access to the distribution channel of the company you are covering. For example, if you are building a model for Nike Inc, then consider visiting a sample of shoe retailers and check to see how the latest Nike shoe is doing. It may be out of stock at every store you visit. On the other hand, if it is not well received by consumers, you may find it available everywhere at deep discounts. If you have the resources, you may be able to sample availability and prices of the product across the country, or better yet, globally. This data could prove extremely useful in predicting the next quarter’s results. Step 18d—Have Any Other Factors Changed After the Last Time Management Disclosed Guidance? This item covers a broad range of potential factors, including things like foreign currency rates, interest rates, and the geopolitical backdrop. Try to think of any factors which could affect the next release, and incorporate some form of adjustment to capture an estimate of the financial impact. Wherever possible you will want to ground your adjustment to some datapoint, or have some other reasonable basis (at the very least a directional basis) for the adjustment. Step 18d.1—Leading Indicators: Most Industries have some form of trade publication which track important metrics for the companies within that industry. Sometimes the data is published by a government organization. For example, the U.S. Department of Commerce, Bureau of Economic Analysis publishes industrial production, GDP, and consumer spending data. This is the source used by the FedEx management team in their economic outlook. Given that management has disclosed a correlation with their business in the 10-K, the data could provide insight about the company, and may help forecast earnings ahead of the release. If you are planning to use economic statistics or trade data in your model to help inform your decision of future input estimates, there are a few limitations you should consider. The remaining points in Step 18d.2 through 18d.7 will cover some of these important points. Step 18d.2—Data Revisions: There is always a margin of error involved with forecasting. As a result trade data estimates are almost always revised after the date of the original published report. This can cause issues if you are relying on the originally published data to forecast the next quarter’s earnings, and the revision comes after the quarterly earnings release. Step 18d.3—Reliability of the Data: As with all of the estimates you use in your earnings model, you should take steps to gain comfort over the reliability of the data. One way to do this, is to review the historic estimates relative to actual results. In this review, be sure to make the comparison against the pre-revised version of the estimates to capture the true value in forecasting results with the data. In other words, comparing a revised estimate to actual results does not provide a useful metric for modeling purposes. Step 18d.4—Changes in Data Methodology: It is important to consider how the trade data has changed over time. For example, assume a research firm tracking widget shipments had originally included red, green, and yellow widgets in their shipment estimates. Later the firm changed the tracking methodology to only include red and green widgets. This would make it difficult to draw conclusions when comparing the shipments over time. Usually significant methodology changes are disclosed, but you may need to do some additional research if you find that the correlation of the data and the model input has started to decouple without explanation. Step 18d.5—Timing of Data Release: If the data is typically released after the company you are covering reports quarterly earnings results, then it may not provide much insight for forecasting. Step 18d.6—Number of Observations: It can be difficult to recognize a trend or establish an opinion about the accuracy of the data, if the dataset is limited to a few observations, or if the sample within the population has changed overtime leaving few useful data points for analysis.

107

Step 18d.7—Regression Analysis: You may want to consider regression analysis if you would like to determine the correlation between trade data and an attribute within your model, and use the data point to project an estimate for your future earnings forecast. This will also give you an idea of how accurate the forecast has been over time. For example, if you want to determine the correlation between global GDP and FedEx’s International Express subsegment shipping volume, you could run a regression on the two variables (refer to Appendix 1: Using Regression Analysis to Predict Earnings for details on how to create a regression model). Part of the output from the regression would be an equation to predict the segment’s package volume, which could be incorporated into your earnings model ahead of the earnings release. The regression model would also give you a standard error based on historic results. Regression analysis may be useful if a specific metric for your company is not available. Meaning if there is a correlation between two variables, one could be considered as a proxy for the data point you need for your model. There are many limitations associated with this type of analysis. Perhaps most importantly is the fact that this approach would ignore changes in market share, competition, and other market forces. Be sure to refer to the limitations described in Appendix 1 for additional items to consider. FAQ 4—How do I know if trade data exists for my company? Trade data is not ground breaking information, and does not represent the missing link required to unlock the secret of a company’s future earnings. It is usually reviewed by all analysts covering a company, so it should be relatively easy to find with a simple internet search of general industry information. Step 18e—Consider the “Out-Years”: Up until this point the focus of our forecast has been the next two fiscal years. After that our equations in the blue input cells basically keep the “status-quo” into years three through five, meaning the growth rates, operating margin, and other assumptions, match the average of the preceding year, or consider the seasonal effect of the previous year. This is a reasonable approach since the further out in time we go, the less certain we are in our forecast; however, if you plan to use a Discounted Cash Flow (DCF) valuation approach, you may want to start to normalize the results in the last two years. This is the period which approaches the constant growth stage of the model used to calculate the terminal value component of the valuation. This is a personal preference of mine, and you may disagree, or use your own approach if you prefer. Conceptually the way I think about the DCF is as follows: Assuming the company you are covering will reach a constant growth stage in five years, which is of course an unreasonable assumption; however, as time passes and you add additional years to the end of your model, the forecast years four and five will be continually pushed back. So if we keep the year four and five calibration (meaning growth against years one, two, and three) relatively consistent overtime, and earnings continue to grow, so will the equity valuation through the terminal value calculation. This concept will be discussed further in Chapter 6. For the time being you will need to decide what “normalized earnings” represent for your company. To do this consider the review of the financial statements you performed in Step 13. Which line items stood out as breaking trends with historic results? These items should be adjusted in the last two years of your model as you approach the constant growth stage. The remaining items in Step 18e are a summary of the adjustments I have made to the FedEx model in forecast years four and five. Step 18e.1—Consider the “Out-Years” (Debt): The FedEx debt balance has spiked in recent years as the company took advantage of historically low interest rates, and purchased an annuity contract to retire a portion of the pension obligation (refer to Step 9f for details). To normalize the debt balance I have reduce the average debt-to-equity ratio by 50 basis points per quarter in forecast years four and five. Step 18e.2—Consider the “Out-Years” (Capex): Capital expenditures have been relatively high over the last few years as the company has been investing in shipping hub automation and a new fleet of airplanes. In the out-years I have brought the capex-to-revenue ratio back toward 8%, where it was prior to fiscal year 2015, and set the ratio equal to 8% for the constant growth stage. Step 18e.3—Consider the “Out-Years” (Volume and Yield): Revenue has been increasing significantly as a result of the TNT acquisition and the economic recovery. Overtime economic cycles will peak and trough, and new competitive pressures will be exposed. Based on the company’s history, the long-term revenue growth rate is probably lower than what has been achieved in the last few years. I have reduce the average volume and yield growth rates for each segment in the last two forecast years to reflect this point. 108

Chapter 4: Model Calibration & Forecasting Step 18e.4—Consider the “Out-Years” (Fuel Assumptions): To reflect an expectation of improved fuel efficiency in the new airplane fleet, I have decreased the number gallons of jet fuel used per volume shipped by 5 percentage points starting in 2020. Step 18e.5—Consider the “Out-Years” (Share Repurchases): Share repurchases will also likely decline from the current level of approximately $1B per year, to a much lower amount. I have reduced the dollar purchase amount to approximately $0.4B. Step 18e.6—Consider the “Out-Years” (Dividend Growth): As net income growth begins to stabilize, the dividend growth rate will likely approach the long-term earnings growth rate. I have reduced the dividend growth rate from the 30% rate in fiscal year 2019 down to 15% by forecast year five. Follow Along in the Spreadsheet: Refer to “File 8-Forecast (Steps 13 through 18)” for details on what your model should look like after the forecast estimates from Step 13 through Step 18e.6. Refer to the "How to Use This Textbook" section of Chapter 1 for instructions on how to access the spreadsheet files. Bring it all Together: After you have your forecasts entered in your model, you can write the narrative that explains your numbers rather easily. For example, with our projections for the U.S. Overnight Box sub-segment we could describe our forecast as follows: “For fiscal year 2019 we expect ADV growth to remain muted, as a result of increased competitive pressure which we anticipate will offset a positive U.S. macroeconomic tailwind. Despite the relatively flat volume, we expect revenue for the U.S. Overnight Box business to increase year-over-year on improved yield, which reflects the price increases that went into effect in January of 2018.” If you write similar comments for the other segments, capital expenditures, and the primary expense line items you will have a comprehensive research report in no time.

109

CHAPTER 4 WRAP-UP Takeaways •

The primary difference between the Gutenberg Modeling Framework and the Calibration & Forecasting steps is that the framework deals with choosing the metrics which will be used to develop a future projection, while the Calibration & Forecasting steps determine what values to input for the metrics.

•

There are multiple approaches which can be used to develop an earnings forecast. These approaches typically consider historic trends, the effects of seasonality, and any changes in company specific or broader market conditions.

•

Changes in circumstances may impact the future period forecast differently, depending on the nature of the change (i.e. “big bang” event, gradual structural changes, or non-recurring items).

•

Management’s earnings guidance is among the most important inputs into a model’s forecast. Guidance performance should be back-tested before reliance is placed on the estimates.

•

Some of the items which should be considered in the future period financial statement model include: o Whether or not the competitive landscape has changed, or will change in the future. o Any insight provided by the supply chain. o Any insight provided by distribution channel data. o Changes in market conditions subsequent to the latest guidance provided by management. o Macroeconomic or market leading indicators.

•

Models typically include near-term and longer-term projection stages. It is important to consider the effects both stages will have on earnings and share valuation.

Concept Quiz Instructions: Answer each of the following questions as “true” or “false” 1) A company announces a merger with a competitor during an investor conference call. This is an example of a non-recurring event. 2) Restructurings, acquisitions, and new product launches are all examples of events which would impact the comparability of past and future financial results. 3) Leading indicators have many shortcomings including future revisions, changes in methodology over time, and timing of data releases. 4) Management’s guidance is always the first item to consider in an earnings model forecast. Instructions: Use the following summary to answer questions 5 through 8. Chatty Bird Incorporated is a social media company which generates revenue from advertisements placed on its website. Revenue is earned on a per-click basis, meaning the company only receives a payment when users click on the 110

Chapter 4 Wrap-Up advertisements. Carrie, an equity research associate from Big Bank LLC has created the following model to project Chatty Bird’s earnings for the next four quarters. Columns C through J in Carrie’s model represent the historic periods, and columns K through N represent the future quarterly forecast periods. Carrie has shaded the cells she will use to input assumptions blue.

5) Using only historic seasonality to estimate future results, what is the most appropriate quarter-over-quarter growth rate to enter into cell K15? A. -1.5% B. 2.5% C. 3.0% 6) On Chatty Bird’s last quarterly earnings conference call, management announced a new algorithm which makes advertisements much more relevant to users, and increases the likelihood of clicks by 20%. The new advertising approach will go into effect immediately in the first quarter of 2019. Based on this information and historic trends, what is the most accurate revenue per Monthly Active User (MAU) estimate which should be entered in cell K17? A. $7.00 B. $7.28 C. $9.35 7) What is the most appropriate estimate for the ratio of operating expenses-to-revenue for 4Q2019 in cell N20? A. 32% B. 35% C. 40% 8) On January 15, 2019, in response to data privacy concerns, many U.S. government representatives have been discussing limiting the type of data gathering and use, which makes Chatty Bird’s advertising program’s 20% improvement possible (from question 6). What is the most prudent approach Carrie should take in her model? A. Remove the 20% benefit from the advertising program in her future period forecast. B. Reduce the 20% benefit from the advertising program in her future period forecast. C. Take no action in the model, and mention the developments in her research report.

Concept Quiz Answers 1) False. This is an example of a “big bang” event which will change the earnings profile indefinitely. 2) True. Restructurings, acquisitions, and new product launches are all examples of events which would impact the comparability of past and future financial results.

111

3) True. Leading indicators have many shortcomings including future revisions, changes in methodology over time, and timing of data releases. 4) False. Management’s guidance must first be tested to determine how reliable the data is. Also considering historic trends prior to management’s guidance can be an effective approach to prevent the guidance from causing a bias in your analysis. 5) A. The question references historic trends, so a seasonal effect of an increase in Monthly Average Users (MAUs) during the fourth quarter and decrease in the first quarter should be considered. Since Carrie is using quarterover-quarter growth rates to project future MAUs, this historic 1Q quarter from the first quarter of the previous year is the closest to a correct value. The other two answers would incorporate some form of growth, however the questions states to use historic seasonality only. 6) C. This is a big bang event that will change the comparability of the forecast against historic results. The answer should be somewhere between the historic growth rate of ~15% and 35% (historic growth plus 20% increase in clicks). Answer C is the only answer which incorporates the impact of the new advertising approach. 7) A. Seasonally, the fourth quarter is the strongest quarter and the operating expenses are spread out across a larger revenue base, so the percentage tends to be smaller although the dollar amount of operating expenses has been higher in the fourth quarter. 8) C. Government officials discussing a topic is much different from working on actual legislation. There is not currently enough information available to make a change to the forecast, and the impact is not likely to come in the next four quarters presented in the model. She should however disclose the risk in his report. Solution model for questions 5 through 8.

Follow Along in the Spreadsheet: Refer to “File 19-Chapter Wrap-Ups”, worksheet “Chapter 4” to access the spreadsheet used for the questions above. Refer to the "How to Use This Textbook" section of Chapter 1 for instructions on how to access the spreadsheet files.

112

CHAPTER 5: THE DCF INPUTS (BETA, ERP, CAPM, & WACC) Step 19: Calculate the Equity Risk Premium

Step 20: Derive Beta Using the Regression Function in Excel Step 21: Calculate the Required Return on Equity using Beta and the CAPM Step 22: Calculate the Weighted Average Cost of Capital Chapter 5 Overview: Many investors struggle with the concept of linking risk metrics, particularly volatility, as well as interest rates and general market return expectations, to individual stock valuation. This is due in part to the complexity of valuation techniques which incorporate these measures, relative to a simplistic market multiple-based approach. Adopting a Discounted Cash Flow (DCF) valuation technique is an effective way to link risk and interest rates to equity valuation. This chapter will discuss the approach used in a DCF model which incorporates the change in volatility and interest rates to determine a share value, through the Equity Risk Premium.

Step 19: Calculate the Equity Risk Premium

The market Equity Risk Premium (ERP) is a measure of the total return an investor requires in excess of the risk-free rate (typically a U.S. Treasury security), as compensation for the additional risk of an equity investment. There is a direct relationship between the ERP and required return, which means that as an investment’s risk increases, investors will expect a higher return. Conversely, as risk decreases the required return on equity will also decrease. There are many different approaches to measure the market’s ERP. Each method has different assumptions, pros, cons, and results. I use the Constant Sharpe approach. Step 21 will demonstrate this method. Note that for this exercise I will use the S&P500 Index as a proxy for the equity market return. Follow Along in the Spreadsheet: Refer to “File 16-Equity Risk Premium Model (Step 19)” for details of the calculations discussed in this Chapter, and to follow along as the approach is developed. Refer to the "How to Use This Textbook" section of Chapter 1 for instructions on how to access the spreadsheet files. Step 19a—Calculate the Historic Sharpe Ratio: There are two primary steps in the calculation of the Constant Sharpe ERP. First, is a calculation of the market’s historic average Sharpe ratio, which is the average S&P500 total return, less the risk-free rate (I use the 10-year U.S. Treasury rate), divided by the standard deviation of S&P500 returns. In Exhibit 102 below, you can see I used data through 1963 to calculate the excess return over the risk-free rate each year (refer to column E), then took the average (cell E65), and divided it by the standard deviation in returns (cell E66), to arrive at the Constant Sharpe Ratio of 0.325 for the index (cell E67).

113

Exhibit 102—Constant Sharpe Ratio Calculation

Step 19b—Calculate the Current ERP: The second step in the Constant Sharpe-based ERP calculation, is the application of volatility to the Constant Sharpe Ratio. The volatility measure for the S&P500 is the Chicago Board Options Exchange (CBOE) volatility index known as the VIX (refer to Exhibit 102 above, column H). The VIX is a measure of implied shortterm volatility of S&P500 options, quoted in percentage terms. As market uncertainty increases, the VIX increases, and equity valuations fall. When market uncertainty decreases, the VIX decreases, and equity valuations rise. To link the ERP to market volatility, multiply the Constant Sharpe by the VIX. Using the Constant Sharpe ratio of 0.325 calculated above, and the current average VIX index value of 13.96% (12-month trailing average as of the third quarter of 2018 in cell J16). The resulting estimate for the ERP is 4.5% (refer to cell R16). Notice that I use the trailing 12-month average VIX in my Constant Sharpe estimate. I do this for two reasons: 1) Long-Term Investment Horizon: When I am analyzing investments, I am typically trying to estimate a 12-month target share price, with the intention of holding the investment for at least one year. If I used a daily, or even quarterly average VIX in my Constant Sharpe ERP estimate, my target share valuation would fluctuate substantially, which would not produce much value from an analytical standpoint. 2) Match Volatility with Rates of Return: Since I am using a one-year S&P500 return horizon, and a one-year average rate of the 10-year U.S. Treasury security in my equation, I should attempt to match the horizon with my volatility profile.

114

Chapter 5: The DCF Inputs (Beta, ERP, CAPM, & WACC) Step 19c—Review of the Historic Constant Sharpe ERP: The chart below plots the Constant Sharpe-based ERP estimate through history. Two periods stand out when the ERP increased well above the long-term average for a sustained period of time. Both were driven by events which triggered periods of greater financial uncertainty. The first spike began after the tech bubble burst in the late 1990s. Then as the ERP began to approach the normal average for a short period of time, it peaked again after the September 11, 2001 terrorist attacks in New York. The second period occurred at the end of 2008 as a result of the U.S. recession and credit crisis. The increased equity volatility during these periods drove higher discount rates, which resulted in lower net present value of cash flows (since the discount rate is included in the denominator of the discounting equation), and lower stock valuations to reflect the heightened level of risk. Note that in the chart below the peaks come shortly after the events have occurred. This is due to the fact that I have used the trailing 12-month VIX in my ERP estimate. This measure also smooths out the peak and troughs which would exist if we used a shorter period for the VIX input. If your investment horizon is shorter than one year, you may prefer to use the quarterly average VIX. If you want the ERP to align exactly with the historic stress events, then use the daily VIX, but keep in mind this will produce extreme fluctuations in your share valuation projections when you incorporate the ERP into your cash flow discount rate through the required return on equity component of the weighted average cost of total capital.

Exhibit 103—Historic Constant Sharpe-Based ERP Estimate

Step 19d—Forecasting the ERP: Since share valuation is a forward looking metric, you may want to forecast the future ERP based on your market expectations, and incorporate this forecast into your share valuation. To do this, you will need to develop forecasts for the following blue input cells in the ERP model as shown in Exhibit 102 above: 1) The Federal Funds rate, which is the rate for deposits at the Federal Reserve, lent overnight from one bank to another (cells K7 through K15), 2) The spread between the Federal Funds rate and the 10-year U.S. Treasury Rate (cells L7 through L15), 3) Volatility (refer to cells H7 through H15), and 4) Equity market return projections (cells P7 through P15). The remainder of Step 21 will demonstrate how to input these assumptions into the ERP model, and calculate a future ERP estimate. Step 19d.1—Forecasting the ERP (The Federal Funds Rate): To develop an estimate for the Federal Funds rate, start by reviewing the expectation of the Federal Reserve’s Federal Open Market Committee (FOMC) members. The FOMC is made up of 12 members who meet 8 times a year (or more often if warranted by economic conditions). The members are the individuals who set the rate, so their forecasts are critical to your interest rate estimates.

115

The Monetary Policy section of the Federal Reserve’s website (www.federalreserve.gov/monetarypolicy) offers most of the data you need for your analysis. There are three key items to download: 1) the FOMC statement, 2) the FOMC meeting minutes, and 3) the FOMC members’ median Federal Funds estimate for the year. Start your analysis with the FOMC Statement. Analysts tend to dissect each word in the document, looking for any change from one meeting to the next, which could indicate that the Committee is shifting its view on rates. The Federal Reserve’s dual mandate is to maintain stable prices by controlling inflation, and to maximize employment. The first two paragraphs of the statement typically address these objectives. The following paragraph is where the interest rate decision is disclosed. In the example below from the September 26, 2017 meeting, the Committee decided to increase the Federal Funds target rate to a range of 2% and 2.25% (refer to Exhibit 104 below).

Exhibit 104—Example FOMC Statement

Source: FederalReserve.gov, FOMC Statement, September 26, 2018, retrieved September 24, 2018.

At the time this material was first published (refer to the “Timing of Publication” section of Chapter 1 for details on the reporting dates used in this textbook, updates are posted in the Appendix) the FOMC had increased the Federal Funds target rate eight times, by 25 basis points each occurrence, from the target rate range of 0% to 0.25% established after the credit crisis. The Committee will have two additional meetings in 2018 on November 7, 2018 and December 18, 2018. The following section is a summary of each of the meeting statements, starting with the last meeting when interest rates were held flat on October 28, 2015, and including each of the meetings since, which resulted in a change in interest rates through September 26, 2018.

116

Chapter 5: The DCF Inputs (Beta, ERP, CAPM, & WACC) Notice how the phrases used by the committee change from meeting to the next. I have underlined some of the points I find particularly important in the context of the Fed’s dual mandate. It is also important to keep track of which members vote against the FOMC decision. Reviewing this history will help you to develop an understanding of how to read the Fed’s statements, and assess for yourself whether or not you should incorporate future interest rate changes into your model, based on the language used in the last FOMC statement.

1) Statement from October 28, 2015: No change in the target rate of between 0% and 0.25% FOMC Committee Statement: Information received since the Federal Open Market Committee met in September suggests that economic activity has been expanding at a moderate pace. Household spending and business fixed investment have been increasing at solid rates in recent months, and the housing sector has improved further; however, net exports have been soft. The pace of job gains slowed and the unemployment rate held steady. Nonetheless, labor market indicators, on balance, show that underutilization of labor resources has diminished since early this year. Inflation has continued to run below the Committee's longer-run objective, partly reflecting declines in energy prices and in prices of non-energy imports. Market-based measures of inflation compensation moved slightly lower; survey-based measures of longer-term inflation expectations have remained stable. Consistent with its statutory mandate, the Committee seeks to foster maximum employment and price stability. The Committee expects that, with appropriate policy accommodation, economic activity will expand at a moderate pace, with labor market indicators continuing to move toward levels the Committee judges consistent with its dual mandate. The Committee continues to see the risks to the outlook for economic activity and the labor market as nearly balanced but is monitoring global economic and financial developments. Inflation is anticipated to remain near its recent low level in the near term but the Committee expects inflation to rise gradually toward 2 percent over the medium term as the labor market improves further and the transitory effects of declines in energy and import prices dissipate. The Committee continues to monitor inflation developments closely. To support continued progress toward maximum employment and price stability, the Committee today reaffirmed its view that the current 0 to 1/4 percent target range for the Federal Funds rate remains appropriate. In determining whether it will be appropriate to raise the target range at its next meeting, the Committee will assess progress both realized and expected toward its objectives of maximum employment and 2 percent inflation. This assessment will take into account a wide range of information, including measures of labor market conditions, indicators of inflation pressures and inflation expectations, and readings on financial and international developments. The Committee anticipates that it will be appropriate to raise the target range for the Federal Funds rate when it has seen some further improvement in the labor market and is reasonably confident that inflation will move back to its 2 percent objective over the medium term. Voting for the FOMC monetary policy action were: Janet L. Yellen, Chair; William C. Dudley, Vice Chairman; Lael Brainard; Charles L. Evans; Stanley Fischer; Dennis P. Lockhart; Jerome H. Powell; Daniel K. Tarullo; and John C. Williams. Voting against the action was Jeffrey M. Lacker, who preferred to raise the target range for the Federal Funds rate by 25 basis points at this meeting. Source: FederalReserve.gov, FOMC Statement, October 28, 2015, retrieved September 24, 2018. Notice in the December 16, 2015 meeting that the decision to raise rates was unanimous, and the committee changed its view on the risk to the economic outlook from “nearly balanced” to “balanced”.

2) Statement from December 16, 2015: Raised target range to between 0.25% to 0.50% FOMC Committee Statement: Information received since the Federal Open Market Committee met in October suggests that economic activity has been expanding at a moderate pace. Household spending and business fixed investment have been increasing at solid rates in recent months, and the housing sector has improved further; however, net exports have been soft. A range of recent labor market indicators, including ongoing job gains and declining unemployment, shows further improvement and confirms that underutilization of labor resources has diminished appreciably since early this year. Inflation has continued to run below the Committee's 2 percent longer-run objective, partly reflecting declines in energy prices and in prices of non-energy imports. Market-based measures of inflation compensation remain low; some survey-based measures of longer-term inflation expectations have edged down. 117

Consistent with its statutory mandate, the Committee seeks to foster maximum employment and price stability. The Committee currently expects that, with gradual adjustments in the stance of monetary policy, economic activity will continue to expand at a moderate pace and labor market indicators will continue to strengthen. Overall, taking into account domestic and international developments, the Committee sees the risks to the outlook for both economic activity and the labor market as balanced. Inflation is expected to rise to 2 percent over the medium term as the transitory effects of declines in energy and import prices dissipate and the labor market strengthens further. The Committee continues to monitor inflation developments closely. The Committee judges that there has been considerable improvement in labor market conditions this year, and it is reasonably confident that inflation will rise, over the medium term, to its 2 percent objective. Given the economic outlook, and recognizing the time it takes for policy actions to affect future economic outcomes, the Committee decided to raise the target range for the Federal Funds rate to 1/4 to 1/2 percent. The stance of monetary policy remains accommodative after this increase, thereby supporting further improvement in labor market conditions and a return to 2 percent inflation. In determining the timing and size of future adjustments to the target range for the Federal Funds rate, the Committee will assess realized and expected economic conditions relative to its objectives of maximum employment and 2 percent inflation. This assessment will take into account a wide range of information, including measures of labor market conditions, indicators of inflation pressures and inflation expectations, and readings on financial and international developments. In light of the current shortfall of inflation from 2 percent, the Committee will carefully monitor actual and expected progress toward its inflation goal. The Committee expects that economic conditions will evolve in a manner that will warrant only gradual increases in the Federal Funds rate; the Federal Funds rate is likely to remain, for some time, below levels that are expected to prevail in the longer run. However, the actual path of the Federal Funds rate will depend on the economic outlook as informed by incoming data. The Committee is maintaining its existing policy of reinvesting principal payments from its holdings of agency debt and agency mortgage-backed securities in agency mortgage-backed securities and of rolling over maturing Treasury securities at auction, and it anticipates doing so until normalization of the level of the Federal Funds rate is well under way. This policy, by keeping the Committee's holdings of longer-term securities at sizable levels, should help maintain accommodative financial conditions. Voting for the FOMC monetary policy action were: Janet L. Yellen, Chair; William C. Dudley, Vice Chairman; Lael Brainard; Charles L. Evans; Stanley Fischer; Jeffrey M. Lacker; Dennis P. Lockhart; Jerome H. Powell; Daniel K. Tarullo; and John C. Williams. Source: FederalReserve.gov, FOMC Statement, December 16, 2015, retrieved September 24, 2018. In the December 15, 2016 meeting the comment about “under utilization of labor resources” was removed as the employment environment improved. The decision to raise rates was unanimous.

3) Statement from December 15, 2016: Raised target range to between 0.50% to 0.75% FOMC Committee Statement: FOMC Committee Statement: Information received since the Federal Open Market Committee met in November indicates that the labor market has continued to strengthen and that economic activity has been expanding at a moderate pace since mid-year. Job gains have been solid in recent months and the unemployment rate has declined. Household spending has been rising moderately but business fixed investment has remained soft. Inflation has increased since earlier this year but is still below the Committee's 2 percent longer-run objective, partly reflecting earlier declines in energy prices and in prices of non-energy imports. Market-based measures of inflation compensation have moved up considerably but still are low; most survey-based measures of longer-term inflation expectations are little changed, on balance, in recent months. Consistent with its statutory mandate, the Committee seeks to foster maximum employment and price stability. The Committee expects that, with gradual adjustments in the stance of monetary policy, economic activity will expand at a moderate pace and labor market conditions will strengthen somewhat further. Inflation is expected to rise to 2 percent over the medium term as the transitory effects of past declines in energy and import prices dissipate and the labor market strengthens further. Near-term risks to the economic outlook appear roughly balanced. The Committee continues to closely monitor inflation indicators and global economic and financial developments. 118

Chapter 5: The DCF Inputs (Beta, ERP, CAPM, & WACC) In view of realized and expected labor market conditions and inflation, the Committee decided to raise the target range for the Federal Funds rate to 1/2 to 3/4 percent. The stance of monetary policy remains accommodative, thereby supporting some further strengthening in labor market conditions and a return to 2 percent inflation. In determining the timing and size of future adjustments to the target range for the Federal Funds rate, the Committee will assess realized and expected economic conditions relative to its objectives of maximum employment and 2 percent inflation. This assessment will take into account a wide range of information, including measures of labor market conditions, indicators of inflation pressures and inflation expectations, and readings on financial and international developments. In light of the current shortfall of inflation from 2 percent, the Committee will carefully monitor actual and expected progress toward its inflation goal. The Committee expects that economic conditions will evolve in a manner that will warrant only gradual increases in the Federal Funds rate; the Federal Funds rate is likely to remain, for some time, below levels that are expected to prevail in the longer run. However, the actual path of the Federal Funds rate will depend on the economic outlook as informed by incoming data. The Committee is maintaining its existing policy of reinvesting principal payments from its holdings of agency debt and agency mortgage-backed securities in agency mortgage-backed securities and of rolling over maturing Treasury securities at auction, and it anticipates doing so until normalization of the level of the Federal Funds rate is well under way. This policy, by keeping the Committee's holdings of longer-term securities at sizable levels, should help maintain accommodative financial conditions. Voting for the FOMC monetary policy action were: Janet L. Yellen, Chair; William C. Dudley, Vice Chairman; Lael Brainard; James Bullard; Stanley Fischer; Esther L. George; Loretta J. Mester; Jerome H. Powell; Eric Rosengren; and Daniel K. Tarullo. Source: FederalReserve.gov, FOMC Statement, December 15, 2016, retrieved September 24, 2018. In the March 15, 2017 meeting we see the first dissenting vote against the increase in interest rates.

4) Statement from March 15, 2017: Raised target range to between 0.75% to 1.00% FOMC Committee Statement: Information received since the Federal Open Market Committee met in February indicates that the labor market has continued to strengthen and that economic activity has continued to expand at a moderate pace. Job gains remained solid and the unemployment rate was little changed in recent months. Household spending has continued to rise moderately while business fixed investment appears to have firmed somewhat. Inflation has increased in recent quarters, moving close to the Committee's 2 percent longer-run objective; excluding energy and food prices, inflation was little changed and continued to run somewhat below 2 percent. Market-based measures of inflation compensation remain low; survey-based measures of longer-term inflation expectations are little changed, on balance. Consistent with its statutory mandate, the Committee seeks to foster maximum employment and price stability. The Committee expects that, with gradual adjustments in the stance of monetary policy, economic activity will expand at a moderate pace, labor market conditions will strengthen somewhat further, and inflation will stabilize around 2 percent over the medium term. Near-term risks to the economic outlook appear roughly balanced. The Committee continues to closely monitor inflation indicators and global economic and financial developments. In view of realized and expected labor market conditions and inflation, the Committee decided to raise the target range for the Federal Funds rate to 3/4 to 1 percent. The stance of monetary policy remains accommodative, thereby supporting some further strengthening in labor market conditions and a sustained return to 2 percent inflation. In determining the timing and size of future adjustments to the target range for the Federal Funds rate, the Committee will assess realized and expected economic conditions relative to its objectives of maximum employment and 2 percent inflation. This assessment will take into account a wide range of information, including measures of labor market conditions, indicators of inflation pressures and inflation expectations, and readings on financial and international developments. The Committee will carefully monitor actual and expected inflation developments relative to its symmetric inflation goal. The Committee expects that economic conditions will evolve in a manner that will warrant gradual increases in the Federal Funds rate; the Federal Funds rate is likely to remain, for some time, below levels that are expected to prevail in the longer run. However, the actual path of the Federal Funds rate will depend on the economic outlook as informed by incoming data.

119

The Committee is maintaining its existing policy of reinvesting principal payments from its holdings of agency debt and agency mortgage-backed securities in agency mortgage-backed securities and of rolling over maturing Treasury securities at auction, and it anticipates doing so until normalization of the level of the Federal Funds rate is well under way. This policy, by keeping the Committee's holdings of longer-term securities at sizable levels, should help maintain accommodative financial conditions. Voting for the FOMC monetary policy action were: Janet L. Yellen, Chair; William C. Dudley, Vice Chairman; Lael Brainard; Charles L. Evans; Stanley Fischer; Patrick Harker; Robert S. Kaplan; Jerome H. Powell; and Daniel K. Tarullo. Voting against the action was Neel Kashkari, who preferred at this meeting to maintain the existing target range for the Federal Funds rate. Source: FederalReserve.gov, FOMC Statement, March 15, 2017, retrieved September 24, 2018. In the June 14, 2017 meeting Neel Kashkari again votes against the increase in interest rates. Comments were added with regards to the Balance Sheet normalization program. This is a response to the quantitative easing strategy the Fed put into effect during the credit crisis, where bonds were purchased to increase liquidity and stimulate the economy. It is difficult to project what the impact of the shrinking Balance Sheet will now have on interest rates. In our model we will include the estimate in the spread between the Federal Funds rate and the 10-year U.S. Treasury security, monitor our spread estimates over time, and adjust as necessary.

5) Statement from June 14, 2017: Raised target range to between 1.00% to 1.25% FOMC Committee Statement: Information received since the Federal Open Market Committee met in May indicates that the labor market has continued to strengthen and that economic activity has been rising moderately so far this year. Job gains have moderated but have been solid, on average, since the beginning of the year, and the unemployment rate has declined. Household spending has picked up in recent months, and business fixed investment has continued to expand. On a 12-month basis, inflation has declined recently and, like the measure excluding food and energy prices, is running somewhat below 2 percent. Market-based measures of inflation compensation remain low; survey-based measures of longer-term inflation expectations are little changed, on balance. Consistent with its statutory mandate, the Committee seeks to foster maximum employment and price stability. The Committee continues to expect that, with gradual adjustments in the stance of monetary policy, economic activity will expand at a moderate pace, and labor market conditions will strengthen somewhat further. Inflation on a 12-month basis is expected to remain somewhat below 2 percent in the near term but to stabilize around the Committee's 2 percent objective over the medium term. Near-term risks to the economic outlook appear roughly balanced, but the Committee is monitoring inflation developments closely. In view of realized and expected labor market conditions and inflation, the Committee decided to raise the target range for the Federal Funds rate to 1 to 1-1/4 percent. The stance of monetary policy remains accommodative, thereby supporting some further strengthening in labor market conditions and a sustained return to 2 percent inflation. In determining the timing and size of future adjustments to the target range for the Federal Funds rate, the Committee will assess realized and expected economic conditions relative to its objectives of maximum employment and 2 percent inflation. This assessment will take into account a wide range of information, including measures of labor market conditions, indicators of inflation pressures and inflation expectations, and readings on financial and international developments. The Committee will carefully monitor actual and expected inflation developments relative to its symmetric inflation goal. The Committee expects that economic conditions will evolve in a manner that will warrant gradual increases in the Federal Funds rate; the Federal Funds rate is likely to remain, for some time, below levels that are expected to prevail in the longer run. However, the actual path of the Federal Funds rate will depend on the economic outlook as informed by incoming data. The Committee is maintaining its existing policy of reinvesting principal payments from its holdings of agency debt and agency mortgage-backed securities in agency mortgage-backed securities and of rolling over maturing Treasury securities at auction. The Committee currently expects to begin implementing a Balance Sheet normalization program this year, provided that the economy evolves broadly as anticipated. This program, which would gradually reduce the Federal Reserve's securities holdings by decreasing reinvestment of principal payments from those securities, is described in the accompanying addendum to the Committee's Policy Normalization Principles and Plans.

120

Chapter 5: The DCF Inputs (Beta, ERP, CAPM, & WACC) Voting for the FOMC monetary policy action were: Janet L. Yellen, Chair; William C. Dudley, Vice Chairman; Lael Brainard; Charles L. Evans; Stanley Fischer; Patrick Harker; Robert S. Kaplan; and Jerome H. Powell. Voting against the action was Neel Kashkari, who preferred at this meeting to maintain the existing target range for the Federal Funds rate. Source: FederalReserve.gov, FOMC Statement, June 14, 2017, retrieved September 24, 2018. Notice in the December 13, 2017 meeting the language around economic activity has shifted from “moderate” to “solid”. Also there are now two dissenting members against the rate increases.

6) Statement from December 13, 2017: Raised target range to between 1.25% to 1.50% FOMC Committee Statement: Information received since the Federal Open Market Committee met in November indicates that the labor market has continued to strengthen and that economic activity has been rising at a solid rate. Averaging through hurricane-related fluctuations, job gains have been solid, and the unemployment rate declined further. Household spending has been expanding at a moderate rate, and growth in business fixed investment has picked up in recent quarters. On a 12-month basis, both overall inflation and inflation for items other than food and energy have declined this year and are running below 2 percent. Market-based measures of inflation compensation remain low; survey-based measures of longer-term inflation expectations are little changed, on balance. Consistent with its statutory mandate, the Committee seeks to foster maximum employment and price stability. Hurricane-related disruptions and rebuilding have affected economic activity, employment, and inflation in recent months but have not materially altered the outlook for the national economy. Consequently, the Committee continues to expect that, with gradual adjustments in the stance of monetary policy, economic activity will expand at a moderate pace and labor market conditions will remain strong. Inflation on a 12‑month basis is expected to remain somewhat below 2 percent in the near term but to stabilize around the Committee's 2 percent objective over the medium term. Near-term risks to the economic outlook appear roughly balanced, but the Committee is monitoring inflation developments closely. In view of realized and expected labor market conditions and inflation, the Committee decided to raise the target range for the Federal Funds rate to 1-1/4 to 1‑1/2 percent. The stance of monetary policy remains accommodative, thereby supporting strong labor market conditions and a sustained return to 2 percent inflation.

In determining the timing and size of future adjustments to the target range for the Federal Funds rate, the Committee will assess realized and expected economic conditions relative to its objectives of maximum employment and 2 percent inflation. This assessment will take into account a wide range of information, including measures of labor market conditions, indicators of inflation pressures and inflation expectations, and readings on financial and international developments. The Committee will carefully monitor actual and expected inflation developments relative to its symmetric inflation goal. The Committee expects that economic conditions will evolve in a manner that will warrant gradual increases in the Federal Funds rate; the Federal Funds rate is likely to remain, for some time, below levels that are expected to prevail in the longer run. However, the actual path of the Federal Funds rate will depend on the economic outlook as informed by incoming data. Voting for the FOMC monetary policy action were Janet L. Yellen, Chair; William C. Dudley, Vice Chairman; Lael Brainard; Patrick Harker; Robert S. Kaplan; Jerome H. Powell; and Randal K. Quarles. Voting against the action were Charles L. Evans and Neel Kashkari, who preferred at this meeting to maintain the existing target range for the Federal Funds rate. Source: FederalReserve.gov, FOMC Statement, December 13, 2017, retrieved September 24, 2018. In the March 21, 2018 meeting the voting decision shifted to a unanimous result. The language around economic activity also changed back to “moderate” from “solid”.

7) Statement from March 21, 2018: Raised target range to between 1.50% to 1.75% FOMC Committee Statement: Information received since the Federal Open Market Committee met in January indicates that the labor market has continued to strengthen and that economic activity has been rising at a moderate rate. Job gains have been strong in recent months, and the unemployment rate has stayed low. Recent data suggest that growth rates of household spending and business fixed investment have moderated from their strong fourth-quarter readings. On a 12-month basis, both overall inflation and inflation for items other than food and energy have continued to run 121

below 2 percent. Market-based measures of inflation compensation have increased in recent months but remain low; survey-based measures of longer-term inflation expectations are little changed, on balance. Consistent with its statutory mandate, the Committee seeks to foster maximum employment and price stability. The economic outlook has strengthened in recent months. The Committee expects that, with further gradual adjustments in the stance of monetary policy, economic activity will expand at a moderate pace in the medium term and labor market conditions will remain strong. Inflation on a 12-month basis is expected to move up in coming months and to stabilize around the Committee's 2 percent objective over the medium term. Near-term risks to the economic outlook appear roughly balanced, but the Committee is monitoring inflation developments closely. In view of realized and expected labor market conditions and inflation, the Committee decided to raise the target range for the Federal Funds rate to 1-1/2 to 1-3/4 percent. The stance of monetary policy remains accommodative, thereby supporting strong labor market conditions and a sustained return to 2 percent inflation. In determining the timing and size of future adjustments to the target range for the Federal Funds rate, the Committee will assess realized and expected economic conditions relative to its objectives of maximum employment and 2 percent inflation. This assessment will take into account a wide range of information, including measures of labor market conditions, indicators of inflation pressures and inflation expectations, and readings on financial and international developments. The Committee will carefully monitor actual and expected inflation developments relative to its symmetric inflation goal. The Committee expects that economic conditions will evolve in a manner that will warrant further gradual increases in the Federal Funds rate; the Federal Funds rate is likely to remain, for some time, below levels that are expected to prevail in the longer run. However, the actual path of the Federal Funds rate will depend on the economic outlook as informed by incoming data. Voting for the FOMC monetary policy action were Jerome H. Powell, Chairman; William C. Dudley, Vice Chairman; Thomas I. Barkin; Raphael W. Bostic; Lael Brainard; Loretta J. Mester; Randal K. Quarles; and John C. Williams. Source: FederalReserve.gov, FOMC Statement, March 21, 2018, retrieved September 24, 2018. In the June 13, 2018 meeting the voting decision remained unanimous. The language around economic activity again shifted back to “solid” from “moderate”.

8) Statement from June 13, 2018: Raised target range to between 1.75% to 2.00% FOMC Committee Statement: Information received since the Federal Open Market Committee met in May indicates that the labor market has continued to strengthen and that economic activity has been rising at a solid rate. Job gains have been strong, on average, in recent months, and the unemployment rate has declined. Recent data suggest that growth of household spending has picked up, while business fixed investment has continued to grow strongly. On a 12month basis, both overall inflation and inflation for items other than food and energy have moved close to 2 percent. Indicators of longer-term inflation expectations are little changed, on balance. Consistent with its statutory mandate, the Committee seeks to foster maximum employment and price stability. The Committee expects that further gradual increases in the target range for the Federal Funds rate will be consistent with sustained expansion of economic activity, strong labor market conditions, and inflation near the Committee's symmetric 2 percent objective over the medium term. Risks to the economic outlook appear roughly balanced. In view of realized and expected labor market conditions and inflation, the Committee decided to raise the target range for the Federal Funds rate to 1-3/4 to 2 percent. The stance of monetary policy remains accommodative, thereby supporting strong labor market conditions and a sustained return to 2 percent inflation. In determining the timing and size of future adjustments to the target range for the Federal Funds rate, the Committee will assess realized and expected economic conditions relative to its maximum employment objective and its symmetric 2 percent inflation objective. This assessment will take into account a wide range of information, including measures of labor market conditions, indicators of inflation pressures and inflation expectations, and readings on financial and international developments. Voting for the FOMC monetary policy action were Jerome H. Powell, Chairman; William C. Dudley, Vice Chairman; Thomas I. Barkin; Raphael W. Bostic; Lael Brainard; Loretta J. Mester; Randal K. Quarles; and John C. Williams.

122

Chapter 5: The DCF Inputs (Beta, ERP, CAPM, & WACC) Source: FederalReserve.gov, FOMC Statement, June 13, 2018, retrieved September 24, 2018.

9) Statement from September 26, 2018: Raised target range to between 2.00% to 2.25% FOMC Committee Statement: Information received since the Federal Open Market Committee met in August indicates that the labor market has continued to strengthen and that economic activity has been rising at a strong rate. Job gains have been strong, on average, in recent months, and the unemployment rate has stayed low. Household spending and business fixed investment have grown strongly. On a 12-month basis, both overall inflation and inflation for items other than food and energy remain near 2 percent. Indicators of longer-term inflation expectations are little changed, on balance. Consistent with its statutory mandate, the Committee seeks to foster maximum employment and price stability. The Committee expects that further gradual increases in the target range for the Federal Funds rate will be consistent with sustained expansion of economic activity, strong labor market conditions, and inflation near the Committee's symmetric 2 percent objective over the medium term. Risks to the economic outlook appear roughly balanced. In view of realized and expected labor market conditions and inflation, the Committee decided to raise the target range for the Federal Funds rate to 2 to 2-1/4 percent. In determining the timing and size of future adjustments to the target range for the Federal Funds rate, the Committee will assess realized and expected economic conditions relative to its maximum employment objective and its symmetric 2 percent inflation objective. This assessment will take into account a wide range of information, including measures of labor market conditions, indicators of inflation pressures and inflation expectations, and readings on financial and international developments. Voting for the FOMC monetary policy action were: Jerome H. Powell, Chairman; John C. Williams, Vice Chairman; Thomas I. Barkin; Raphael W. Bostic; Lael Brainard; Richard H. Clarida; Esther L. George; Loretta J. Mester; and Randal K. Quarles. Source: FederalReserve.gov, FOMC Statement, September 26, 2018, retrieved September 26, 2018. After you are comfortable with the language in the FOMC statements, the next step is to read through the FOMC meeting minutes. This is a much longer document compared to the FOMC statement, typically about 20 pages. The minutes describe all of the economic metrics discussed during the meeting, how they have changed, how the committee members interpret the results, and whether or not the members are in agreement about FOMC’s course of action. You can access the Projection Materials within the minutes, or with the FOMC Statement, which is released before the meeting minutes are available. The Projection Materials include the various macroeconomic indicators which the Committee uses to set the target rate, so this should be your primary focus for estimating the future Federal Funds rate. I typically do not deviate from the median committee members’ projections, although, you may decide to if you have an opposing view. Besides the Federal Funds rate, there are four key macroeconomic variable projections published: 1) GDP, 2) unemployment, 3) Personal Consumptions Expenditure (PCE) inflation, and 4) core inflation (excluding food and energy). If you watch these four variables over time, you can develop an expectation for the FOMC’s actions. For example, if you notice that GDP has increased above the FOMC’s median forecast for 2019 of 2.5% (refer to the September 26, 2018 Projection Material example in Exhibit 105 below), the unemployment rate has decreased below the median projection of 3.5%, and inflation/core inflation have increased above the 2.1% forecast, then you may expect the FOMC to increase interest rates at the next Committee meeting.

123

Exhibit 105—FOMC Projection Materials Example

Source: FederalReserve.gov, FOMC Projections Materials, September 26, 2018, retrieved September 26, 2018.

Now, to derive the expected increase in the Federal Funds rate from the Projection Materials, simply take the Median Federal Funds projection and subtract it from the current midpoint of the Federal Funds target rate: 2.4% minus 2.125% = 0.275%. I usually divide the increases evenly for the remaining quarters of the year. Since the FOMC will meet two additional times in 2018 I have assumed that they will raise by 25 basis points in one of these meetings. I enter this increase in 4Q2018E (refer to Exhibit 102, cell K15). The Median Federal Funds target rate for 2019 is 3.1% which is 0.7% higher than the 2018 forecast of 2.4%. This means the Fed expects to raise rates approximately three times in 2019. In my model I have spread these increases to be three25 basis point increases in the: first quarter of 2019 (refer to Exhibit 102 cell K14), the second quarter of 2019 (refer to Exhibit 102 cell K13), and the third quarter of 2019 (refer to Exhibit 102 cell K12). In the FOMC projection material the Committee’s forecast for 2020 increases 30 basis points to 3.4%. I have incorporated one 25 basis point increase in the first quarter of 2020 (refer to Exhibit 102 cell K10). Before moving on take a step back and review your Federal Funds target forecast in Exhibit 102 cells K8 through K16. Note that in my version of the model, the rates do not align exactly to the FOMC Projection Material. This is driven by three factors: 1) rounding of the decimal places, 2) the fact that the Federal Funds target rate is a range, and 3) I am using an average rate over the quarter which takes some time for the average to reach the rate at the end of the period. Alternative Approach to forecasting the Federal Funds Rate (Futures): Federal Fund futures are actively traded on the Chicago Mercantile Exchange (CME). You could look to the futures contracts to determine the probability of a future rate change. To do this simply take the Federal Funds future price for the most recent month and subtract it from 100. For example, assume that it is July 1st, the current Federal Funds rate is 2.40%, and the FOMC will meet in the middle of this month. If the July 31st Federal Funds future contract price is 97.85, this would imply an expected Federal Funds rate of 2.15%, or 100 – 97.85. The current Federal Funds rate in this example is 25 basis points higher compared to the rate implied by the futures contract (2.40% versus 2.15%). This implies that the market expects a 25 basis point decrease in the target Federal Funds rate at the next FOMC meeting.

124

Chapter 5: The DCF Inputs (Beta, ERP, CAPM, & WACC) To calculate a rough estimate of the implied probability of a rate change, take the difference between the rate implied by the futures contract and the current rate, and divide it by the difference in the expected rate if the FOMC were to change the rate, from the current rate. In our example this would be (2.15% - 2.40%) ÷ (2.15% - 2.40%) = 100 percent probability of a rate cut based on the market’s pricing of futures contracts. This is a simplified approximation. The CME publishes a probability distribution of rate increases based on future prices which is far more precise. The projection tool is available on the CME website (https://www.cmegroup.com/trading/interest-rates/countdown-to-fomc.html). Alternative Approach to forecasting the Federal Funds Rate (NY Fed Survey): The New York Federal Reserve publishes a survey of market participant expectations for future Federal Funds rates. The survey results could be used in conjunction with your analysis of the FOMC Statement, and futures prices to help form an expectation of future interest rates. The results of the market participant survey are available on the New York Federal Reserve Bank’s website (https://www.newyorkfed.org/markets/survey_market_participants). Why Put so Much Effort Into Watching the Fed? Let’s revisit the Fed’s dual mandate to control inflation and maximize employment, and the impact these factors have on equity valuation. The economic indicator charts below plot GDP, equity valuations, volatility, interest rates, inflation, and unemployment through the last recession. As demonstrated in the charts, during periods of economic stress, equity valuations decline significantly. In fact, equities lost nearly half of their value from peak to trough. The potential impact of such a downside scenario should show up in our modeling efforts through lower earnings and cash flow projections (a result of lower economic activity, measured by GDP), and higher discount rates driven by increased volatility, which offset the interest rate decreases that typically occur during a recession. Alternatively, if you are not using a DCF-based approach, you should see the effect from lower earnings estimates and lower Price/Earnings (P/E) multiples. The difficulty with incorporating a recession into a model is that the decline in equity prices during recessions are extreme, yet before a recession begins, during stable economic periods, the market and projection models only incorporate a low probability of recession. This is one of the key drivers of why sell-side price targets deteriorate so rapidly when the economy starts to show signs of weakness. The flip side of the coin comes during recovery periods when the Fed begins to combat inflation by raising interest rates. This also increases the discount rate and leads to lower equity valuations, unless the effects are offset by decreases in volatility, and higher projected cash flows. Generally speaking, equity prices are directly related to stability, growth in economic activity, and indirectly related to interest rates, and volatility; however, the natural balance between these factors means that you cannot look at each in a vacuum. For example, stating that GDP growth alone is favorable for equity prices is probably not a fair comment since some of this factor will likely lead to an inflationary environment. This will in turn result in an increase in interest rates, which will offset some of the increase in equity value. The significant potential impact of these economic factors on equity prices is why we must watch the Fed, and economic indicators, so closely.

Economic Indicators During the Last Recession—GDP

125

Economic Indicators During the Last Recession—S&P500 Index (% change)

Economic Indicators During the Last Recession—Volatility (VIX in %)

Economic Indicators During the Last Recession—Interest Rates (%)

126

Chapter 5: The DCF Inputs (Beta, ERP, CAPM, & WACC)

Economic Indicators During the Last Recession—Inflation (PCE, % change)

Economic Indicators During the Last Recession—Unemployment Rate (%)

Step 19d.2—Forecasting the ERP (Spread): When discussing spreads in this section, it is important to recognize that we are not referring to credit spreads, which reflect compensation for credit risk (usually a topic of discussion for corporate bonds not sovereign debt). We are referring instead, to the spread between the short-end of the yield curve and the intermediate to long-term portion of the curve. If you plot each maturity point for the U.S. Treasury rate and include the Federal Funds rate (overnight, 3-month, 6-month, 1-year, 2-year, 3-year, 5-year, 7-year, 10-year, 20-year, and 30-year bonds), the result will be a curve, which is typically referred to as the yield curve, or term structure of interest rates. The charts below in Exhibit 106 show the term structure at various points in time. First, look at the historic yield curve from February 2007, prior to the start of the 2008 recession. Notice that the curve was negative sloping. This is typically referred to as an inverted yield curve. The yield curve inverts because investors expect future interest rates to decline. In response, they are willing to pay more for long-term maturities, which drives the longer-term bond prices up, and yields down (bond yields and prices are inversely related). Next, look at the yield curve from December of 2015, which was around the time of the first Federal Funds rate increase after the recession. Notice that the curve was positive sloping. This is generally an indication that the market expects future interest rates to increase, which reflects an expectation of future economic prosperity. In November of 2016 after the U.S. Presidential election the yield curve steepened except for the short-term maturities which remained relatively flat. This demonstrates that moves do not always reflect parallel shifts in the curve. As of October of 2018 the curve has flattened a bit, and the overall level of interest rates has increased.

127

Exhibit 106—Changes in Term Structure

Source: FederalReserve.gov, data retrieved September 24, 2018.

Since we are using the 10-year U.S. Treasury note as our estimate of the risk-free rate, and the Federal Funds rate to project future interest rates, we need to add a spread in our estimate to recognize the difference in these two points along the yield curve. Keep in mind that the historic results (in cells N16 through N70 in Exhibit 102) already include this spread. The chart below in Exhibit 107, shows the historic spread between the Federal Funds rate and 10-year note. The spread has averaged about 1.7% over the last 10+ years, and has averaged about 1.3% in the last two years. If the market expects an economic recession and future interest rates to fall, the spread will tighten or turn negative (inverted yield curve). If the market expects an economic expansion and future interest rates to increase, the spread will widen. Following this logic, if you expect economic strength in the next few quarters, you may want to add a few basis points to the risk-free rate estimates in Exhibit 102 cells L7 through L15 to reflect a widening of the spread between the 10year treasury and Federal Funds rate. Conversely, if you expect economic weakness, you may want to subtract a few basis points from the estimates. I have incorporated five basis points of additional spread per quarter, which will move the spread closer to the long-term average through 2020.

128

Chapter 5: The DCF Inputs (Beta, ERP, CAPM, & WACC)

Exhibit 107—Historic 10-year to Fed Funds Rate Spread

Source: FederalReserve.gov, data retrieved September 24, 2018.

Step 19d.3—Forecasting the ERP (Volatility): Volatility is very difficult to predict. From 1990 through September of 2018 the VIX has averaged just over 19%. Through the first half of 2018, volatility has remained well below the historic average. Volatility tends to increase in times of economic uncertainty, and decrease with economic stability. If you expect an economic downturn, then you would want to increase your forecast of volatility in cells H7 through H15 in the ERP model (refer to Exhibit 102). If you have a more positive view of the market, you would want to keep volatility stable, or decrease the future estimate.

Exhibit 108—Historic Volatility

Source: FederalReserve.gov, data retrieved September 24, 2018.

Step 19d.4—Forecasting the ERP (Market Returns): The last step in the process, is to enter assumptions for the quarterly total market return (the S&P500 index is used as the market proxy). These estimates are input into cells P7 through P15 within the ERP model (refer to Exhibit 102). You can use the historic S&P500 returns and your opinion of the future economic/market conditions to decide what an appropriate future market return will be. Step 19d.5—Forecasting the ERP (Bringing it All Together): After you have made forecasts for the Federal Funds rate, spread, volatility, and market returns, the model will automatically derive the future period forecast for the ERP using the following calculations (refer to Exhibit 109 below): 129

Calculation A) Calculate the annual market return as the geometric mean of the last four quarter projected or reported values. C9 = ((1+P15)*(1+P16)*(1+P17)*(1+P18))^(1/1)-1 Calculation B) The quarterly average 10-year U.S. Treasury rate in cell N15 = the forecasted Federal Funds rate in cell K15 + the forecasted spread in cell L15. The trailing average 10-year Treasury rate in cell O15 is based on the average of the last four quarters, which is brought to cell D9. Calculation C) The excess market return over the 10-year U.S. Treasury rate is calculated in cell E9 as C9 – D9. Calculation D) The new projected market Sharpe Ratio is calculated in cell E68, as the new average excess return from cells E9 through E64 including the latest projected year, divided by the standard deviation of the market returns in cells C9 through C64. Calculation E) The final ERP forecast for the next quarter is calculated as the projected market Sharpe Ratio from cell E68 × the trailing 12-month average VIX in cell J15 (which is a function of the VIX forecast in cell H15 and the last reported values in cells J16 through J18). Whether you intend to use a future forecast for the ERP or the 12-month trailing version, you will need the ERP to calculate the required return on equity for your company (we will cover this in Step 21), and the weighted average cost of capital (covered in Step 22).

Exhibit 109—Forecasting the Future ERP

Step 20: Derive Beta Using the Regression Function in Excel

The ERP from Step 21, represents the risk premium for the entire market. A risk premium for a specific company is required for the valuation of an individual stock. A stock’s beta can be used to link the broader market return, to that of an individual company. Beta is a measure of correlation between the change in value of a specific equity security, and the change in value for an index. Many financial data providers publish beta estimates for stocks. If you do not have access to these sources, or if you would like to have control over the inputs, you can derive beta by regressing the stock’s returns against the S&P500 (the resulting regression beta coefficient is equal to the stock’s beta). This section demonstrates how the calculation works using Excel and FedEx share price as an example. Follow Along in the Spreadsheet: Refer to “File 17-FDX Beta Calculation (Step 20)” for details of how beta is calculated using Excel’s regression function. Refer to the "How to Use This Textbook" section of Chapter 1 for instructions on how to access the spreadsheet files. Step 20a—Obtain Historic Price Data: We will be deriving two beta coefficients: a short-term beta, which we will use in our stage-one DCF (discussed in Chapter 6), and a long-term beta for stage-two. For the short-term beta we will use the last 12 months of monthly return data, excluding dividends. Download the FedEx share price and the S&P500 index value at the end of each month for the 12 months, and calculate the percentage change for each. Step 20b—Install Data Analysis Add-In: You will need the Data Analysis Tool Pack in Excel. It is available for free with Excel, but must be downloaded. If you have not installed the Tool Pack, open an Excel file, select “File”, “Options”, “Add-ins”, “Analysis Tool Pack”, “Go” to install.

130

Chapter 5: The DCF Inputs (Beta, ERP, CAPM, & WACC) Step 20c—Run the Regression: In Excel, select “Data Analysis” on the “Data” tab, then select “Regression”. Enter the percentage change in FedEx shares column for the first variable, and the percentage change in the index column for the second variable. Select an output range and click “OK” to run.

Exhibit 110—Deriving Beta Using Excel (Short-Term Beta)

Exhibit 110 above shows the regression output using changes in share value over the last year. The beta is the slope coefficient in the regression equation, which is 1.121 (cell H15). The interpretation of the result is that the slope, or beta represents the expected percentage change in the value of FedEx shares, based on the percentage change in the S&P500 index. For example, if the S&P500 increased by 5%, we would expect FedEx shares to increase by 5.6% (1.121 × 5% = 5.6%). Of course this assumes we isolate our analysis to include only the market return and beta. Notice that the R Square (cell H4) is relatively low, at just 29%. The interpretation of the R Square is that approximately 29% of the variation in FedEx shares is explained by changes in the S&P500 index. The fact that this correlation is low makes sense, since there are many other factors which cause FedEx shares to increase or decrease in value, other than how the general market performs. We will use the data from the last year as our stage-one, or short-term beta, in our DCF calculation. If we run the same beta regression calculation on a long-term basis with data through January of 2008, the resulting beta coefficient is 1.268. We will use this value for our stage-two discount rate calculation. For additional details of how to interpret regression output, refer to the regression section in Appendix 1: Using Regression Analysis to Predict Earnings.

Step 21: Calculate the Required Return on Equity using Beta and the CAPM

Beta and the ERP are not direct inputs into the Weighted Average Cost of Capital (WACC) calculation, which is the discount rate used to value shares. Instead, these metrics are used to derive the required Return on Equity (ROE). To calculate the required ROE, use the Capital Asset Pricing Model (CAPM) which states that the required return on equity for a stock is equal to the risk-free rate, plus beta times the ERP. After inputting the beta, ERP, and risk-free rate into the CAPM equation, we can solve for the required return. In our model we can add a section below the Cash Flow Statement where we can calculate the required return on equity for FedEx. The required return on equity is 8.90% calculated as the risk-free rate of 3.02% (taken from our ERP model, refer to Exhibit 109), plus a beta of 1.121 (refer to Step 20), times the market risk premium of 5.25%, based on an average VIX of 16.14%. The calculation is shown in Exhibit 111 below. Notice that 8.90% is a relatively low required return on equity compared to the historic average, which reflects the historic low volatility and interest rates. We will use this rate only for the stage-one portion of the valuation, and a different long-term average required return for the stage-two valuation (refer to Chapter 6 for details).

131

Exhibit 111—Calculating the Required Return on Equity

Add footnotes to your model to explain the Constant Sharpe and Volatility: (b) This model uses the Constant Sharpe approach to estimate the Equity Risk Premium (ERP). The S&P500 Constant Sharpe is calculated by taking the excess return on the index over the risk-free rate, divided by the standard deviation of returns. The Constant Sharpe ratio is then multiplied by the estimate of implied volatility to calculate the ERP. (c) The VIX is quoted in percentage points and measures the implied annualized volatility for the S&P500. The VIX is a forward looking measure of implied volatility, however, single day volatility would have too much of an impact on the overall discount rate. For this reason the 12-month trailing average is used.

Follow Along in the Spreadsheet: Refer to “File 9-Valuation (Steps 21 through 30)” for details on the CAPM-based required return on equity calculation, as well as the market multiple and DCF valuation calculations. Refer to the "How to Use This Textbook" section of Chapter 1 for instructions on how to access the spreadsheet files.

Step 22: Calculate the Weighted Average Cost of Capital

The required return on equity is used to calculate the Weighted Average Cost of Capital (WACC). Continuing with the DCF section of our model, we can add additional rows to perform this calculation. First, calculate the weight of equity in the company’s capital structure. Refer to Exhibit 112 below. Cell C379 performs the calculation by dividing the market capitalization in cell C372, by the total capital including debt and equity (cells C372 + R257 + R258 + R263). Next apply the weight of equity to the cost of equity calculated in Step 21. This calculation is the first part of the calculation within cell C383, which is cell C378 times C379. The second part of the stage-one WACC in cell C383 is the after-tax cost of debt times the weight of debt. The final WACC will be used as the discount rate in stage-one of our DCF valuation in Chapter 6. The equation is: WACC = (% of Equity × Required Return on Equity)+ (% of Debt × After Tax Cost of Debt) This demonstration represents one of many approaches used to derive a discount rate. Rather than calculating the WACC, some analysts use a competitor’s WACC, a competitor’s beta in the calculation, or use a sector average discount rate. You can change any of the inputs in the model to suit your needs. If you choose one of these options, keep in mind that you may need to make an adjustment to account for the differences in capital structure between the companies included in your derivation of a discount rate (i.e. de-lever and re-lever beta).

132

Chapter 5: The DCF Inputs (Beta, ERP, CAPM, & WACC)

Exhibit 112—Calculating the Weighted Average Cost of Capital*

*Note at this point in the process we are using the current share price to calculate the market capitalization for the equity to total capital ratio. In Chapter 6 we will change this to the target share price.

Follow Along in the Spreadsheet: “File 9-Valuation (Steps 21 through 30)” includes the details of the WACC calculation. Refer to the "How to Use This Textbook" section of Chapter 1 for instructions on how to access the spreadsheet files.

133

CHAPTER 5 WRAP-UP Takeaways •

The Equity Risk Premium (ERP) represents the return required for investors, in excess of the risk-free rate, as compensation for the additional risk of equity securities. o When perceived market risk/uncertainty increases, during economic downturns, the required return and ERP increases. o When perceived market risk/uncertainty decreases, the required return and ERP decreases.

•

One approach to estimating the ERP is to apply a measure of volatility to the Constant Sharpe ratio (the ratio of historic market returns less the risk-free rate, divided by the standard deviation of returns).

•

The market ERP can be linked to a specific company’s equity shares through the Beta coefficient.

•

Beta is a measure of correlation between the change in value of a particular equity security and the change in value of the market.

•

The Capital Asset Pricing Model (CAPM) is used to calculate the required return on equity. o CAPM equation: Required return on equity = risk-free rate + Beta × ERP

•

The Weighted Average Cost of Capital (WACC) = (weight of equity × required return on equity) + (weight of debt × after tax cost of debt).

Concept Quiz Instructions: Answer each of the following questions as “true” or “false” 1) Changes in interest rates do not impact the required return on equity. 2) During a recession the ERP would likely increase. 3) The U.S. Federal Reserve sets the risk-free rate. 4) An inverted yield curve is a situation when the yield of long-term U.S. Treasury securities is lower than shortterm Treasury Securities.

134

Chapter 5 Wrap-Up Instructions: Use the following chart of the historic ERP to answer Question #5.

5) The spike in the ERP around the end of 2008 was likely caused by which of the following? A. An increase in interest rates. B. An increase in volatility. C. A decrease in credit spreads. Instructions: Use the following information to answer questions 6 through 10. Molly Gatawny is a research intern at International Bank Corp. She is covering Joe’s Hot Dog Stand Inc (JHDS) and has gathered data to assist in determining the correct discount rate to use in a Discounted Cash Flow (DCF) analysis for JHDS. First, she notes that the Constant Sharpe ratio is 0.325, and volatility for the market is at 15%, which is below the ten-year average of 18%. She notes the current spread between the Federal Funds rate of 2.25% and the 10-year U.S. Treasury security is 1.00%. She is not sure which security to use so she plots the entire yield curve:

Molly also regresses the daily returns of Joe’s Hot Dog Stand Inc (JHDS) against the S&P 500 and saves the following results in her data to be checked by the lead analyst:

135

6) What is the Beta for JHDS shares? A. 0.25 B. 0.24 C. 1.12 7) Molly’s boss instructs her to use the 10-year U.S. Treasury security as the risk-free rate. What is the final riskfree rate Molly should use? A. 1.00% B. 2.25% C. 3.25% 8) Calculate the market ERP. A. 4.875% B. 5.460% C. 8.710% 9) What is the required return on equity for JHDS shares? A. 4.875% B. 5.460% C. 8.710% 10) What is the most accurate projection of future economic conditions based only on Molly’s yield curve plot? A. The economy will likely expand. B. The economy will likely contract. C. The economy is likely headed for a recession.

136

Chapter 5 Wrap-Up

Concept Quiz Answers 1) False. The required return on equity is calculated using the CAPM, Required return on equity = risk-free rate + Beta × ERP. Changes in interest rates impact the risk-free rate and the ERP (risk premium over the risk-free rate). 2) True. When perceived market risk/uncertainty increases, during economic downturns, the required return and ERP increases. 3) False. The U.S. Federal Reserve’s FOMC sets the Federal Funds rate. There is no such thing as a set risk-free rate. Most finance professionals use a U.S Treasury security to approximate the risk-free rate, but the maturity of the security which is most appropriate, is a topic of debate. 4) True. An inverted yield curve is a situation when the yield of long-term U.S. Treasury securities is lower than short-term Treasury Securities. 5) B. During the 2008 credit crisis volatility spiked which resulted in an increase in the market ERP. 6) C. Beta is the regression coefficient in cell C16. 7) C. 3.25% which represents the Federal Funds rate of 2.25% plus the spread of 1.00%. 8) A. The market ERP = Constant Sharpe × the VIX = 0.325 × 15% = 4.875%. 9) C. The required return on equity = risk-free rate + Beta × ERP = 3.25% + 1.12 × (0.325 × 15%) = 8.710%. 10) A. positive sloping yield curve is generally an indication that the market expects future interest rates to increase, which reflects an expectation of future economic prosperity.

137

CHAPTER 6: DISCOUNTED CASH FLOW VALUATION Step 23: Calculate the Stage-One DCF

Step 25: Calculate the DCF Valuation

Step 24: Calculate the Stage-Two DCF

Step 26: Understanding the DCF Valuation

Chapter 6 Overview: In this chapter, we will discuss how to use the inputs we calculated in Chapter 5, to derive a DCFbased share valuation. Keep in mind that the DCF calculation is much more complicated compared to a market multiple-based approach, with many more required inputs and assumptions. Despite the relative complexity, DCF analysis is useful for demonstrating the fundamental concept that any asset’s fair value is equal to the present value of the future cash flows it produces…Do not confuse this with the statement that fair value is what someone is willing to pay for an asset. If you are projecting the same cash flows as the market’s overall consensus forecast, and incorporating all potential risk in the discount rate that the market considers, theoretically you should get to the same answer. Which is what someone is willing to pay for that asset. The DCF has three basic parts: 1) projected cash flows, 2) a future growth rate, and 3) a discount rate. This chapter will cover all three, and demonstrate the full DCF calculation. Key Concept 4—Valuation is in the Eye of the Beholder: There are many different opinions of market efficiency and valuation theory. Before beginning the DCF and market multiple valuation chapters, revisit the “Thoughts on Valuation” section of the introduction, and as you read consider customizing this approach to suit your ideas on valuation.

Step 23: Calculate the Stage-One DCF

There are many different approaches to calculating a DCF-based valuation. I use a two-stage approach, with five years of cash flow projections in the first stage, which represents the high-growth period for the company. Then, I calculate a terminal value in stage-two, with the assumption that the company reaches a constant growth state in year six. This is a rather large assumption for most companies since they will likely go through many business cycles during their existence, rather than reaching a flat growth rate held in perpetuity. However, once the model is established and calibrated with this assumption, it gives a comparable valuation over time. With each additional year added to the forecast as time progresses, the constant growth stage will be continually pushed out if management is able to keep growth on track. In the model, I calculate the free cash flow at the bottom of the Cash Flow Statement as the cash from operations, less capital expenditures, plus after-tax interest expense. I then discount the free cash flow using the WACC calculated in Step 22. This calculation is performed for each of the five, stage-one years. The sum of the five discounted cash flows is saved in the DCF section of the model. Exhibit 113 below demonstrates the free cash flow calculation. For example, in forecast year two, start with the cash flow from operations in cell AB320. Subtract the capital expenditures, or in this case since capital expenditures are reflected as a cash outflow on the Cash Flow Statement, add the negative value to decrease the free cash flow by the capex amount. Next add back the after-tax cost of debt, similar to capex, interest expense is reported as a negative 138

Chapter 6: Discounted Cash Flow Valuation number in the FedEx model, so we will subtract the negative, which is cell AB29 times 1 minus the effective tax rate estimate in cell C381. The full free cash flow equation is: AB320 - (-AB322) + ((-AB29) × (1-$C$381)) = 8,937 - (+5,869) + -671 × (1 -25%) = 3,572

Exhibit 113—Free Cash Flow Calculation

After the free cash flow is calculated for each of the five stage-one years, we can discount these back using the WACC calculated in Step 22. Excel does have a cash flow present value calculation function. For transparency, I have included the calculation in the model without the Excel shortcut. Refer to Exhibit 114 cell AB342 which equals the free cash flow from cell AB340, divided by one plus the discount rate in cell C383, raised to the number of years, or: AB340/(1+$C$383)^AB341 = 3,572 / (1 + 7.6%) ^ 2 = 3,083 To complete the final stage-one DCF we sum the discounted free cash flows for all five years in cell C391 (W342 + AB342 + AG342 + AL342 + AQ342): = 1,812 + 3,083 + 3,712 + 3,734 + 3,696 = 16,037

139

Exhibit 114—Stage-One DCF Calculation

Step 24: Calculate the Stage-Two DCF

The stage-two calculation estimates the terminal value of the company holding growth constant. The calculation may look daunting at first (refer to Exhibit 115 cell C390): =((((AQ320 × (1 + C386)) - (C387 × AQ13 × (1 + C385)) + (C382 × (AQ263 + AQ257 + AQ258)))) ÷ (C388C385)) ÷ (1 + $C$388)^5 =((((11,382 × (1 + 0.06)) - (0.08 × 86,316 × (1 + 0.06)) + (0.027 × (29,244 + 2,575)))) ÷ (0.117 - 0.06)) ÷ (1 + 0.117)^5 = 56,741 (note a slight difference exists here due to the rounding of decimal places). This is actually a simple perpetuity equation. Breaking the components down, will help to show that it represents the cash flow for the first year of the constant growth stage, divided by the long-term WACC, minus the constant growth rate. Since this is an approximation of the terminal value at the end of the fifth year, we discount the result back five years to present value terms. The numerator, which represents the cash flow of year six, is the complex part with the following three components: Part 1) Cash Flow From Operations (CFO) = CFO of year 5 x (1 + constant CFO growth rate) Part 2) Minus: capex = Average capex to sales ratio x [sales x (1 + constant sales growth rate)] Part 3) Plus: the after-tax cost of debt = After tax cost of debt x long-term debt Notice that I calculate two discount rates, one for stage-one, and another for stage-two. This approach differs from that of other analysts who tend to use a basic market-based WACC which does not change much over time. Instead I use different rates for each stage because I want to maintain the link between changes in the macroeconomic environment (i.e. changes in interest rates, volatility, etc.) and share valuation. At the same time, it is important to recognize that over the long-term, business cycles will come and go, but the key inputs in the valuation should have long-term averages similar to historic results. Since the majority of a DCF-based valuation comes from the terminal value, adjusting the WACC used for the second stage cash flows to reflect short-term changes in interest rates, volatility, or the correlation of shares to the market, would have an unjustified impact on the price target. Maintaining two distinct discount rates for the short-term and 140

Chapter 6: Discounted Cash Flow Valuation long-term portions of the valuation, allow the model user to maintain the macroeconomic link, without over or under estimating the impact on target share price. The stage-two discount rate (refer to Exhibit 115, cell C388) assumes the weight and cost of debt remain constant, and cost of equity reaches the long-term average based on a long-term market ERP of 6.0%, using the historic average VIX of 18.59% and Constant Sharpe of 0.325. The required return on FedEx shares in stage-two is 14%, based on the longterm beta of 1.268, and the historic average 10-year U.S. Treasury rate of 6.3%. The full stage-two WACC equation including the cost of debt is as follows: (C379 × (0.063 + (1.268 × (0.325 × 0.1859)))) + ((1 - C379) × C382) = (79.8% × (0.063 + (1.268 × (0.325 × 0.1859)))) + ((1 – 79.8%) × 2.7%) = 11.7%

Exhibit 115—Stage-Two DCF Calculation

Step 25: Calculate the DCF Valuation

Now add the stage-one and stage-two cash flows in present value terms. Divide the total value by the diluted shares outstanding, and add the net cash/(debt) per share to arrive at the equity valuation per share of $218 (refer to cell C393 in Exhibit 115). There is one final step to complete the DCF valuation process, and that is to change the “current share price” (refer to Exhibit 114, cell C370), to the “target share price”. This is a matter of preference, so you may decide to leave the DCF valuation as is, but consider what would happen if we simply kept the current share price in the DCF calculation. For the sake of this demonstration assume market risk increased from one quarter to the next. As the share price declines, the equity-to-total capital ratio will also decline. As a result the discount rate will decrease, and the share price projected by the DCF would actually increase, despite the heighten level of risk. The reverse effect would occur in a scenario where market risk decreases from one period to the next. From a reasonableness standpoint this simply does not make sense. To demonstrate this point another way, assume the current share price of a stock is $100, and the DCF valuation is projecting a 12-month target price of $150. If the DCF target price of $150 came to pass, then equity would make up a larger percentage of the total capital structure, and the share price would actually be much lower. So in order to have a logical price forecast, we must also forecast the change in the company’s capital structure.

141

To solve this issue, the current share price can be replaced with the target share price in the calculation of equity-tototal capital. This recognizes the balancing effect of changes in circumstances over time. The inherent problem with this approach is that the DCF calculation relies on an estimate of the equity-to-total capital ratio. Since the share price goes into the calculation of the capital ratio, this will naturally cause a circular reference. We could use trial and error to type in share prices between the current price and target price, until the price for the market capitalization calculation equals that of the DCF target. Or we can simply input the equation of cell C370 minus C393, and use the Goal Seek function to change C370 to make C370 equal C393. Note that you cannot simply set C370 equal to C393 as this will cause a circular reference error. This should be the final step in your model, as any changes in the inputs could impact the valuation. After we perform this step, the DCF target price changes from $218 to $224 per share.

Step 26: Understanding the DCF Valuation

At this point, you should be able to recognize how each input impacts valuation under different scenarios. As you think about the impact, keep these basic equations in mind: Market ERP) Required Return on Equity – Risk-Free Rate, or Constant Sharpe × Volatility CAPM) Company Specific Required Return Equity = Risk-Free Rate + Beta × Market Risk Premium WACC) Weight of Debt × After-Tax Cost of Debt + Weight of Equity × Required Return on Equity Using these equations, we can draw the following conclusions about the DCF inputs: Volatility: The higher the VIX, the greater the cost of equity. Increases in volatility will have a greater impact on companies with a greater proportion of their capital structure in equity rather than debt. This scenario can tilt if a large debt balance results in a high probability of default. In this scenario, the company’s credit spread would widen, and the cost of debt and equity would spike. Interest Rates: An increase in market interest rates will increase the required return on equity through the CAPM, and result in lower asset values. If a company has fixed rate liabilities, higher rates will not impact the cost of debt in the WACC when rates change. If the company’s debt has adjustable features, then changes in interest rates can have an impact on the cost of debt. Beta: A high beta coefficient will result in greater variability in the cost of equity. Final Note on DCF Valuation: If you have not used a DCF valuation in the past, after reading this section you may be thinking “I have been using market multiples and missing the risk in my valuation all this time?!” That is not necessarily the case. Market multiples are very useful tools, arguably more useful than discounted cash flow analysis which requires many more assumptions. Just remember that if you are going to use a market multiple approach, you must adjust the multiple during times of increased risk to reflect the greater required return on equity. Regardless of the valuation approach you choose, keeping the fundamental risk and return theory in mind will ultimately lead to better decisions. Follow Along in the Spreadsheet: Refer to “File 9-Valuation (Steps 21 through 30)” for details of the Discounted Cash Flow Valuation calculation. Refer to the "How to Use This Textbook" section of Chapter 1 for instructions on how to access the spreadsheet files.

142

CHAPTER 6 WRAP-UP Takeaways •

Discounted Cash Flow (DCF) analysis is based on the fundamental concept that any asset’s value is equal to the present value of the cash flows it is expected to produce.

•

The basic components of DCF analysis consist of projected cash flows, a future period growth rate forecast, and a discount rate.

•

There are many different ways to calculate a Discounted Cash Flow valuation. This program demonstrates a two-stage approach.

•

o

Stage-one consists of the next five years of projected cash flows.

o

Stage-two estimates a terminal value of the company based on an expected constant growth held in perpetuity.

There are many definitions of “cash flow” used in discount valuation analysis. One approach is to use the free cash flow, which is calculated as: o

•

Cash flow from operations – capital expenditures + after-tax interest expense

The DCF valuation is sensitive to changes in the discount rate driven by the following components: o

Volatility has an inverse relationship with equity valuation. As the VIX increases the required return on equity increases, which drives the WACC up, and decreases the present value of future cash flows.

o

Interest rates increase the risk-free rate in the CAPM, which increase the discount rate. Note that the risk-free rate is also subtracted from the expected equity market return to calculate the ERP which is included in the CAPM, but all else equal, increases in interest rates will increase the discount rate and decrease the present value of future cash flows.

o

Changes in the correlation of shares with the market (Beta) will impact the discount rate. Beta is not a constant. Increases in beta over time will result in greater variability in the required return on equity, higher discount rates, and lower present value of future cash flows.

143

Concept Quiz Instructions: Use the following information to answer questions 1 through 10. Anna is a senior research analyst at Wall Street Bank & Broker Dealer LLC. Her firm will be initiating coverage of the Amusement and Recreational Services Industry, and Anna will be covering Fun Times Inc (FTI), an amusement park which operates in the northeast region of the United States. Anna begins working on the following five-year earnings model for FTI:

144

Chapter 6 Wrap-Up Anna has also collected the following information. She asks you to assist with the highlighted cells.

1) What should Anna input for the market Equity Risk Premium (ERP) in cell M7? A. 4.50% B. 5.04% C. 6.30% 2) What should Anna input for FTI’s required return on equity in cell M10? A. 5.00% B. 6.30% C. 9.55% 3) What should Anna input for FTI’s market capitalization in cell M11? A. $10 B. 350 million shares C. $3,500 million 4) What should Anna input for the ratio of equity-to-total capital in cell M12? A. 89.9% B. 95.2% C. 100% 5) What should Anna input for the Weighted Average Cost of Capital (WACC) in cell M14? A. 3.02% B. 8.88% C. 9.55% 6) What should Anna input for the year-two Free Cash Flow in cell E54? A. 178 B. 186 C. 228

145

7) What should Anna input for the year-two discounted Free Cash Flow in cell E56? A. 150 B. 163 C. 175 8) What should Anna input for Stage-one Net Present Value of cash flows in cell M16? A. 150 B. 180 C. 584 9) What should Anna input for the DCF-based share valuation in cell M23 (assuming Anna has opted to leave the current share price constant in the equity-to-total capital ratio), and what is the appropriate assessment of the current share price based on the DCF analysis? A. $8.25, shares are currently trading above the DCF valuation. B. $8.55, shares are currently trading below the DCF valuation. C. $8.55, shares are currently trading above the DCF valuation. 10) What should Anna input for the DCF-based share valuation in cell M23 (assuming Anna has opted to change the share price to represent the target price in the equity-to-total capital ratio), and what is the appropriate assessment of the current share price based on the DCF analysis? A. $8.25, shares are currently trading above the DCF valuation. B. $8.56, shares are currently trading below the DCF valuation. C. $8.56, shares are currently trading above the DCF valuation. 11) The FOMC has decided to increase the Federal Funds rate. What is the most likely impact on the discount rate and share valuation? A. The WACC will likely increase and share valuation will decrease. B. The WACC and share valuation will remain unchanged. C. The WACC and share valuation will increase.

Concept Quiz Answers 1) B. The market ERP is calculated as volatility × the constant Sharpe ratio, or 0.325 × 15.5% = 5.04%. 2) C. The required return on equity using the Capital Asset Pricing Model (CAPM) = risk-free rate + beta × the ERP, or 3.25% + 1.25 × 5.04% = 9.55%. 3) C. Market capitalization equals shares outstanding × market price per share, or $10 × 350 million shares = $3.5 billion. 4) A. Equity to total capital is calculated as market cap ÷ (market cap + debt) = $3,500 ÷ ($3,500 + $395) = 89.9%. 5) B. The WACC is calculated as the weight of equity × the required return on equity + the weight of debt × the after-tax cost of debt, or 89.9% × 9.55% + (100% -89.9%) × 3.02% = 8.88%. 6) A. Free cash flow is calculated as cash flow from operations – capital expenditures + after-tax interest expense, or $211M - $25M + ($9.5M × (1 – 20%)) = $178M. 7) A. The discounted free cash flow is calculated as $178M ÷ (1 + 8.88%) ^ 2 = $150M. 8) C. The stage-one Net Present Value of Cash flows is the sum of the five years of discounted free cash flows, or $71M + $150M + $106M + $133M + $124M = $584M. 9) C. The final DCF valuation per share is the Stage-one NPV + the Stage-two NPV ÷ shares outstanding + Net cash/(debt) per share = ($584M + $2,195M)/ 350M shares + $0.61 = $8.55 per share. Since the DCF value is lower than the current market value of $10, this means shares are currently trading above the DCF valuation. 146

Chapter 6 Wrap-Up 10) C. To complete this additional step, setup an equation of cell M3 minus M23. Then use the goal seek function to change cell M4 until cell M3 minus M23 equals zero. The goal seek will return a target share price of $8.56. Since the DCF value is lower than the current market value of $10, this means shares are currently trading above the DCF valuation. 11) A. Although the Fed Funds rate is not a direct component, all else equal an increase in the Fed Funds rate will result in an increase in the risk-free rate and the WACC. Solution model for questions 1 through 10.

Follow Along in the Spreadsheet: Refer to “File 19-Chapter Wrap-Ups”, worksheet “Chapter 6 & 7” to access the Excel file used for the questions above. Refer to the "How to Use This Textbook" section of Chapter 1 for instructions on how to access the spreadsheet files.

147

CHAPTER 7: MARKET MULTIPLE-BASED VALUATION Step 27: Choose a Multiple

Step 29: The Historic Multiple

Step 28: Separate Net Cash

Step 30: Create a Price Band

Chapter 7 Overview: Most companies can be valued by applying a market multiple to forecasted earnings. In this chapter, we will discuss some of the key factors to consider when choosing a multiple, and demonstrate how to use the earnings forecasted in the modeling chapters to derive a market multiple-based valuation. We will also discuss the importance of developing a price band around your target share valuation.

Step 27: Choose a Multiple

When choosing an appropriate multiple to use for valuation, the following factors should be considered: Industry/Sector: Each sector has a set of metrics which are widely accepted by analysts and investors, including which multiples to use for valuation. For banks, it is the Price-to-Book Value multiple, for energy companies, it is the Enterprise Value-to-EBITDA multiple (EV/EBITDA), and for developed companies with relatively stable earnings, it is the Price-toEarnings (PE) multiple. Cyclical vs Non-Cyclical: If the company’s industry is currently in decline, then the company could be operating at a loss. In this case, a PE multiple cannot be used for valuation. As an alternative, it may be appropriate to use a Price-toEBITDA ratio or Price-to-Sales ratio. Another alternative would be to forecast earnings on a normalized basis, assuming a return to average market conditions at some point in the future, and apply a PE ratio to this later stage in the business cycle. Lifecycle Stage: Companies in the development stage have different earnings profiles compared to those which have reached maturity. Developing companies may not be profitable or may have volatile earnings streams with high growth rates, in which case it may make more sense to apply a multiple to a later stage of earnings, rather than using a Next Twelve Month (NTM) PE. For example, Netflix trades at a PE multiple of about 80x the NTM consensus EPS estimate, while FedEx trades between 15x and 18x (excluding net debt per share). The reason that investors are willing to pay more for Netflix earnings, is that the company is currently in a hyper-growth stage. As a result, investors are valuing shares based on the ultimate long-term earning capacity, which will not be reached for many years. Therefore, the NTM earnings are less relevant for valuation purposes. For companies in a hyper-growth stage, a growth adjustment could be used make the PE ratio more meaningful. Other than using a historic average multiple, you may want to consider using the multiple of a competitor, a sector average, or a cyclical peak/trough multiple. The multiple is easy to change in the model. Feel free to try different multiple choices to see what the impact is on valuation.

Step 28: Separate Net Cash/(Net Debt)

Next, review the company’s Balance Sheet and decide if any of the company’s assets or liabilities should be removed and valued separately. A good example of assets which should be valued separately, is an investment in shares of a 148

Chapter 7: Market Multiple-Based Valuation distinct company. For example, when Yahoo was a public company, it owned a significant stake in Alibaba. After Alibaba went public, it was easy for analysts to value Yahoo’s stake in Alibaba separate from the rest of Yahoo’s enterprise value. Another important item you may want to value separately, is cash and investments (net of debt). I typically separate net cash and investments from the equity valuation in my models after deducting debt, because I do not want to give a valuation benefit to companies that rack-up large debt balances, and let the cash sit on the Balance Sheet. This approach will also adjust the valuation to reflect dividend distributions, since the dividend will reduce the cash balance after it is paid. Some companies, generate substantial cash balances in other countries. If this foreign domiciled cash was brought back to the U.S., the company would be subject to a repatriation tax. You can make adjustments to the net cash amount to approximate this impact, although after the U.S. corporate tax reform legislation in 2018, the effect of repatriation is not as significant as it has been in the past.

Step 29: Calculate the Historic Multiple

Now that we have the inputs, the calculation is straight forward: the PE multiple times the sum of the next four quarters of diluted EPS, plus adjusted net cash/(debt) per share, equals the implied share valuation. This calculation is demonstrated in cell C364 of Exhibit 116 below, which equals the sum of cells S44 + T44 + U44 + V44 times 16.7 divided by the shares outstanding, plus the net debt per share of -$48.92 = $238 per share. Next we can apply a 50% weight to the DCF-based valuation from Chapter 6, and a 50% weight to the market multiple valuation to arrive at the final implied 12-month target share price of $231 per share (Exhibit 116 cells C6 through C8).

Exhibit 116—Market Multiple Valuation

Step 30: Create a Price Band

It is important to recognize that we have made a significant number of assumptions in the model. For this reason, having one specific target price may misrepresent the level of certainty in the share valuation. To reflect this fact, I like to create a price band around the implied target value, based on the historic changes in share price over the last 12 months. To do this, first calculate the mean monthly return for shares over the last 12 months, which is 0.03% for FedEx. Next, calculate the standard deviation in monthly returns, which is 5.58%. The standard deviation is a measure of dispersion around the mean monthly return. The larger the historic standard deviation, the greater the volatility in share price. Using a normal distribution, approximately 95% of observations fall within two standard deviations of the mean. Therefore, at the 95% confidence interval, the lower bound is the mean return, minus 2 times the standard deviation, or 0.03% ­ (2 × 5.58%) = -11.13%. The upper bound is the mean return, plus 2 times the standard deviation, or 0.03% + (2 × 5.58%) = 11.19%. Now that we have our upper lower bounds, we can apply these percentage changes to our target price. Exhibit 117 demonstrates this calculation using our 50/50 DCF/Multiple target valuation of $231 per share, which results in a target price band of $206 to $257. 149

This approach has multiple limitations, including the fact that is assumes historic results can predict future return characteristics. The approach also assumes the stock's returns are normally distributed. Tail events, that is events which are not included in the 95% range of the distribution, will occur, and are not be picked up in our band. In addition, since we are only including the last 12 months of data to create the band, any significant changes such as price increases or decreases which have only occurred once in many years, will not be included in the band. Despite these limitations, creating a band offers some insight into how wide the level of certainty is around the valuation expectation, and to reiterate the point that valuation is a delicate concept with many forces at work. Due to the nature of forecasting and valuation we have a very low level of confidence in the ability to project the value of a particular security.

Exhibit 117—Target Price Band

Follow Along in the Spreadsheet: Refer to “File 9-Valuation (Steps 21 through 30)” for details of the market multiple valuation and target price band. Refer to the "How to Use This Textbook" section of Chapter 1 for instructions on how to access the spreadsheet files. Follow Along in the Spreadsheet: If you would like to see an example of the standard deviation calculation, refer to “File 22-FDX Model (Appendix 2-Step 1 F2Q2019 Preview)”, worksheet “Std Dev & Mean Return”. Refer to the "How to Use This Textbook" section of Chapter 1 for instructions on how to access the spreadsheet files.

150

CHAPTER 7 WRAP-UP Takeaways •

There are many different market multiples which can be used to value equity shares.

•

Most industries or sectors have widely accepted multiples which are used by most investors.

•

A Price-to-Earnings (PE) ratio based on Earnings Per Share (EPS) will not be useful in the case of companies operating at a net earnings loss. Instead an Enterprise Value-to-EBITDA, Price-to-EBITDA, or Price-to-Revenue multiple may be used for valuation purposes. Essentially the approach is to move up the Income Statement until you find a positive value which would result in a meaningful multiple.

•

Multiples are “relative” valuation metrics, which means the share value produced is relative to another value. o For example, if a multiple based on an industry’s average multiple is used, the share value produced will be relative to the industry. o Or if a historic average multiple is used, then the value is relative to the average value which existed in the past.

Concept Quiz Instructions: Use the following information to answer questions 1 through 5 (Note this case is a continuation of the example used from Chapter 6). Anna is a senior research analyst at Wall Street Bank & Broker Dealer LLC. She is covering Fun Times Inc (FTI), an amusement park which operates in the Northeast region of the United States. Anna has produced the following earnings model for FTI:

1) Assuming Anna’s full year 2019 Earnings Per Share (EPS) estimate of $0.34 is equal to the consensus estimate, what is the current trailing 12-month Price-to-Earnings (PE) ratio? A. 27.64x B. 28.46x C. 29.45x

151

2) Assuming Anna’s full year 2019 Earnings Per Share estimate of $0.34 is equal to the consensus estimate, what is the current forward Next Twelve Month (NTM) Price-to-Earnings (PE) ratio? A. 27.64x B. 28.46x C. 29.45x 3) Assuming Anna’s full year 2019 Earnings Per Share estimate of $0.34 is equal to the consensus estimate, what is the current forward Next Twelve Month (NTM) Price-to-Earnings (PE) ratio after removing net cash per share? A. 27.64x B. 28.46x C. 29.45x 4) FTI’s closest competitor Adventure Land Corp (ADC) has a similar capital structure as FTI and trades at 24x NTM EPS excluding net cash per share. The average forward PE for the Amusement and Recreational Services Industry (excluding net cash) is 22x. What would the value of FTI shares be at the multiple of ADC? A. $8.15 B. $8.55 C. $8.77 5) Over the last three months FTI’s shares have traded at a peak of $11.00, a low of $7.75, and an average of $8.00. The NTM consensus EPS has not changed during this time. Calculate the three-month average NTM forward PE (excluding cash), and what is the assessment of the current share price based on this multiple? A. 23.53x, shares are trading below the three-month average. B. 21.74x, shares are trading above the three-month average. C. 21.74x, shares are trading below the three-month average.

Concept Quiz Answers 1) B. The trailing 12-month Price-to-Earnings ratio is the current market price divided by the previous 12-month actual EPS, or $10 ÷ $0.35 = 28.46x. 2) C. The forward NTM Price-to-Earnings ratio is the current market price divided by the NTM consensus EPS estimate, or $10 ÷ $0.34 = 29.45x. 3) A. The forward NTM Price-to-Earnings ratio excluding net cash is the current market price minus net cash per share divided by the NTM consensus EPS estimate, or $10 - $0.61 ÷ $0.34 = 27.64x. 4) C. The valuation would equal $0.34 × 24 + $0.61 = $8.77. 5) B. The multiple equals the average price minus net cash per share divided by the NTM consensus EPS estimate, or $8.00 - $0.61 ÷ $0.34 = 21.74x. Shares are currently trading at a multiple of 27.64x which is greater than 21.74 average. Follow Along in the Spreadsheet: Refer to “File 19-Chapter Wrap-Ups”, worksheet “Chapter 6 & 7” to access the Excel file used for the questions above. Refer to the "How to Use This Textbook" section of Chapter 1 for instructions on how to access the spreadsheet files.

152

FINAL THOUGHTS ON THE COMPLETED MODEL Congratulation You Have Completed Your Model…But Does it Make Sense? Take some time to revisit the fundamental simplicity vs complexity concept. Have all of our inputs and assumptions come together to form a coherent forecast, or are we missing the forest for the trees? To see if your model passes the test, think about the forecasted revenue, EPS, and free cash flow growth you are projecting. At a high-level, do the growth rates make sense based on historic trends, competition, the current macroeconomic environment? Building a model is an exercise of learning about the company, and documenting our forecast assumptions. At the end of the process a few simplistic checks can be a refreshing exercise. Not to mention it gives you a chance to stand back and admire all of your hard work. How About the Share Valuation Forecast? Now that we have completed the valuation section, let’s reflect on what we have accomplished. Consider all of the various inputs used in the DCF calculation (interest rates, volatility, equity market projections, company cash flows, etc.). Do you feel comfortable that all of these assumptions entered in the forecast are accurate? You should not. Think about how many “experts” spend countless hours analyzing data simply to determine whether or not the Fed will raise interest rates at the next FOMC meeting. And that is just one of the many inputs into our model. Next consider the market multiple approach. Much easier than the DCF, right? Wrong. All of the inputs in the DCF valuation are included in determining what the correct multiple should be, we simply do not have visibility into the impact each factor has on the multiple. The ease of the multiple application is an illusion. The reality is that projecting earnings and cash flows is nearly impossible, even before attempting to determine a value of these earnings and cash flows. This is not a hopeless effort. You can take comfort in the fact that you have done your due diligent, and have established a reasonable basis for your projection. Then sit back and enjoy the spectacle of news commentators, bloggers, economists, and analyst, as they attempt to bring order to the chaos and confusion that is forecasting. The Bigger Picture: The equity markets are in a perpetual state of balance, with offsetting forces constantly at work. Each factor plays an important role in this balance, much like an individual instrument takes part in a symphony played simultaneously by an orchestra. And just as with the symphony, there is beauty in the market’s balance; however, when an individual tries to tame the market with his or her limited information, the result can look haphazard or chaotic. Do not let this deter you from modeling, as your analysis plays an important role in the natural balance of the market, and the fact is, one instrument played by itself can sound very beautiful indeed! When you put all the individual factors that go into a forecast together you may feel overwhelmed, and begin to think that equity research as a career is impossible. In fact it is very possible, as long as you understand the purpose of the field. No single analyst could possibly have the most accurate price target for every stock covered. If that were the case, there would be no market since all participants would follow every recommendation of this analyst, and be on the same side of every trade. The purpose is to use your knowledge to articulate your opinion. My hope is that you have come away from this experience with a respect and appreciation for the difficulty in forecasting, and feel empowered by this new understanding to begin developing your own projection models. What’s Next? Models are like living creatures (Key Concept 5). They should not sit idle on the shelf. They are tools which should be utilized, and of course, updated for new information. The final chapters will demonstrate how to use and maintain your model. 153

CHAPTER 8: HOW TO USE YOUR EARNINGS MODEL Step 31: Run Scenario Analysis

Step 34: Earnings Simulation—Prepare

Step 32: Perform Sensitivity Analysis

Step 35: Earnings Simulation—Update

Step 33: Analyze Guidance/Consensus

Step 36: Regular Model Maintenance

Chapter 8 Overview: Now that you have completed your earnings model, you can use it to analyze the company’s future prospects. This is the point where all the hard work put into creating a dynamic, equation-based forecast pays off. In this chapter, I will explain some of the tasks you can perform with your model, such as running scenario and sensitivity analysis, analyzing management’s guidance, and the consensus analyst estimates. Note: In Step 35 we will simulate an earnings release, and then discuss how to update and maintain your model over time as the company continually releases new financial results.

Step 31: Run Scenario Analysis

If you look at your finished model, you can easily identify all of the key inputs which we have shaded blue. You can change any of these inputs and instantly see how the change would impact earnings and the financial statements. Using the input cells, you can create multiple upside and downside scenarios. For example, you may want to see what the impact on earnings would be if the FedEx International Express sub-segment growth accelerated due to faster than expected integration of TNT, and better then expected synergies between the two businesses. You can now simply increase the growth rates on volume and yield in the International segment, and decrease the operating expenses to reflect this scenario. On the other hand if you think that Amazon’s shipping program will result in a general decrease in express shipping for the entire industry, and FedEx will be forced to increase automation efforts to keep up, then you can decrease the future period price (package yield) assumptions, and increase the capex-to-revenue ratio projections. You can also consider the knock-on effects such changes to your model would have. For example, if revenue in dollars increases, then operating margin percentage would generally improve, since there is a wider revenue base to apply the costs to, a sort of “economies of scale” situation, which you may want to incorporate into your projections. On the flipside, if revenue decreases significantly, then the need for additional capex may actually decrease, since shipping volume would probably be lower. As a result the infrastructure required for the lower volume may change. After you create your upside/downside scenarios, or any other variations you would like to see in the model, you can review the impact on the Next Twelve Month (NTM) EPS, the free cash flows, and the target share valuation under the market multiple and DCF calculations. Besides running company specific scenarios, you could also analyze the impact of macroeconomic changes. For example, if you want to see the impact of better than expected global trade and shipping, you could increase the volume and yield assumptions across the board. You could also input higher interest rates, changes in correlation with the 154

Chapter 8: How to Use Your Earnings Model market (beta), increases in volatility, or simply change the market multiple or discount rate to reflect a changing required return on equity. The possibilities are endless. Take some time to input different assumptions in your model and review the impact.

Step 32: Perform Sensitivity Analysis

Sensitivity analysis attempts to capture and quantify the impact of changes in a sample of variables, when there is uncertainty about their future value. The primary limitation of sensitivity analysis is that the calculation only works for one or two variables at a time, whereas scenario analysis allows you to change all variables at the same time. Despite the variable limitation, sensitivity analysis is a useful exercise to perform prior to an earnings release to establish a quick view of the potential outcomes. The “Data Table” function in Excel can be used to create quick sensitivity charts for two variables. For example, let’s assume that FedEx will release fiscal first quarter 2019 results tonight after the market closes. The consensus estimate for revenue and non-GAAP operating margin are $16.87B and 8.5%, and we are modeling $16.86B and 8.5% (after consideration of the 90 basis point impact from the pension expense accounting change). You can run a sensitivity table to show what the impact on EPS would be if the revenue and operating margin come in at different points above or below the consensus estimate, or our estimate. Or if there are specific attributes within the model that you are uncertain of, you may want to use sensitivity analysis to estimate the impact of those variables on EPS. To demonstrate we can assume that investors are nervous about the FedEx International Express sub-segment given global macroeconomic concerns and TNT merger execution risk. We can take the International sub-segment with the largest impact on revenue, which is the International Priority Express business, and run a sensitivity table on the volume and yield inputs from our model shown in Exhibit 118 below (cells S92 and S94).

Exhibit 118—International Priority Express Sub-Segment: Volume and Yield Forecast

To do this, start by setting up a chart with one variable across the top (Average Daily Volume, ADV) and the other running along the left side (yield—revenue per package). Refer to the “Sensitivity Analysis” section of the model, or Exhibit 119 below. Next, input the values you want to see for each variable. I like to have my base case fall in the middle of the chart. To do this, set the value for each variable in the middle cells equal to the value in the model. For ADV growth, this value from our estimate is 5.1%, shown in cell S92 (Exhibit 118 above), and 5.9% for yield growth. Input these two values in cells G425 and C429 (Exhibit 119). Now set the remaining points for the data table with equal increments above and below the midpoint. For example, I have set the ADV growth estimates that I want to see in my chart at 2.5% increments, so cell H425 equals cell G425 plus 2.5%, cell I425 equals H425 plus 2.5%, and cell J425 equals cell I425 plus 2.5%. On the downside of the data table cell F425 equals G425 minus 2.5%, cell E425 equals F425 minus 2.5%, and cell D425 equals E425 minus 2.5%.

155

Follow the same process with yield growth rate, except use increments of 1.5%. Cell C430 equals cell C429 plus 1.5%, cell C431 equals C430 plus 1.5%, and cell C432 equals C431 plus 1.5%. On the bear-side of the chart, cell C428 equals cell C429 minus 1.5%, cell C427 equals C428 minus 1.5%, and cell C426 equals C427 minus 1.5%.

Exhibit 119—FedEx EPS Sensitivity to Express Priority ADV and Yield: Data Table Setup

After the variables are setup, you can link the table to the model. To do this, set an equation which points to the result in the model you are looking to solve for, in this case EPS in our earnings model (cell S44 of Exhibit 118). Place this equation in the upper left-hand corner of the chart where the variables meet (cell C425 in Exhibit 119 should be set to equal cell S44 of Exhibit 118). This is a reference used by the data table. If you want to hide the result, change the color of the text to be the same as the shade of the cell it is in (which will be white text if you are not using color in the chart header). Keep in mind that the table will not calculate if the variables are not hardcoded values, so cells S92 and S94 in Exhibit 118 must be saved as values not equations. After you run the data table you can reset the cell back to an equation. Also the sensitivity table must be in the same worksheet as the model. Next, highlight the entire chart (cells C425 through J432 in Exhibit 119). From the “Data” tab select “What-If Analysis”, and “Data Table”. This will invoke the data table input selection window. In the “Row input cell” range select the International Express Priority yield growth cell (Exhibit 118, cell S94) within the model. In the “Column input cell” range select the International Express Priority ADV growth cell within the model (Exhibit 118, cell S92). Click “OK” to run the table. For each point in the data table, Excel will take the yield and ADV assumptions across the top and left hand side of the table, plug them into the earnings model, recalculate the EPS, and populate the result in the data table. When using the data table analysis function, all other variables are held constant to the assumptions contained in the model (i.e. operating expenses, tax rate, etc.).

Exhibit 120—Link Data Table to the Model and Run

156

Chapter 8: How to Use Your Earnings Model The results in Exhibit 120 show the different fiscal first quarter non-GAAP EPS estimates using the different growth rates for the International Express Priority sub-segment. Notice that our original EPS estimates of $3.53 lies at the midpoint of our table since cells G425 and C429 represent the growth rates in our original forecast. The upper left-hand quadrant of the table represents our downside sensitivity, and the lower right-hand quadrant is our upside sensitivity. The key takeaway is that a 2.5% increase in ADV or 1.5% increase in yield, adds about 2 to 3 cents to EPS. Now that you have the estimates of the impact on EPS for different possibilities, you could apply the Next Twelve Month (NTM) Price-to-Earnings multiple that shares are trading at ahead of the press release to calculate the approximate impact these outcomes could have on share price. Keep in mind, this holds all other factors constant, and if there is a significant beat or miss of the revenue or margin estimates, then the consensus for the next few quarters would probably need to be revised. Despite the ever-changing estimates of future earnings, a data table can be a very useful tool for a quick view on possible outcomes. Follow Along in the Spreadsheet: Refer to “File 10-Sensitivity (Step 32)”for the details of the sensitivity calculation. Refer to the "How to Use This Textbook" section of Chapter 1 for instructions on how to access the spreadsheet files.

Step 33: Analyze Guidance & Consensus

In Step 16, we analyzed management’s guidance relative to historic results. Before an earnings release, you should compare management’s guidance to the latest consensus estimate, and your final model assumptions. This will help you understand where the market estimates stand relative to the company’s internal forecast, and whether or not there have been any new market developments since management last gave guidance. You should also take this opportunity to prepare a table to analyze the actual results against your forecast and the consensus, after the earnings report is released. Consider the format in Exhibit 121 below. There are three sections in this table: Section 1 (rows 7 through 16) is designed to keep track of your model estimates prior to the release. The model estimates for the next five quarters are in cells E8 through E16, J8 through J16, M8 through M16, P8 through P16, and S8 through S16, are linked to the model. Note I use five quarters in my table because after the release, the Next Twelve Month (NTM) EPS estimate will be pushed out one quarter, and I like to see how the “new” twelve month forecast has changed in my model. Since the estimates will change as you update the model, you will want to copy and paste as values in this table to maintain the integrity of forecast prior to the earnings release. The consensus estimates before the release are also included in cells D8 through D16, I8 through I16, L8 through L16, and O8 through O16. When the earnings results are released, you can drop the details in cells F8 through F16 and quickly compare how close your model forecast and the consensus estimate was to the actual results. Section 2 (rows 20 through 29) track the changes to the next four quarter forecast after you adjust the model to include the latest actual results from the earnings release, as well as the changes you make to the forecast quarters based on the information provided on the latest earnings conference call. These cells should be linked to the model so the values will carry over directly to this table as you change your model. Section 3 (rows 33 through 42) will compare the actuals from the latest release to the consensus and your model forecast for the latest quarter. Section 3 also compares your new model forecasts for the next four quarters, to your previous model forecasts for these quarters before the release. Section 3 is also setup to make the same comparison for the consensus; however, the new consensus estimates are not designed to pull automatically, so the new consensus will need to be downloaded and updated after analysts have had a chance to update their models including the results from the latest earnings release.

157

Exhibit 121—Prepare Consensus & Model Forecast Analysis for the Next Report

Step 34: Earnings Simulation—Prepare for the Release

In Steps 34 and 35 we will simulate an earnings release. For this exercise we will assume you are a new sell-side equity research associate at an investment bank. Your research team, which covers the shipping industry, consists of just you and the lead analyst Jane Smith. Let’s assume it is the morning of September 17, 2018. FedEx is expected to issue the fiscal first quarter 2019 earnings press release at 4:00pm after the market closes, and will hold the earnings conference call at 5:00pm. Jane calls you into her office to discuss expectations: Good morning. I hope you are excited about today’s press release. Given that this is your first release as a new associate I wanted to go over our approach for today. Some of our most important hedge fund clients have already expressed interested in discussing the FedEx results with me after the release. As you know research regulations prevent me from disseminating information to select clients, so we will need to publish a quick earnings recap note before I can hold calls with them. We can write a comprehensive report later tonight after the conference call, but we must get the recap note out as soon as possible so I can take some client calls before the FedEx conference call begins at 5:00pm. I will work on writing the note. I want you to focus on updating the model to include the latest actuals. Timing will be critical since we will need to load the updated model into our research database before we can publish our recap. To make sure you get the data as soon as possible start refreshing the SEC webpage and the FedEx investor relations site at 4pm on the dot, and take whichever source posts the press release first. When the earnings conference call starts I want you to begin adjusting the forecast based on what management discusses on the earnings call. Be sure to focus on any changes to guidance that management includes in the press release, on the conference call, or in the earnings presentation slides. I’ll prepare our questions for management, but if you have any questions based on your modeling, send them to me and I will raise them on the conference call.

158

Chapter 8: How to Use Your Earnings Model We have a phone meeting scheduled with the FedEx management team at 7:00pm. It will most likely be with the head of investor relations, but the CFO may also attend. Let’s try to have our model forecast finished by then, since this will be our last chance to ask management any questions. After the 7:00pm call I will review your model and tweak it as needed. We should plan to have it re-uploaded into our model database by 9:00pm, assuming there are no big announcements tonight, and we can publish our final comprehensive report at that point. Are you comfortable updating the model? My Experience: Take a minute to think about this exercise before you start. If you are heading for a career in research, this will probably be very similar to the experience you have. How would you answer your bosses question? Are you ready to begin updating models on your own? One of my primary goals for this textbook is to prepare my readers for this moment. Learning to update a model quickly and accurately after an earnings release takes practice. When I was a new associate in research I found this to be among the most stressful tasks, given the tight timeline for completion. One of the most important things you can do as a new associate is to get familiar with the models of the companies you are covering. You may want to practice updating the model using the previous quarter’s press release. This will also help you get familiar with where the information is presented by the company. Some metrics are buried in text, while others are easy to find in tables or slides. Preparation is the key to success when it comes to updating an earnings model. Jane gives you the rest of the day to prepare for the earnings release at 4:00pm. You list out the each of the steps you want to complete before and after the earnings release hits the wires. These steps include the following: Before the Release (Step 34): • Review the relevant leading indicators, and latest news from the investor relations page as a final check to determine if there are additional items to consider in the forecast, ahead of the earnings release. •

Update the valuation metrics to reflect the latest data including: the current share price, last three-month forward Price-to-Earnings multiple, stage-one beta, volatility, interest rates, mean monthly return and standard deviation.

•

Consider the consensus estimate. Are there any points in the consensus earnings forecast which we may be missing in our model? Look to the high-end, and low-end of the consensus range. How has the consensus revenue, EPS estimate, and share price target changed since the last quarter? How does the change in the consensus estimate compare to the change in macroeconomic conditions over the course of the quarter?

•

How does the share value look relative to recent history? Is the Price-to-Earnings ratio above or below the historic average?

•

Save a copy of the existing model in an archive folder.

•

Update the model headers.

•

Change the forecast equations to values, and insert ratio equations.

•

Copy and paste values on the results comparison table (refer to Exhibit 121).

Review last quarter’s press release, earnings transcript, and presentation slides as a refresher on the important topics to watch for in the next press release. After the Release (Step 35): • Input the new actuals into the fiscal first quarter 2019 column. •

•

Compare the latest quarter to the previous projections.

•

Recast the historic results for the pension expense accounting change.

•

Update the references to the forecast and actual columns. For example, when the next forecast period moves from F4Q2018 to F1Q2019 the following changes will be required: o Share count in cell C376 will change to S42. o Net cash/(Debt) per share in cell C368 will move to S351. o The Next Twelve Month (NTM) Earnings Per Share (EPS) reference will move from cells S45 + T45 + U45 + V45 to T45 + U45 + V45 + X45. o The total debt included within cell C384 will move from R263 + R268 + R262 to S263 + S268 + S262. 159

•

Update the forecast periods based on the performance in the latest quarter, insight from the earnings call, and new guidance from management.

•

Add an additional year to the forecast if required.

Step 34a—Perform Final Leading Indicator Check: FedEx does not have specific channel-checks or supply chain earnings results to analyze for an earnings preview. Instead we can look to the following four datapoints as leading indicators for FedEx: 1) macroeconomic data through the quarter, 2) change in the U.S. Dollar (the reporting currency) against other currencies, 3) change in fuel prices, and 4) any press release from the investor relations home page or other news which is relevant for the company. Generally speaking U.S. GDP growth has continued to expand through the quarter, while Europe and Asia have slowed slightly. This may be a one-off dip, and growth may continue to expand again next quarter, so it is probably too early to make drastic changes in our forecast. The quarterly average U.S. Dollar exchange rate has remained relatively stable against most major currencies. Jet fuel prices have ticked higher over the quarter, however, not high enough to raise significant concern. Key Concept 6—Be a Proactive Analyst: If you see the macroeconomic indicators have changed since management last gave guidance, then adjust your projections to reflect this before the next earnings release. A reactive analyst will play it safe, and wait for the earnings release to see if some new company-specific development comes which could offset the macro effect. A proactive analyst will adjust based on the information available, and if they are proved wrong the next day, own it and re-adjust the forecast again. Similarly, if the risk environment for the market has changed ahead of the earnings release, i.e. volatility and the ERP has increased during the quarter, then adjust your valuation target to reflect this before the release. Step 34b—Update the Valuation Metrics—Beta: A re-run of the last 12 monthly returns for FedEx shares against the S&P500 index shows the stage-one short-term beta coefficient has decreased from 1.121 (as of August 31, 2018) to 1.065 (as of September 14, 2018). The stage-two long-term beta regression run against monthly returns through January of 2008, has remained relatively stable at 1.268 (as of September 14, 2018). Both beta coefficients must be updated in the model. For details on how to calculate beta refer to Chapter 5 Step 20. Follow Along in the Spreadsheet: Refer to “File 20-FDX Beta Calculation (Chapter 8 Step 34 F1Q2019 Preview)” for the updated beta calculation prior to the F1Q2019 earnings release. Refer to the "How to Use This Textbook" section of Chapter 1 for instructions on how to access the spreadsheet files. Update the Valuation Metrics—All Other Items: The following metrics have been incorporated into the model to reflect the latest data available as of September 14, 2018. • The estimated equity-to-total capital ratio is 78.6%. •

Market volatility (VIX, quarterly average) is 16.1% (12-month trailing average).

•

The quarterly average 10-year U.S. Treasury rate is 3.02%.

• The three-month average, Next Twelve Month (NTM) forward Price-to-Earnings ratio is 16.7x. Step 34c—Save an Archive Copy of Your Model: Archive and draft model files are important to maintain throughout the process. You may have to revert to a previous version if a file becomes corrupt, or to revisit prior period assumptions. Be sure to save copies of your models at the key stages including before and after an earnings release. Step 34d—Update the Model Headers: This may seem like a basic step (and it is) but it is critically important to do any of the housekeeping items ahead of the release if possible. It will also help you stay organized after the press release is made available. In my models, I shade the historic columns dark gray, and the forecast columns light gray with an “E” in the date to indicate that the column values represent estimates (as discussed in Chapter 2, Step 2). You do not have to use this exact methodology, however, your model should have some indication in the headers distinguishing the forecast columns from the actuals. Before the earnings release, I remove the “E” for the column which will be reported, and change the color to dark gray. I also highlight the column above the header and freeze the panes so I do not lose track of which column I am updating when the release comes out. Refer to Exhibit 122 below for the before and after

160

Chapter 8: How to Use Your Earnings Model headers. To prepare for the release I make these changes to each of the header rows (Income Statement, Segments, Balance Sheet, Cash Flow Statement, etc.).

Exhibit 122—Update the Headers

Step 34e—Change the Forecast Equations to Values: In Chapter 1 we discussed the concept of creating a dynamic model driven by equations which are fed by input assumptions. Remember that the equations generally move downward in the historic columns, and upward in the forecast columns. After the quarterly results are reported, the equations will need to change to calculate the “inputs”, which are used to forecast the next quarter’s results. To demonstrate this point, let’s look at one quarter of historic results and one quarter of projections for the FedEx Ground Segment in Exhibit 123 below. Notice that in the historic column, Ground Average Daily Freight LB in cell Q135 is a hard coded number (since it came from the company’s actual filed results), and the forecast value for this row in cell S135 is an equation based on the blue input cell.

Exhibit 123—Change the Forecast Equations

In the moments leading up to the earnings press release, it is no longer important to maintain the equations for the next quarter. At this point, the primary objective is to prepare the model to consume the new results. To do this, go through the equations which point to an input assumption. Copy and paste these inputs as values. For example, in Exhibit 123 we would paste the values for cells S135, S137, and S139. Start at the top of the Income Statement and work your way down to the end of the model. Be sure to maintain the equations for metrics which should not be hard coded. For example, in the Income Statement the operating profit line should remain as an equation which equals revenue minus operating expenses. If you are having trouble determining which metrics should be values and which should be equations, just look to the last historic quarter and mimic the structure. After you complete the hard coding, you will need to copy over the equations for the input fields. For example, in Exhibit 123 we would change the blue input cell S136 to an equation that calculates the year-over-year growth, similar to cell Q136. Again, look to the last reported quarter if you are having trouble with the equations. 161

There are two benefits of following this process ahead of the release: 1) You will be able to update the model much faster since all the equations will already be setup, and 2) as you fill in the actual values you will instantly see how close your estimates were for each line item. On the other hand, if you kept the original equations in the model rather than saving the values, you would have lost the visibility of your estimates as soon as you input the first value for revenue. This is the result of the ratio based equations which are driven by revenue. Saving the inputs as values before entering any new data, will prevent the loss of visibility as you update the model. Step 34f—Copy/Paste Values on the Results Comparison Table: This is just a reminder of what was discussed in Step 33. Once you begin updating the model it will be too late to perform this step, so keep it on your list of things to do. Step 34g—Review Last Quarter’s Details: If you have time while you wait for the earnings to be released, review the press release, earnings transcript, and presentation slides from the previous quarter. This will help remind you where each of the data points is located so you can be better prepared to update the model quickly. Follow Along in the Spreadsheet: For details on what your model should look like after Step 34 refer to “File 11–Release Prep (Steps 33 to 34)”. The cells highlighted yellow have been saved as values and will need to be updated after the release, as discussed in Step 34c. Refer to the "How to Use This Textbook" section of Chapter 1 for instructions on how to access the spreadsheet files.

Step 35: Earnings Simulation—Updating The Model

It is now 4:00pm and the fiscal first quarter 2019 results have just been published on the FedEx investor relations website. You download the earnings press release and latest statistical book, and notice that the company has not posted the earnings presentation slides to the investor relations page yet. You were expecting the slides to be available, but being the resourceful associate that you are, you decide to take screen-shots of the slides during the webcast of the conference call starting at 5:00pm, just in case you or Jane miss any of the points presented on the slides. You print up the press release and get to work. 35a—Update the F1Q2019 Actuals: Manually typing in the values from the press release may seem like a primitive approach to updating the model. There are many software companies trying to automate the process, however, they rely on technologies which have not been fully developed. For example, as discussed in Step 3, there is an XBRL taxonomy process managed by the Financial Accounting Standards Board (FASB), which is used by third party data service providers to populate financial data automatically. One major limitation to this process is that companies are not yet required to tag the details in their 8-K SEC filings (quarterly earnings press releases are published as 8-K filings). This means there is no standard setting body overseeing the automation of the quarterly press release data. When you consider this point along with the fact that all parties receive the press release at the exact same moment, there is currently no faster, more accurate, and reliable way of accessing the data and getting it into your model then the old fashion way of taking it from the press release yourself. There are many groups working on improving this process, so this may not be the case in the not so distant future. For the time being the practice used by most sell-side research teams is to manually type in the results. For an example of how the columns appear before and after the release refer to Exhibit 124 below. Notice that none of the blue input assumptions appear in column “S(new)” since the results have already been reported. Cells S(new)152, 154, 156, 158, 161, 162, and 163 are all now input values from the press release. Cells S(new)153, 155, 157, 159, 164, 165, and 166 are now equations based on the newly reported results.

162

Chapter 8: How to Use Your Earnings Model

Exhibit 124—Update the Actuals Column: Freight Segment Example

Follow Along in the Spreadsheet: Refer to “File 12–F1Q2019 Actuals (Step 35a)” for details on what your model should look like after adding the latest reported actuals as discussed in Step 35a. Refer to the "How to Use This Textbook" section of Chapter 1 for instructions on how to access the spreadsheet files. 35b—Compare the F1Q2019 Actuals to Your Previous Projections (Key Concept 7—Own Your Projections): One of the most important things you can do as an analyst, is to hold yourself accountable to your projections. There are many behavioral finance concepts working against analysts. We tend to be overconfident in our ability to predict future results, and often do not fully incorporate new data into our existing projections. We also tend to focus on areas where we are correct in our predictions, and tend to forget the areas where we are wrong. Try to keep your psychological biases in mind as you compare your projections to the reported results of the company you are covering. Celebrate the accurate points in your projections, but recognize when you are wrong. If necessary correct your mistake and move on with your new forecast. For this exercise we will revisit a portion of the chart from Exhibit 121, which has been updated to include the latest release in Exhibit 125 below. Note that we have not completed our future period forecast so we can only compare the F1Q2019 actual results to our original forecast. Our revenue estimate was relatively close to the actual results within 1% difference from the actuals; however, our non-GAAP operating profit margin of 8.3% was much higher than the actual non-GAAP operating margin of just 7.0%. We were expecting some puts and takes between the operating margin line and net income given the expected change in accounting for pension expense, but even after considering this point we still underestimated the operating expenses.

Exhibit 125—Compare F1Q2019 Actuals to the Original Forecast

163

To dig into the specific reason for the difference, open a previous version of the model and compare each line until you find the main source of the difference between the actual reported result and the forecast. In this case there were two primary drivers of the $289M difference between our operating profit and the actual operating profit: 1) Fuel Expense: Approximately $100M lower in our forecast compared to the actual. What Went Wrong in Our Forecast? We had projected the Express Segment would use 287 million gallons of jet fuel, however 315 million gallons were actually used. There was also a slight difference in the projected fuel price (we estimated $2.25 per gallon versus the actual of $2.30), but the primary driver of the overall difference was due to the number of gallons. To project the number of gallons I used a ratio of gallons to Express ADV, from the last reported quarter. I then applied that ratio to the ADV I was projecting in F1Q2019 to get to the projected number of gallons. Had I used the comparable quarter, F1Q2018, instead of the last reported quarter the projected number of gallons would have been 330M, which would have been much closer to the F1Q2019 actual result. This is due to the fact that the comparable quarter captures the seasonality effects of fuel efficiency (meaning peak times may be more fuel efficient compared to non-peak, or vice versa). How to Fix the Issue in Our New Projection: Change the blue input cell for jet fuel gallons in the Express segment to apply the ratio from the prior comparable quarter rather then the last reported quarter. 2) Salary Expense: Approximately $200M lower in our forecast compared to the actual. What Went Wrong in Our Forecast? This line item is a bit more convoluted compared to the fuel expense line, since we do not carve out salary expense separately from other operating expenses in our model. Instead we project all other operating expenses, and allocate back to each line based on the average allocation from the comparable quarter (in this case F1Q2018). The problem with this approach in our F1Q2019 forecast is that it underestimated the impact of the two employee wage effects described in the F1Q2019 press release: • Substantially higher variable compensation accruals compared to the F1Q2018 quarter which were impacted by the TNT cyberattack, and • The accelerated wage increase due to the company’s decision to pass on some of the positive effect of the Tax Cuts and Jobs Act (TCJA) U.S. tax legislation benefit to employees. How to Fix the Issue in Our New Projection: For F2Q2019, and F3Q2019, increase the expected percentage of total operating expense to revenue (for the Freight and Ground Segments only since the Express forecast was very close to the actual), and increase the allocation to the salaries and wage line item in the Income Statement. Notice I will not be adjusting the F4Q2019 forecast since the TCJA effect began in F4Q2018, so the comparable period-based forecast for that quarter should be relatively reasonable. 35c—Recast the Historic Results for the Pension Expense Accounting Change: It is very common for companies to adopt new accounting standards, alter their reportable segments, change their product detail disclosures, or change other items which would result in a needed change to the model. In this case FedEx has adopted ASU 2017-07 Compensation—Retirement Benefits: Improving the Presentation of Net Periodic Pension Costs and Net Periodic Postretirement Benefit Cost, which we discussed in Chapter 4, Step 15h. In Step 35c we will recast our historic results to include the impact of this accounting change. FedEx issued an 8-K filing with the SEC which shows how to recast past results under the new standard, refer to Exhibit 126 below for a summary. In this step, we will use this filing to update our model. The recast is a three step process: First, reduce the operating income for each segment in Exhibit 126 by increasing the “all other operating expense” line for the amount in Exhibit 126. Second, insert a new line in the Income Statement for other retirement plan income/(expense), and put the offset of the amount from the first step into the line, so there is no net impact on net income. Third, adjust the non-GAAP items using the non-GAAP recast details from the latest press release.

164

Chapter 8: How to Use Your Earnings Model

Exhibit 126—Recast Historic Results for the Impact of ASU 2017-07

Pitfall: Sometimes companies will recast historic results for accounting changes to provide an apples-to-apples comparison under the new standard; however, if the accounting change is made on a “prospective basis” meaning it is applied only to future financial statements, then the recast values will not tie to the financials from the SEC website. In this case the impact of ASU 2017-07 was applied on a “retrospective basis” meaning historic results were revised to account for the service cost component of the pension expense. You can read the ASU from FASB’s website for the guidance on whether the ASU should be applied prospectively or retrospectively. 35d—Update the Future Quarter Forecast: Now that you have input the latest quarterly results, the values in your forecast quarters have probably changed quite a bit. This is driven by the fact that the future growth rates are based in part on the historic results, which have changed after the latest release. For example, we projected the F1Q2019 segment ADV would grow based on average growth rates from the last few reported quarters. Now the F1Q2019 actuals are available and the F2Q2019 ADV projections are setup to recalculate based on a new average, which includes the F1Q2019 actuals. As a result, it may be necessary to adjust our future forecast to align it with our new expectations. We must also consider incorporating the new guidance provided by management into our forecast, and input the

165

updates based on our analysis of our previous forecast from Step 35b. Here are the primary changes I made to my forecast: •

Changed the Express jet fuel projection as discussed in Step 35b, and kept a fuel efficiency factor which reduces the gallons used by 2.5% in fiscal year 2019, and 5% in 2020 to reflect the benefit of airplane fleet modernization.

•

Increased the Freight and Ground “other operating expenses” by 50 basis points in F2Q2019, and 25 basis points in F3Q2019 as discussed in Step 35b.

•

Tweaked the Ground Segment operating expenses in fiscal year 2020.

Follow Along in the Spreadsheet: Refer to “File 13–Pension Recast & Updated Forecast (Steps 35c & 35d)” for details on what your model should look like after the recast of the pensions expense and the new forecast estimates. Refer to the "How to Use This Textbook" section of Chapter 1 for instructions on how to access the spreadsheet files. 35e—Review and Update the New Price Target: Now that we have an additional quarter of results in our model, there are two changes we must make to our valuation section: 1) Pushout out the cells which represent the Next Twelve Month (NTM) EPS reference in the market multiple section by one quarter. 2) Update the link to the latest reported net cash/(debt) per share in the market multiple and DCF valuation sections. 3) Update the reference to the total shares outstanding in cell C374. 4) Update the reference to the debt balances in the equity-to-total capital ratio in cell C382. Conceptually the 12-month price target is a forward looking metric. Using the valuation method discussed in Chapter 7, we base our valuation on the next 12-month EPS estimate; therefore, we must constantly push the references in our model forward. Refer to Exhibit 127 below for a demonstration of how to perform these two steps. Notice that in the previous version of our model, the market multiple valuation was based on the NTM EPS estimate of cell S44 + T44 + U44 + V44. Now we must update the reference in cell C364 to equal (T44 + U44 + V44 + X44) × the Price-to-Earnings ratio in cell C360 + the new reported net debt per share in cell V346. Now, on the night of the earnings release when we pushout the NTM EPS estimate reference, the three-month average NTM Price-to-Earnings ratio is not directly relevant since it is referring to a different twelve month period. To reflect this point you may choose to haircut the ratio slightly based on how the reported results compared to your forecast, and whether or not the updated guidance was different from your expectation. Before the F1Q2019 release, we were using the three-month average Price-to-Earnings ratio, and the results were slightly worse then what we were modeling, so let’s decrease our Price-to-Earnings ratio to the three-month low of 15.7 times.

166

Chapter 8: How to Use Your Earnings Model

Exhibit 127—Pushout the Next Twelve Month EPS and Net Cash/(Debt) Per Share

Once a year we must also add an additional year to our forecast so that we have approximately five years in our stageone DCF valuation, pushout the reference to our free cash flow estimates by one year, and check the reasonableness of the effective tax rate and after tax cost of debt in the DCF valuation section. Technically you could change the discount calculation to reflect a portion of the year rather than a full year of discounting, but this implies a very high degree of precision in our DCF valuation estimate. Instead I keep it simple and assume a full year of discounting. Then push-out the next year at some point between the second and third quarter. Follow Along in the Spreadsheet: Refer to “File 14-NTM EPS (Step 35e)” for details on what your model should look like after the NTM EPS and net cash/(debt) per share cell reference updates. Refer to the "How to Use This Textbook" section of Chapter 1 for instructions on how to access the spreadsheet files. …And Now Back to the Earnings Simulation: It is now 9:00pm. You sent your model to Jane fifteen minutes ago for her review. She calls you into her office to discuss: Good job on the model. I see we were a bit low on the fuel and wage expenses and you have corrected for it and recast the historic results to include the impact of the pension expense accounting change. I have incorporated a few of the data points from your model into my report which is ready to be published. Please submit the model into our database to complete the publishing process. After that please save down a “values only” version [copy and paste values without the embedded equations] for our clients to access. Great job today. Try to get some studying in tonight for your FINRA exams, then get some sleep. We have another earnings report to work on tomorrow!

Step 36: Regular Model Maintenance

Everyday, new information is released which may impact your model. Economic data may be published which could impact the Equity Risk Premium (ERP) used in the Discounted Cash Flow (DCF) share valuation. Company specific news may also be released, such as a new product release, or the issuance of debt or equity, which would need to be incorporated into the model.

167

You should use judgment when deciding how often to update your model. If you have a long-term horizon and you are modeling a company as a passive hobby, you may not feel the need to update your model for changes throughout the quarter. Instead you could just update the model when new earnings are released. On the other hand, if you are a sellside analyst and your clients are relying on your model, you should keep it as up-to-date as possible. Going forward I will add Appendix entries to the end of this textbook in future editions each time there is a new quarterly earnings release. These updates will cover the changes I make to the FedEx model, and will focus on the following items: 1) Updates for the latest reported earnings results. 2) Recalibration of the forecast to incorporate the latest company and market developments, to meet the latest guidance from management, and consideration of the consensus estimates. 3) Updates to the Equity Risk Premium (ERP) model which will include the latest quarter of interest rate and volatility data, and any beta adjustments. 4) Updates to the Price-to-Earnings ratio for the market multiple valuation. 5) Updates to the monthly mean return and standard deviation for the target price band. Follow Along in the Spreadsheet: Refer to “File 15-FDX Model (Step 36 after F1Q2019)” for details of the changes made to the forecast after the Fiscal first quarter earnings release. Refer to the "How to Use This Textbook" section of Chapter 1 for instructions on how to access the spreadsheet files.

168

CHAPTER 8 WRAP-UP Takeaways •

Key Concept 5—Models are Living Tools: After a model has been built, it should not sit idle. It should be used to perform analysis, debate potential company outcomes, and explore market possibilities. Most importantly, it must be maintained as new information is released.

•

Key Concept 6—Be a Proactive Analyst: A proactive analyst takes risks and adjusts his or her forecast based on the information available. A reactive analyst plays it safe, waits for the subject company to issue results, and potentially revise guidance before going out on a limb with their forecast.

•

Key Concept 7— Own Your Projections: One of the most important things you can do as an analyst, is to hold yourself accountable to your projections. There are many behavioral finance concepts working against analysts. We tend to be overconfident in our ability to predict future results, and often do not fully incorporate new data into our existing projections. We also tend to focus on areas where we are correct in our predictions, and tend to forget the areas where we are wrong. Try to keep your psychological biases in mind as you compare your projections to the reported results of the company you are covering. Celebrate the accurate points in your projections, but recognize when you are wrong. If necessary, correct your mistake and move on with your new forecast.

Concept Quiz Instructions: Use the following information to answer questions 1 through 5. Ice Cream Factory Inc (ICF) sells ice cream cones across the country. The company’s sales tend to peak in the summer months, and decline in the winter. The March quarter tends to be volatile, depending on whether or not the spring arrives early or late. The weather was mild in the first quarter of 2018, so the March 2018 quarterly numbers represent an average result for a first quarter. Fiona Jane covers ICF and would like to run scenario analysis ahead of the first quarter 2019 earnings conference call. She compiles the following earnings model for her analysis:

169

1) Fiona starts by creating an upside scenario for the March quarter. What estimate is she most likely to use for gross margin in this scenario? A. 50% B. 51% C. 55% 2) What gross margin estimate would Fiona select for a downside scenario in the March 2019 quarter? A. 45% B. 49% C. 50% 3) What estimate is Fiona most likely to use for average revenue per cone in an upside scenario? A. $1.50 B. $1.55 C. $1.75 4) Fiona basis her share valuation on a Next Twelve Month (NTM) forward Price-to-Earnings ratio. After the 1Q2019 earnings are released, which cells will she reference in her valuation calculation? A. I12, J12, K12, M12 B. I12, J12, K12, L12 C. H12, I12, J12, K12 5) Fiona runs sensitivity analysis on cone growth (year-over-year) and gross margin. What is the most accurate assessment based on the table below?

A. A 50-basis point decrease in gross margin will result in a $0.01 to $0.02 decrease in EPS, and an increase in cone growth of 5 percentage points will increase EPS by $0.01 to $0.02. B. A 50-basis point increase in gross margin will result in a $0.01 to $0.02 decrease in EPS, and an increase in cone growth of 5 percentage points will increase EPS by $0.01 to $0.02. C. A 50-basis point decrease in gross margin will result in a $0.01 to $0.02 decrease in EPS, and an increase in cone growth of 5 percentage points will decrease EPS by $0.01 to $0.02.

Concept Quiz Answers 1) A. Historic results show that gross margin increases with increases in volume. For example, the June 2018 quarter gross margin was higher than the December 2018 quarter (most likely due to discounts for purchasing supplies in bulk). Since the March quarter of 2018 was an average first quarter, the gross margin for an upside scenario will likely be higher than the 50% gross margin from March 2018, but probably not higher than the 53% seen in the peak quarter of June 2018; therefore, 51% is most likely the gross margin selected for an upside scenario.

170

Chapter 8 Wrap-Up 2) B. Historic results show that gross margin decreases with decreases in volume. For example, the December 2018 quarter gross margin was lower than the June 2018 quarter. Since the March quarter of 2018 was an average first quarter, the gross margin will likely be lower than the 50% gross margin from March 2018 for a downside scenario, but probably not lower than the 46% seen in the trough quarter of December 2018; therefore, 49% is most likely the gross margin selected for a downside scenario. 3) C. The most likely average revenue per cone in the upside scenario is $1.75, which is the only choice that is higher than the average from the first quarter of 2018. 4) A. Prior to the release the Next Twelve Month EPS estimate is H12 + I12 + J12 + K12. After the release this fourquarter period is pushed out by one quarter I12 + J12 + K12 + M12 (be careful not to reference column L, which represents a full year). 5) A. A 50-basis point decrease in gross margin will result in a $0.01 to $0.02 decrease in EPS, and an increase in cone growth of 5 percentage points will increase EPS by $0.01 to $0.02. Follow Along in the Spreadsheet: Refer to “File 19-Chapter Wrap-Ups”, worksheet “Chapter 8” to access the Excel file used for the questions above. Refer to the "How to Use This Textbook" section of Chapter 1 for instructions on how to access the spreadsheet files.

171

APPENDIX 1: USING REGRESSION ANALYSIS TO PREDICT EARNINGS Step R1: Select the Data Set Step R2: Run the Regression Step R3: Test the Model

Step R4: Analyze the Output Step R5: Perform Back-Testing Step R6: Model Limitations

Appendix 1 Overview: Regression analysis is a mathematical process which estimates the relationship between two or more variables. As equity analysts, we must forecast our expectations for future periods. Regression models can be helpful in the projection process when reliable data is available ahead of a company’s press release. In this appendix, I will demonstrate how to use regression models for projections with examples from the FedEx Ground Segment. Notice in this section the step count has started over at “R1” (Regression-1). This is because the financial model has been completed in the previous chapters. The steps in this appendix are only relevant for those who wish to use regression analysis to predict inputs into their earnings model (or to put it another way, estimate the inputs in our blue cells). The reason this section appears in the appendix and not in the forecasting and calibration chapter, is that the limitations of regression, reduce the usefulness of the approach. Due to these limitations (explained in Step R6) most analysts do not employ regression statistics in their earnings models. Still in some cases where high correlations exist between variables, and the reporting environment is stable, regression can be useful.

Step R1: Select the Data Set

The first step is selecting the appropriate variables to use in the regression model. It can be difficult to find variables with strong explanatory power. Ideally, we should select metrics with limited impact from multiple forces. Given the nature of financial data, this is much easier said then done. We must select a dependent variable from our model which is “dependent” on an external factor (or multiple factors), which will be our independent variable(s). For example, our FedEx earnings model has four primary segments: FedEx Express, FedEx Ground, FedEx Freight, and Services and Other. Let’s assume that we would like to project the operating income for the FedEx Ground Segment. We could run a regression of the FedEx Ground Segment operating income, against a series of macroeconomic variables; however, there are many forces at play on the net operating income. These include: changes in profitability due to changes in fuel prices (which could also be impacted by changes in fuel efficiency overtime), wages, depreciation (which brings changes in capex and fixed assets into the equation). In addition, there are larger forces at work on the revenue line, such as changes in competitive pressure over time, market share, foreign currency rates, changes in financial reporting, business acquisitions/divestitures, and the effect of changes in underlying customer manufacturing trends (proximity to customers and the effect on shipping). All of these factors could distort our regression model. 172

Appendix 1: Using Regression Analysis to Predict Earnings Instead let’s try to isolate just one of the FedEx Ground factors. Our earnings model requires an estimate (blue input cell) of Average Daily Freight Pound (ADFlb) for the FedEx Ground Segment. We will use this as our dependent variable for this regression example. Next, we must select metrics which can assist in predicting the ADFlb. The FedEx management team has already provided support for this item in their macroeconomic outlook which highlights the key factors they track and use to make decisions about their business. The four main variables that management tracks are: 1) U.S. Gross Domestic Product (GDP), 2) Global GDP, 3) U.S. industrial production, and 4) U.S. consumer spending (refer to Step 15k for details). Since we are focused on the FedEx Ground Segment we will test the relationship of ADFlb and the three U.S. variables. Before we get started we should determine the most appropriate method of comparing the variables. Should we compare the quarter-over-quarter change in ADFlb to the quarter-over-quarter change in GDP, or year-over-year? To answer this question we must determine if there is any seasonal effect in the FedEx Ground ADFlb which may not be present in the macro variables. Management explains the impact of seasonality in the 10-K under the Management’s Discussion and Analysis (MD&A) section. For FedEx Ground, the fall is the busiest shipping season. We will test the accuracy of this point in Step R2. For now we can concluded that given the seasonality in the FedEx Ground business, it is more appropriate to compare yearover-year changes in the variables rather than quarter-over-quarter.

Exhibit 128—FedEx Corp 10-K Description of Seasonality

Source: SEC.gov, FedEx Corp 10-K, filing date July 16, 2018, retrieved September 24, 2018.

Data Limitation: We will discuss the list of general regression limitations in Step R6, however it is important to first understand the inherent data limitation in the method we are developing in this step. FedEx reports under fiscal quarters which do not align with the calendar quarters used for the macroeconomic variables. There are two approaches we could take to address this. First, we could attempt to adjust the FedEx results to align with the calendar quarters, although there is no way to prove that we have a reasonable approach to accomplish this, so we will not be using this approach. The second approach, which we will use, is to regress the FedEx fiscal first quarter results (which end August 31st), against the second calendar quarter macroeconomic variables (which end June 30th). There is some dislocation in the time periods, however, in regression analysis we often introduce lags. In this case, we are introducing a two-month lag between the FedEx results and the macro variable. If we did not use this approach then the model would be useless for forecasting since the economic variables we would use to predict the FedEx data, would only be available after the FedEx quarterly earnings release. The real question is whether or not the macro variables have any predictive power in the one month overlap, or is shipping a leading indicator which could predict economic statistics, rather than economic statistics predicting shipping. We will soon find out as we test the regression output. Data Preparation: We start by pulling the macroeconomic details. GDP and personal consumption come from the U.S. Department of Commerce, Bureau of Economic Analysis (https://www.bea.gov), and Industrial production which is available on the U.S. Federal Reserve’s Economic Data website (https://fred.stlouisfed.org). Next, convert the data into year-over-year percentage changes, and save the details with the FedEx Ground ADFlb statistics (refer to Exhibit 129 below).

173

Exhibit 129—Data for the Regression

Step R2: Run the Regression

The next step is to run the regression models for the various metrics to test whether or not statistically significant relationships exist between the dependent and independent variables. We will run the following five regression models: 1) ADFlb/Peak Season: A simple linear regression of FedEx Ground ADFlb, against a dummy variable which represents the FedEx fiscal second quarter to test the point made in the 10-K that the fall quarter tends to be the peak delivery season for FedEx Ground. 2) ADFlb/GDP: A simple linear regression of FedEx Ground ADFlb (% change), against the year-over-year percentage change in U.S. GDP. 3) ADFlb/Production: A simple linear regression of FedEx Ground ADFlb (% change), against the year-overyear percentage change in U.S. industrial production. 4) ADFlb/Consumption: A simple linear regression of FedEx Ground ADFlb (% change), against the yearover-year percentage change in U.S. personal consumption. 5) ADFlb/GDP & Consumption: A multiple regression of FedEx Ground ADFlb (% change), against the yearover-year percentage change in U.S. GDP, and the percentage change in U.S. personal consumption. I will be using Excel for this analysis but feel free to use your favorite statistical software. If you have not installed the Excel Data Analysis add-in, refer to the installation instructions in Chapter 5, Step 20b. Regression 1—ADFlb/Peak Season: The following steps (R2a through R2f) explain how to run the first regression: Step R2a: Convert the fall peak season into a “dummy” variable, which means each observation will have a value of either zero or one. A one will represent the fall peak shipping season, so each fiscal second quarter will have a value of one, and all other quarters will be set to zero. In Excel, on the “Data” tab, click on “Data Analysis”, then “Regression”. Select the FedEx Ground’s ADFlb column for the Y input range (dependent variable), and the FedEx Ground peak cycle dummy variable column for the X input range (independent variable). Step R2b: Check the “Labels” box, which indicates that the Y and X ranges include the data headings. Step R2c: For this regression, we want to set the constant in the regression equation equal to zero. To do this check the “Constant is Zero” box. You can try running the regression with this box unchecked to see which result is better. The constant is the term before the beta coefficient in the regression equation, which will be explained in greater detail in Step R4.

174

Appendix 1: Using Regression Analysis to Predict Earnings Step R2d: Select the “Output Range” where you would like the regression results to be populated. In my Excel file I selected a new worksheet. Step R2e: Click the residual details you would like to analyze, which will be discussed in Step R3. Step R2f: Click “OK” to run the regression. Follow Along in the Spreadsheet: Refer to “File 18-Regression Models (Appendix 1)”. The details of each economic variable are saved in individual worksheets within the file named: “US GDP”, “Industrial Production (Qtrly)”, and “Personal Consumption”. The summary statistics chart is in the worksheet named “Data”, and the output of each of the regressions is saved with the following names: “#1 Peak”, “#2 GDP”, “#3 Production”, “#4 Consumption”, and “#5 Multiple”. Refer to the "How to Use This Textbook" section of Chapter 1 for instructions on how to access the spreadsheet files. Regressions 2 through 5: The remaining regressions follow similar steps as the ADFlb/Peak Season regression except for the selection of the dependent and independent variables (Step R2a). •

Regressions 2—ADFlb/GDP: Set the Y input variable range equal to the FedEx Ground ADFlb percentage change year-over-year. Set the X input variable range equal to the year-over-year percentage change in U.S. GDP. Steps R2b through R2f remain the same.

•

Regressions 3—ADFlb/Production: Set the Y input variable range equal to the FedEx Ground ADFlb percentage change year-over-year. Set the X input variable range equal to the year-over-year percentage change in U.S. industrial production. Steps R2b through R2f remain the same.

•

Regressions 4—ADFlb/Consumption: Set the Y input variable range equal to the FedEx Ground ADFlb percentage change year-over-year. Set the X input variable range equal to the year-over-year percentage change in U.S. personal consumption. Steps R2b through R2f remain the same.

•

Regressions 5—ADFlb/GDP & Consumption: Set the Y input variable range equal to the FedEx Ground ADFlb percentage change year-over-year. Since we are running a multiple regression for this example, we must align the two variables in two consecutive columns and select both the U.S. GDP and U.S. personal consumption as the X input variable. Steps R2b through R2f remain the same.

Step R3: Test the Regression Model

Before we can draw any conclusions about the correlation between the variables, we must test the validity of the regression models. To do this, we perform a series of statistical reviews, including the following: Step R3a—t-Stat and P-Value Review: These two statistics are used to test whether or not the relationship, or correlation, between the variables is different from zero. If the t-stat is greater than +2.74 or less than -2.74, then a statistically significant relationship exists at the 99% confidence interval (also known as the 1% level of significance). The 2.74 figure comes from the Critical t-value of a distribution chart representing 0.005 probability in each tail with 32 degrees of freedom (number of observations minus one, in our FedEx example 32 -1 = 31). Exhibits 130 through 134 show the Excel regression output for each set of variables. All of the simple linear regressions (regressions 1 through 4) have t-stats greater than 2.74; therefore, a statistically significant relationship exists between the variables. A pure statisticians would not say that a relationship between the variables exists, but that we can reject the notion that no relationship exists, or we reject the null hypothesis. The t-stats in the multiple regression of GDP with Consumption (Regression 5 in Exhibit 134) are not greater than 2.74, therefore we cannot reject the null hypothesis in this case. The P-value represents the lowest level of significance the regression model can reach while still showing a statistically significant relationship. The results of the P-Value review are similar to the t-stat. The P-value is close to zero for each regression (numbers 1 through 4), which means we would reach the same conclusion even at a confidence interval greater than 99%.

175

Exhibit 130—Regression 1 ADFlb/Peak Season: P-Value & t-Stat

Exhibit 131—Regression 2 ADFlb/GDP: P-Value & t-Stat

Exhibit 132—Regression 3 ADFlb/Production: P-Value & t-Stat

176

Appendix 1: Using Regression Analysis to Predict Earnings

Exhibit 133—Regression 4 ADFlb/Consumption: P-Value & t-Stat

Exhibit 134—Regression 5 ADFlb/GDP & Consumption: P-Value & t-Stat

Step R3b—Breusche-Pagan Test: To make sure the independent variables do not explain variation in the residuals (a situation known as heteroskedasticty), we run the Breusche-Pagan test by regressing the independent variable against the squared residuals. The results of each of the regressions is shown in Exhibits 135 through 138. For the regressions of ADFlb/Peak Season, and ADFlb/Production, the Breusche-Pagan R squares are relatively low with high p-values, and t-stats between the critical-t; therefore, heteroskedasticity is not an issue with these models. To put it another way, the correlation between the peak season dummy variable or U.S. production and the squared residuals is not statistically different from zero. For the regressions of ADFlb/GDP, and ADFlb/ Consumption, the Breusche-Pagan R squares are relatively low, however, the p-values are nearly zero, and t-stats are above the critical-t values, which is an indication that some level of heteroskedasticity exists.

177

Exhibit 135: Breusche-Pagan Regression—ADFlb/Peak Season

Exhibit 136: Breusche-Pagan Regression—ADFlb/GDP

Exhibit 137: Breusche-Pagan Regression—ADFlb/Production

178

Appendix 1: Using Regression Analysis to Predict Earnings

Exhibit 138: Breusche-Pagan Regression—ADFlb/Consumption

In addition to the Breusche-Pagan test, we can also plot the residual with a linear trend line showing the R Square to determine if the independent variables explain variation in the residuals (this was the box we checked in Step R2e). Using this methodology there appears to be some correlation in the residuals for each regression except ADFlb/Peak Season, which has a trend line and R Square close to zero (indicating that the residuals and independent variable are not correlated).

Exhibit 139: Residual Plot—ADFlb/Peak Season

Exhibit 140: Residual Plot—ADFlb/GDP

179

Exhibit 141: Residual Plot—ADFlb/Production

Exhibit 142: Residual Plot—ADFlb/Consumption

Step R3c—Serial Correlation Test: Serial correlation is the relationship of a variable with itself over time. If serial correlation is present, past observations will influence future results. To ensure the model does not exhibit signs of serial correlation, run the Durbin-Watson (DW) test, by calculating the DW statistic. To calculate the Durbin-Watson statistic, first square the residuals from the regression output. Then, sum the difference of the actual residuals, and the residuals on a one-period lag squared. Next, divide the result by the squared residuals. Durbin-Watson statistics range in value between 0 and 4. A value near 0 indicates positive serial correlation, a value near 4 indicates negative serial correlation, and a value near 2 indicates no significant serial correlation. Note that these ranges represent approximations. For exact measures refer to a DW Significance Table. The following is a breakdown of the Durbin-Watson statistics for Regressions 1 through 4: •

Regressions 1—ADFlb/Peak Season: Durbin-Watson stat = 1.999. No significant serial correlation.

•

Regressions 2—ADFlb/GDP: Durbin-Watson stat = 2.421. Serial correlation is not likely an issue.

•

Regressions 3—ADFlb/Production: Durbin-Watson stat = 2.714. Serial correlation is not likely an issue.

•

Regressions 4—ADFlb/Consumption: Durbin-Watson stat = 2.676. Serial correlation is not likely an issue.

•

Regressions 5—ADFlb/GDP & Consumption: Durbin-Watson stat = 2.48. Serial correlation is not likely a significant issue.

Follow Along in the Spreadsheet: Refer to “File 18-Regression Models (Appendix 1)” for the details of the Durbin-Watson statistics which are calculated in each regressions worksheet: “#1 Peak”, “#2 GDP”, “#3 Production” and “#4 Consumption”. Refer to the "How to Use This Textbook" section of Chapter 1 for instructions on how to access the spreadsheet files. Step R3d—The Normal Distribution Assumption: To test that the data does not violate the assumption of a normal distribution, calculate the skew and kurtosis. A distribution is considered to be skewed if more observations are found on one side of the mean. A perfect normal distribution would have a skew of 0. A skew of +/-1 can be considered 180

Appendix 1: Using Regression Analysis to Predict Earnings approximately normally distributed. Kurtosis is the measure of the distribution’s peak. A kurtosis of less than 3 can be considered approximately normally distributed. Excel has equations for these formulas: =SKEW(residuals) and =KURT(residuals). If you feel you need additional evidence, you can also run a Chi-sq test, and plot the residuals in a histogram to see if they appear to be normally distributed. The following is a breakdown of the skew and kurtosis statistics for Regressions 1 through 4: •

Regressions 1—ADFlb/Peak Season: The skew is -0.20 (which is below +/-1), and the kurtosis is -0.95 (which is below 3). These statistics show that the sample is approximately normally distributed.

•

Regressions 2—ADFlb/GDP: The skew is 0.24 (which is below +/-1), and the kurtosis is -0.58 (which is below 3). These statistics show that the sample is approximately normally distributed.

•

Regressions 3—ADFlb/Production: The skew is 0.51 (which is below +/-1), and the kurtosis is -0.67 (which is below 3). These statistics show that the sample is approximately normally distributed.

•

Regressions 4—ADFlb/Consumption: The skew is 0.29 (which is below +/-1), and the kurtosis is -0.30 (which is below 3). These statistics show that the sample is approximately normally distributed.

•

Regressions 5—ADFlb/GDP & Consumption: The skew is 0.25 (which is below +/-1), and the kurtosis is -0.50 (which is below 3). These statistics show that the sample is approximately normally distributed.

Step R3e—Model Stability: To test the stability of the model, remove two random observations from the sample and re-run the regression. The new regression should pass all the prior tests with a similar standard error compared to the first model. This will prove that the regression model is fairly stable. The first set of tests from Step R3a through R3e above are relevant for all of our regressions. For the multiple regression of ADFlb/GDP & Consumption there are a few additional tests we must perform. These include: Step R3f—F-Stat Review: The F-statistic is similar to the t-statistic, except it is used for multiple regressions. If the F-stat is greater than or less than the critical F-values, then a statistically significant relationship exists. Critical F-values are published by degrees of freedom on a critical F-chart, which is available in most statistics textbooks. In the ADFlb/GDP & Consumption model, the F-stat of 34 (refer to Exhibit 134) is well above the critical F-value of 2.07; therefore, a statistically significant relationship exists between the variables, although in this case the pvalue and the t-stat of the individual variables were not significant. Step R3g—Significance F: The Significance F for a multiple regression has the same interpretation as the P-value for a simple linear regression. It represents the lowest level of significance the model can reach while still showing a statistically significant relationship. In the ADFlb/GDP & Consumption model, the Significance F is close to zero, although in this case the p-value and the t-stat of the individual variables were not significant. Step R3h—Multicollinearity: To test for multicollinearity, regress the two independent variables against each other to ensure they are not correlated. In the case of GDP and Consumption, the resulting regression between the independent variables has a t-stat of 29, a P-value of near zero and an R Square of 96%. This indicates that the correlation between the two independent variables is statistically significant at the 99% confidence interval; therefore, multicollinearity is a significant issue.

Step R4: Analyze the Output

If the results of our statistical tests in Steps R1 through R3 prove that our regression models are valid, we can analyze the model output and use the details to predict future period ADFlb, which can be included in the earnings model built in the previous chapters. There are three primary metrics from the regression output screens (Exhibits 130 through 134) which we use for forecasting: 1) R Square also known as the coefficient of determination, represents the amount of variation in the dependent variable, explained by the independent variable. 2) Coefficients used to create the regression model equation, which will be used to project future period ADFlb. 3) Standard Error which is the standard deviation of the sampling distribution used to give the predicted range of the regression model (similar to how we used the standard deviation of share returns to created a target share price band in Chapter 7 Step 30).

181

While each of the five regressions we have tested show some form of weakness, we will still analyze the output for each in this step, then select one of our five regression models which has proved to be the strongest prospect for forecasting relative to the others. •

Regressions 1—ADFlb/Peak Season o The adjusted R square of 0.41 (refer to Exhibit 130) indicates that 41% of the variation in FedEx Ground’s ADFlb, is explained by the dummy variable for the fall peak season. o The resulting regression model equation based on the coefficient in Exhibit 130 is: ADFlb quarter-over-quarter % change = 0.110 x Peak season dummy variable

o •

Regressions 2—ADFlb/GDP o The adjusted R square of 0.66 (refer to Exhibit 131) indicates that 66% of the variation in the yearover-year percentage change in FedEx Ground’s ADFlb, is explained by the year-over-year percentage change in U.S. GDP. o o

•

•

The resulting regression model equation based on the coefficient in Exhibit 131 is: ADFlb year-over-year % change = 1.747 x U.S. GDP year-over-year % change The predicted ADFlb range based on the model standard error is +/- 4.7%.

Regressions 3—ADFlb/Production o The adjusted R square of 0.23 (refer to Exhibit 132) indicates that 23% of the variation in the yearover-year percentage change in FedEx Ground’s ADFlb, is explained by the year-over-year percentage change in U.S. Production. o

The resulting regression model equation based on the coefficient in Exhibit 132 is: ADFlb year-over-year % change = 1.421 x U.S. industrial production year-over-year % change

o

The predicted ADFlb range based on the model standard error is +/- 7.2%.

Regressions 4—ADFlb/Consumption o The adjusted R square of 0.65 (refer to Exhibit 133) indicates that 65% of the variation in the yearover-year percentage change in FedEx Ground’s ADFlb, is explained by the year-over-year percentage change in U.S. Personal Consumption. o The resulting regression model equation based on the coefficient in Exhibit 133 is: o

•

The predicted ADFlb range based on the model standard error is +/- 6.3%.

ADFlb year-over-year % change = 1.724 x U.S. consumption year-over-year % change The predicted ADFlb range based on the model standard error is +/- 4.8%.

Regressions 5—ADFlb/GDP & Consumption o The adjusted R square of 0.65 (refer to Exhibit 134) indicates that 65% of the variation in the yearover-year percentage change in FedEx Ground’s ADFlb, is explained by the year-over-year percentage change in U.S. GDP and U.S. Personal Consumption. o The resulting regression model equation based on the coefficient in Exhibit 134 is: ADFlb year-over-year percentage change = 0.365 x U.S. consumption year-over-year percentage change 1.388 x U.S. GDP year-over-year percentage change o The predicted ADFlb range based on the model standard error is +/- 4.7%.

Other Considerations: Textbook examples of regression analysis tend to be relatively straightforward. In the real world, data and external conditions will change over time, which makes a regression model less powerful, and in most cases ineffective. Even if your model passes all of the statistical tests, it is important to think qualitatively about whether or not there is a true economic reason for the variables’ correlation, and whether of not the direction of the coefficients makes sense. In addition, it is important to assess whether or not the population has changed over time. If there have been significant changes, then the model could not be used (violates the fundamental conditions for a regression model). In the case of FedEx the segment reporting has changed over time, therefore regression analysis is probably not an appropriate 182

Appendix 1: Using Regression Analysis to Predict Earnings method for forecasting. Nevertheless, for the sake of our example the model which shows the most promise is Regression 2—ADFlb/GDP. We can select this regression model over the others based on the following points: •

The t-stat is above the critical-t value at the 99% confidence interval, and the P-value is close to zero.

•

There is no significant sign of heteroskedasticity based on the Breusche-Pagan test, although the plot of the residuals indicates there could be some level of correlation.

•

No significant signs of serial correlation as measured by the Durbin-Watson statistic.

•

The distribution appears to be relatively normal as measured by the skew and kurtosis of the residuals.

•

The model is relatively stable after removing two random observations from the sample and re-running.

•

The adjusted R square of 66% is higher than the other models (although still relatively low for forecasting).

•

Directionally the coefficient is logical, as the positive correlation indicates a direct relationship, meaning as GDP increases, the FedEx Ground ADFlb predicted by the regression model will also increase. This makes sense since increases in GDP should result in increases in shipping volumes.

Now that we have selected our regression model, we can use the regression equation to predict the future percentage change in the FedEx Ground ADFlb by inputting the percentage change in U.S. GDP, which is published before the FedEx quarterly earnings release. For example, the calendar quarter 2Q2018 (June 30,2018) published U.S. GDP was 20,412M which represents a yearover-year percentage change of 5.4%. Plug the GDP growth rate into the regression equation, and solve for the percentage change in ADFlb: ADFlb year-over-year percentage change = 1.747 x 5.4% = 9.4% Next plug in the projected ADFlb into the earnings model as shown in Exhibit 143. This input has become part of our earnings forecast ahead of the earnings release. Now, fast forward to the F1Q2019 FedEx results from Chapter 8, to check how our regression-based projection performed against the reported actual. In this case, the regression model was much less accurate compared to our simplified approach of projecting the Ground Segment ADFlb based on the average of the last four quarters and adjusting up by 1% for an improvement in U.S. economic conditions that we originally used in Chapter 4.

Exhibit 143: Using the Regression Result in the Earning Model

What Went Wrong With Our Regression? The short answer is nothing. The projected result of 9.4% for the fiscal first quarter of 2019 is within the projected result range implied by the model’s standard error, but the model will only project 66% of the variation in ADFlb, leaving the remaining 34% to chance. Perhaps changes in reporting structure overtime have weakened the forecasting power of the regression model. Or the inherent lag in the data due to the fiscal reporting structure of the FedEx results, have rendered the model ineffective. Even if there were a higher correlation between the variables and stable segment reporting over time, we would still need to adjust the output to reflect changes in average economic conditions in the data used to train our regression model. Or to put it in other terms, the time period of the 32 observations used to create the regression model did not have the same economic conditions as the latest quarter. This example shows why most research analysts do not use regression models. They are complex, time consuming, and do not necessarily result in a more accurate forecast.

183

Step R5: Perform Back-Testing

If you are able to find a regression model with a high correlation to a metric in your earnings model, the next step would be to back-test the regression against the historic results. You could manually plug each quarterly GDP observation into the regression equation to back-test how accurate the model would have been for each quarter, or simply review the residuals in the regression output.

Table 12—FedEx ADFlb Regression Model Back-Testing Quarter

Model Predicted ADFlb % Change

Actual Reported ADFlb % Change

Quarter

Model Predicted ADFlb % Change

Actual Reported ADFlb % Change

F1Q2011

7.0%

7.3%

F1Q2015

8.3%

0.8%

F2Q2011

8.0%

9.5%

F2Q2015

9.0%

1.8%

F3Q2011

7.3%

8.8%

F3Q2015

7.7%

2.0%

F4Q2011

6.7%

10.6%

F4Q2015

8.9%

3.5%

F1Q2012

6.7%

10.9%

F1Q2016

8.0%

4.0%

F2Q2012

5.9%

7.3%

F2Q2016

6.0%

9.4%

F3Q2012

6.4%

7.4%

F3Q2016

5.1%

11.2%

F4Q2012

8.4%

7.0%

F4Q2016

4.3%

10.5%

F1Q2013

7.4%

8.3%

F1Q2017

4.0%

10.0%

F2Q2013

7.5%

10.6%

F2Q2017

4.5%

5.0%

F3Q2013

6.2%

15.3%

F3Q2017

5.9%

2.2%

F4Q2013

6.0%

14.5%

F4Q2017

7.2%

3.3%

F1Q2014

5.3%

15.2%

F1Q2018

6.7%

4.0%

F2Q2014

6.4%

8.3%

F2Q2018

7.3%

7.1%

F3Q2014

7.7%

5.7%

F3Q2018

7.8%

5.5%

F4Q2014

5.6%

3.1%

F4Q2018

8.0%

5.5%

Step R6: Model Limitations

As demonstrated with the FedEx model, there are limitations to regression analysis which should be considered when deciding whether or not to use this method to forecast earnings inputs. These include the following items: Sample: Regression analysis draws conclusions based on a limited sample of observations, typically at least 30 observations if possible. As a result, there is always a degree of error expected. Ahead of an earnings release, we quantify the expected error with a range of values. Actual results can exceed the model’s standard error, particularly if the assumptions in the model are not held constant in the actual results. Historic Trends: Regression analysis assumes that historic relationships will hold constant in the future, and does not consider changes such as market share, currency rate fluctuations, or general increases in the Total Addressable Market (TAM) over time. The residuals from the predicted values show instances in the past where the model predicted value differs significantly from the actual result. Similar instances will likely occur in the future. Revisions: Economic data is often revised after it is released, which can result in impaired predictability. Company Policies: Accounting policies can change over time and may not align with historic results. Due to the inherent limitations, most analysts do not utilize regressions in their forecasts. Despite the limitations, regression models can provide important insight, and if nothing else, add a few extra data points to consider in our earnings sensitivity review ahead of a quarterly release.

184

APPENDIX 1 WRAP-UP Takeaways •

Regression analysis is a mathematical process which estimates the relationship between two or more variables.

•

The primary steps of regression analysis are: o Select the variables to test o Collect and prepare the data o Run the regression o Test the regression model o Understand the Limitations of the results

•

Regression analysis can be a useful forecasting tool under the right circumstances; however, due to the limitations of the approach, most equity research analyst do not employ the technique in their earnings projections. These limitations include: o Limited relevant observations available to create a model. o Historic trends tend to change, which violate a primary assumption required for regression. o Data which could be used to create regression models is often revised after the initial release. o Company specific segment reporting or accounting policies change over time, which can cause issues when relying on regression models which were built using historic reporting methods.

Concept Quiz Instructions: Use the following information to answer questions 1 through 6 The Tough Dirt Mining Corp (TDMC) mines precious metals in Africa. The company does not hold any metal inventory, and uses forward contracts to hedge exposure to fluctuating commodity prices. Jack Thomas covers TDMC and would like to see if there is a correlation between the price of gold and TDMC’s earnings. He gathers data through mid-2008 and regresses the quarter-over-quarter percentage change in TDMC’s revenue against the percentage change in gold prices and analyzes the following regression output:

185

1) Jack looks up the Critical t-value on the t-distribution table with 39 degrees of freedom, and a probability of 0.005 in each tail. The resulting Critical t-value is 2.708. Based solely on this critical t-value and the t-stat in cell N17, what conclusion would Jack reach regarding the regression model? A. The Critical t-value of 2.708 indicates that Jack should reject the notion that no statistically significant relationship exists between TDMC’s revenue and gold prices. B. The t-stat of 15.3 is greater than the Critical t-value of 2.708. Jack should reject the notion that no statistically significant relationship exists between TDMC’s revenue and gold prices. C. The t-stat of 15.3 is greater than the Critical t-value of 2.708. Jack should accept the notion that no statistically significant relationship exists between TDMC’s revenue and gold prices. 2) How should Jack interpret the P-value in cell 017? A. A P-value represents the lowest level of significance the regression model can reach while still showing a statistically significant relationship. In this case the P-Value of zero means there is likely no relationship between gold prices and TDMC’s revenue. B. A P-value represents the lowest level of significance the regression model can reach while still showing a statistically significant relationship does not exist. In this case the P-Value of zero means there is likely no relationship between gold prices and TDMC’s revenue. C. A P-value represents the lowest level of significance the regression model can reach while still showing a statistically significant relationship. In this case the P-Value of zero means there is likely a statistically significant relationship between gold prices and TDMC’s revenue. 3) How should Jack interpret the adjusted R square in cell L5? A. R square represents the amount of variation in the dependent variable, explained by the independent variable. Therefore 86% of the quarter-over-quarter percentage change in TDMC’s revenue is explained by the percentage change gold prices. B. R square represents the amount of variation in the dependent variable, explained by the independent variable. Therefore 86% of the quarter-over-quarter percentage change in gold prices is explained by the percentage change TDMC’s revenue. C. R square represents the amount of variation in the dependent variable, explained by the independent variable; however, since the t-stat and P-value do not support a correlation between the variables, we cannot rely on the R square in cell L5 4) What is the regression model equation? A. TDMC revenue (% change) = 0.6988 × Gold price (% change) B. TDMC revenue (% change) = 0.8568 × Gold price (% change) C. TDMC revenue (% change) = 0.0091 + 0.6988 × Gold price (% change) 5) Last quarter TDMC booked $29,293 million in revenue. Jack would like to predict TDMC’s revenue for next quarter using the regression model he created. Gold prices for the quarter have increase 3.4%. What is the revenue predicted by the regression model? A. $961 million B. $30,254 million C. $32,815 million 6) Which of the following are limitations of Jack’s regression model? i. The economic conditions which caused significant changes in the price of gold within the data set used to train Jack’s model, may differ from current economic conditions. ii. The mix of precious metals (gold, silver, platinum, etc.) which TDMC mines may fluctuate over time, and may be significantly different from the mix in the data used to create the regression model. iii. Changes in the terms of hedge contracts overtime may impact the predictive power of the regression model.

186

Appendix 1 Wrap-Up A. i, ii, and iii. B. ii and iii. C. ii only.

Concept Quiz Answers 1) B. The t-stat of 15.3 is greater than the Critical t-value of 2.708. Jack should reject the notion that no statistically significant relationship exists between TDMC’s revenue and gold prices. 2) C. A P-value represents the lowest level of significance the regression model can reach while still showing a statistically significant relationship. In this case the P-Value of zero means there is likely a statistically significant relationship between gold prices and TDMC’s revenue. 3) A. R square represents the amount of variation in the dependent variable, explained by the independent variable. Therefore 86% of the quarter-over-quarter percentage change in TDMC’s revenue is explained by the percentage change gold prices. 4) C. In the examples from the chapter we ran the regressions without a slope intercept. In this case we included the intercept therefore the coefficient must also be included in the regression model equation. 5) B. The regression model predicts the percentage change in TDMC revenue quarter-over-quarter. TDMC revenue (% change) = 0.0091 + 0.6988 × Gold price (% change), so TDMC revenue (% change) = 0.0091 + 0.6988 × 3.4% = 3.28%. $29,293 million × (1 + 3.28%) = $30,254 million. 6) A. All three items listed represent limitations of the regression model.

187

APPENDIX 2: FEDEX FISCAL 2Q2019 EARNINGS RELEASE Step 1: Prepare for the Release

Step 2: Update After the Release

Appendix 2 Overview: Each time a new quarterly earnings report is released, or when significant company or market developments occur, the FedEx earnings model will need to be updated. Going forward these updates will be tracked in the Appendix section of this textbook, and published in future editions. This represents the earnings preview/review cycle, which demonstrates that a model is a living file, and must be maintained continuously.

Appendix 2: Step 1—Prepare for the F2Q2019 Release (Earnings Preview) Note: This section was updated on December 14, 2018 before the earnings release on December 18, 2018. The following is a list of steps to consider, and questions to think about, as part of the quarterly earnings preparation: •

Review the relevant leading indicators, and latest news from the investor relations page as a final check to determine if there are additional items to consider in the forecast, ahead of the earnings release.

•

Update the valuation metrics to reflect the latest data including: the current share price, last three-month forward Price-to-Earnings multiple, stage-one beta, volatility, interest rates, mean monthly return and standard deviation.

•

Consider the consensus estimate. Are there any points in the consensus earnings forecast which we may be missing in our model? Look to the high-end, and low-end of the consensus range. How has the consensus revenue, EPS estimate, and share price target changed since the last quarter? How does the change in the consensus estimate compare to the change in macroeconomic conditions over the course of the quarter?

•

How does the share value look relative to recent history? Is the Price-to-Earnings ratio above or below the historic average?

•

Save a copy of the existing model in an archive folder.

•

Update the model headers.

•

Change the forecast equations to values, and insert ratio equations.

•

Copy/paste special values on the results comparison table (refer to Exhibit 121).

Review last quarter’s press release, earnings transcript, and presentation slides as a refresher on the important topics to watch for in the next press release. Review the Developments Over the Quarter: FedEx does not have specific channel-checks or supply chain earnings results to analyze for an earnings preview. Instead we can look to the following four datapoints as leading indicators for FedEx: 1) macroeconomic data through the quarter, 2) change in the U.S. Dollar (the reporting currency) against other currencies, 3) changes in fuel prices, and 4) any press releases from the investor relations home page. •

Review the Developments Over the Quarter (GDP Growth): When we look at macroeconomic indicators, we can review changes in the trends, from the last time management gave guidance. Given the delay in reporting economic data, this datapoint is more likely to offer insight into whether or not management will change their future guidance, and less likely to offer insight into the current quarter. Global Gross Domestic Product (GDP) is typically used as a barometer for general economic activity, and can be used as a proxy for the amount of goods which will be shipped 188

Appendix 2: FedEx Fiscal 2Q2019 Earnings Release globally. Exhibit 144 below shows that through September of 2018, U.S. growth is continuing, however growth in Europe and China is slowing. As a result we should bring down our International business segment growth rates.

Exhibit 144—Global GDP (percentage change, year-over-year)

Source: U.S. Federal Reserve, https://fred.stlouisfed.org

Review the Developments Over the Quarter (U.S. Dollar): As shown in Exhibit 145 below, the quarterly average U.S. Dollar exchange rate has remained relatively constant. As a result no adjustment is necessary for this metric at this time.

Exhibit 145—U.S. Dollar Exchange Rates (quarterly average)

Source: U.S. Federal Reserve, https://fred.stlouisfed.org

Review the Developments Over the Quarter (Fuel Prices): Note that changes in fuel prices will have two offsetting components: the change in fuel expense, versus the change in fuel surcharge. The company publishes fuel surcharge changes by week. You can monitor the detail, or assume a basic increase or decrease in the net impact. Jet fuel represents the largest component of overall fuel expense. The price of jet fuel has declined toward the end of the quarter (refer to Exhibit 146 below); however, the price remains within the range seen in the last few quarters. Overall a decline in fuel price may offset some of the economic headwind from Europe and China.

189

Exhibit 146—Fuel Prices

Source: U.S. Federal Reserve, https://fred.stlouisfed.org

Update the Valuation Metrics—Beta: A re-run of the last 12 monthly returns for FedEx shares against the S&P500 Index shows the stage-one short-term beta coefficient has increased significantly from 1.065 (as of September 14, 2018) to 1.592 (as of December 14, 2018). The stage-two long-term beta regression run against monthly returns through January of 2008, has also increased from 1.268 (as of September 14, 2018) to 1.297 (as of December 14, 2018). Both beta coefficients have been updated in the earnings model. For details on how to calculate beta refer to Chapter 5, Step 20. Follow Along in the Spreadsheet: Refer to “File 21-FDX Beta Calculation (Appendix 2-Step 1 F2Q2019 Preview)” for the updated beta calculation prior to the F2Q2019 earnings release. Refer to the "How to Use This Textbook" section of Chapter 1 for instructions on how to access the spreadsheet files. Update the Valuation Metrics—All Other Items: The following changes have been incorporated into the model to reflect the latest data available as of December 14, 2018. The comparisons represent changes from the model after the F1Q2019 release reflected in Chapter 8, Step 35: •

The FedEx share price has decreased to $184 from $240. This has resulted in a decrease in the equity-to-total capital ratio of 74.9% from 79.8%.

•

Market volatility (VIX, quarterly average) has increased to 19.96% from 19.00%. The 12-month trailing average increased to 16.38% from 16.14%.

•

The quarterly average 10-year U.S. Treasury rate has decreased to 3.06% from 3.23%.

•

The equity market return assumption for the current quarter has decreased to -8.67% from the previous -8.00% (projected).

•

The three-month average, Next Twelve Month forward Price-to-Earnings ratio has decreased to 13x. Shares are trading at the three-month low heading into the release, down from the prior average of 15.8x.

•

The 12-month mean monthly return in FedEx shares declined to -1.63% from 0.19%, and the standard deviation in returns has increased to 8.97% from 5.57%.

After incorporating each of these items into the model including the change in beta, the target share valuation has decreased to $196 from $226, and the new price band is $158 to $228, driven by the higher standard deviation in average monthly returns. Follow Along in the Spreadsheet: Refer to “File 22-FDX Model (Appendix 2-Step 1 F2Q2019 Preview)” for the updated FDX model prior to the F2Q2019 earnings release. Refer to the "How to Use This Textbook" section of Chapter 1 for instructions on how to access the spreadsheet files.

190

Appendix 2: FedEx Fiscal 2Q2019 Earnings Release Pitfall: Keep in mind that we are updating the FedEx model the week before the earnings call. As noted above there have been significant market changes over the course of the quarter. If you were covering FedEx (or any other company for that matter) you should not wait until the week before the release to update the market risk and valuation factors, if there is a sharp change in market conditions as in this case. Update Model Tables: It is important to maintain the tables within the model. I have included key consensus estimates and price targets which will need to be updated, as well as graphs based on the details in the earnings models. The charts show the latest four quarters of reported results, and four quarters of projections. The cell references must be updated for the next quarterly release. Sensitivity Data Tables: In Chapter 8, Step 32 we ran EPS sensitivity analysis on the FedEx Express Priority ADV and yield estimates. Each quarter the references in the Excel data table used for the sensitivity analysis must be updated to link to the next quarter’s EPS, ADV and yield.

Appendix 2: Step 2—Update After the F2Q2019 Release (Earnings Review) Note: This section was updated on December 18, 2018, after the earnings release. Key Takeaways from the F2Q2019 Conference Call: The primary headline is that the international segments are slowing much faster than management had anticipated. If you were vigilant in tracking the economic developments, particularly in the last week leading up to the release, you may have seen that China’s economy has continued to show signs of weakness, which may cause a drag on overall global growth in the near-term. In addition European economic conditions began to weaken. We incorporated some of the impact of these items in Appendix 2, Step 1; however, we did not decrease the forecast enough, and did not incorporate a downward revision to management’s guidance. Here are some of the important economic comments from the call: As our volumes and revenues demonstrate, FedEx is experiencing strong growth in the U.S. where the economy remains solid; however, our international business, especially in Europe, weakened significantly since we last talked with you during our earnings call in September. In addition, China’s economy has weakened due in part to trade disputes. As a result, we have lowered our fiscal 2019 earnings guidance and are accelerating actions to reduce costs given the uncertainty of global macroeconomic trends. We’re highly confident that we will achieve the benefits expected with the acquisition of TNT Express, although we will not achieve our FedEx profit improvement goal in fiscal 2019. -Frederick W. Smith, FedEx Corp, Chairman, President & CEO, Fiscal second quarter 2019 earnings conference call, December 18, 2018. To offset some of the macroeconomic impact management has initiated cost cutting measures which includes an employee buyout program expected to cost between $450M to $575M, and save between $225M and $275M per year beginning in fiscal 2020. Management revised the EPS guidance downward, and disclosed that they would no longer be providing guidance for revenue or operating margin in fiscal 2019. Regarding our fiscal year 2019 outlook, we are projecting adjusted earnings of $15.50 to $16.60 per diluted share, down from $17.20 to $17.80. This lower guidance is due to a shift in business conditions and service mix at Express, primarily in Europe. We are no longer providing guidance for revenue growth and operating margin for fiscal year 2019. - Alan B. Graf, Jr., FedEx Corp, Executive Vice President & CFO, Fiscal second quarter 2019 earnings conference call, December 18, 2018. Model Updates: Prior to the release we were modeling the international subsegment ADV and yield for the next four quarters, based on the prior four quarter average minus -4.0%, -3.5%, -3.0%, and -1.0% for F3Q2019, F4Q2019, F1Q2020, and F1Q2020 respectively. Based on the international weakness and lower guidance, I have updated the model to reflect declines of -5.0%, -4.0%, -3.5%, and -1.5%. Management stated that they may not be purchasing additional shares in 2019, and the 2019 share count guidance implied by the TNT integration expense forecast per share confirms this expectation, so I have removed the repurchase assumptions for 2019. We are reviewing all aspects of our financial plans, including whether we will repurchase additional shares this year. As a reminder, we spent $11.6B purchasing almost 76M shares over the last 5.5 years, resulting

191

in the nearly 18% reduction in outstanding shares. - Alan B. Graf, Jr., FedEx Corp, Executive Vice President & CFO, Fiscal second quarter 2019 earnings conference call, December 18, 2018. In addition to the revised forecast for ADV and yield, I also incorporated the employee buyout charges, and increased operating expenses until the forecast fell within management’s new guided EPS range. Follow Along in the Spreadsheet: Refer to “File 23-FDX Model (Appendix 2-Step 2 F2Q2019 Review)” for the updated FedEx model after the F2Q2019 earnings release. Refer to the "How to Use This Textbook" section of Chapter 1 for instructions on how to access the spreadsheet files.

Appendix 2: Step 3—Recalibrate to the Post-Earnings Consensus Note: This section was updated on January 2, 2019 after the Equity Risk Premium close for the fourth calendar quarter. Recalibration Logic: At Gutenberg Research we provide consensus-based earnings models to act as a “base-case” scenario which our community members can input their own assumptions into. About a week after the earnings release we revise the model to include the latest consensus estimates and any additional updates. To recalibrate the model to the latest consensus estimates, I have adjusted the international sub-segment growth rates, and the operating expense assumptions. Follow Along in the Spreadsheet: Refer to “File 25-FDX Model (Appendix 2-Step 3 F2Q2019 Recalibration)” for the updated FedEx model after the F2Q2019 earnings release, including the following updates: 1) Details from the 10-Q filing for F2Q2019, 2) the updated Equity Risk Premium data for the end of the fourth calendar quarter, 3) model recalibration adjustments to meet the post-earnings revised consensus estimates. Refer to the "How to Use This Textbook" section of Chapter 1 for instructions on how to access the spreadsheet files. Update the Beta Coefficients: A re-run of the last 12 monthly returns for FedEx shares against the S&P500 Index, including the large decline after the fiscal second quarter results, shows the stage-one short-term beta coefficient has increased significantly from 1.592 (as of December 14, 2018) to 1.79 (as of January 2, 2019). The stage-two long-term beta regression run against monthly returns through January of 2008, has also increased from 1.297 (as of December 14, 2018) to 1.33 (as of January 2, 2019).Overall the increase in beta reflects the market is pricing in higher risk which will increase the required return on equity and decrease the target share valuation. Follow Along in the Spreadsheet: Refer to “File 26-FDX Beta Calculation (Appendix 2-Step 3 F2Q2019 Recalibration)” for the updated beta calculation after the F2Q2019 earnings release. Refer to the "How to Use This Textbook" section of Chapter 1 for instructions on how to access the spreadsheet files. Update the Valuation Metrics: The ERP Model updates through December 31, 2018 are included in Appendix 3, Step 2. After adding the 12-month projected VIX, Federal Funds rate, and the 10-year U.S. Treasury rate estimate to our FedEx model, the 50/50 weighted FedEx target share price is $165 with a price target band of $129 to $190.

192

APPENDIX 3: EQUITY RISK PREMIUM MODEL UPDATE Appendix 3 Overview: Each time the Federal Reserve’s FOMC meets, the ERP model must be updated to reflect the latest details related to the Federal Funds rate. Going forward these updates will be tracked in the Appendix section of this textbook, and published in future editions. Appendix 3 includes the details from the December 19, 2018 FOMC meeting.

Appendix 3: Step 1—FOMC Meeting Update

From the Fed’s point of view, not much has changed from the last Fed meeting in November. The Fed still believes economic activity has been rising at a strong pace. At the December meeting, the Committee voted unanimously to raise the target Fed Funds rate another 25 basis points to a range of 2.25% to 2.50%, from the prior range of 2.00% to 2.25%. In terms of the 2019 target rate, the Fed decreased the number of expected rate hikes from three 25 basis point increases to just two. The market dropped significantly following the press conference with the Fed. This was likely due in part to some market participants looking for a pause in the hikes, as the global economy is showing signs of a slowdown, or a more dovish hike, with one rate increase for 2019 not two. In addition, Chairman Powell stated that he does not expect to change the Balance Sheet runoff plans. The Fed’s Balance Sheet is now at ~$1.1 trillion, with about $50 billion in runoff per month. The Fed Statement and Projection Material are included below. Note that the December meeting Projection Material is in Exhibit 147, and the September meeting in Exhibit 148 for comparison. The ERP model has been updated to include the latest Fed projected rate increases, with two 25 basis point increases in 2019, which have been added to the second and fourth quarters. This assumes the Fed takes a pause for the January 30, 2019 and March 20, 2019 meetings. I have also included one 25 basis point increase in the second quarter of 2020. The ERP model was also updated to include the latest available equity returns, volatility, and spread between the 10year U.S. Treasury security and the Federal Funds rate. I have increased the spread forecast from ~0.6% to ~1.0% by the end of 2020, which includes some effect from the Balance Sheet normalization program. Impact on the ERP Estimate for 4Q2018: The ERP estimate now stands at 5.2% versus the previous estimate of 5.4%. The slight decrease from the last update reflects the impact of lower interest rates (Federal Funds rate and spread) offset by an increase in volatility and lower equity market return projections. The ERP remains below the historic average of 6.2%. Follow Along in the Spreadsheet: Refer to “File 24 Equity Risk Premium Model (Appendix 3 Dec-2018 FOMC Meeting Update)” for the updated ERP Model. Refer to the "How to Use This Textbook" section of Chapter 1 for instructions on how to access the spreadsheet files.

Latest FOMC Statement from December 19, 2018: Raised target range to between 2.25% to 2.50%

Information received since the Federal Open Market Committee met in November indicates that the labor market has continued to strengthen and that economic activity has been rising at a strong rate. Job gains have been strong, on average, in recent months, and the unemployment rate has remained low. Household spending has continued to grow strongly, while growth of business fixed investment has moderated from its rapid pace earlier in the year. On a 12-month basis, both overall inflation and inflation for items other than food and energy remain near 2 percent. Indicators of longer-term inflation expectations are little changed, on balance.

193

Consistent with its statutory mandate, the Committee seeks to foster maximum employment and price stability. The Committee judges that some further gradual increases in the target range for the Federal Funds rate will be consistent with sustained expansion of economic activity, strong labor market conditions, and inflation near the Committee's symmetric 2 percent objective over the medium term. The Committee judges that risks to the economic outlook are roughly balanced, but will continue to monitor global economic and financial developments and assess their implications for the economic outlook. In view of realized and expected labor market conditions and inflation, the Committee decided to raise the target range for the Federal Funds rate to 2-1/4 to 2‑1/2 percent. In determining the timing and size of future adjustments to the target range for the Federal Funds rate, the Committee will assess realized and expected economic conditions relative to its maximum employment objective and its symmetric 2 percent inflation objective. This assessment will take into account a wide range of information, including measures of labor market conditions, indicators of inflation pressures and inflation expectations, and readings on financial and international developments. Voting for the FOMC monetary policy action were: Jerome H. Powell, Chairman; John C. Williams, Vice Chairman; Thomas I. Barkin; Raphael W. Bostic; Michelle W. Bowman; Lael Brainard; Richard H. Clarida; Mary C. Daly; Loretta J. Mester; and Randal K. Quarles. Source: FederalReserve.gov, FOMC Statement, December 19, 2018, retrieved December 20, 2018.

Exhibit 147—Latest Projection Materials Example (December-2018 Meeting)

Source: FederalReserve.gov, FOMC Projections Materials, December 19, 2018, retrieved December 20, 2018.

194

Appendix 3: Equity Risk Premium Model Update

Exhibit 148—Previous Projection Materials Example (September-2018 Meeting)

Source: FederalReserve.gov, FOMC Projections Materials, September 26, 2018, retrieved September 28, 2018.

Appendix 3: Step 2—Data Maintenance

Equity Risk Premium Updates: Each quarter the new interest rate, market return, and volatility results must be input into the Equity Risk Premium (ERP) model. After the close of the fourth calendar quarter on December 31, 2018, our new estimate of the 12-month forward average VIX (on a trailing basis) is 16.50% (based on the 4Q2019 estimate). The 4Q2018 VIX was 21.05, reflecting the higher volatility in the fourth quarter, and the 12-month trailing average VIX at the end of 4Q2018 was 16.65%. Our average 10-year U.S. Treasury rate forecast is 3.74%. This is based on two 25 basis point Federal Funds rate hikes in 2019, and an increase in the spread to just over 1%. Feel free to enter your own assumptions for the ERP estimate. Refer to Chapter 5, Step 19 for details on the ERP Model. We have some flexibility in the ERP and DCF valuation. My approach is to project what I believe the NTM ERP will be, and use these projected metrics in my 12-month forward price target. This assumes that the market is including projections of these measures in the share valuation. Remember that I am using the Federal Funds target from the FOMC material, but the market may be projecting a different interest rate forecast ahead of future FOMC meetings.

195

Exhibit 149—Latest Equity Risk Premium Updates

Follow Along in the Spreadsheet: Refer to “File 27 Equity Risk Premium Model (Appendix 3-December 2018 Update)” for the latest ERP Model, through December 31, 2018. Refer to the "How to Use This Textbook" section of Chapter 1 for instructions on how to access the spreadsheet files.

196

APPENDIX 4: FEDEX FISCAL 3Q2019 EARNINGS RELEASE Step 1: Prepare for the Release

Step 2: Update After the Release

Appendix 4 Overview: Each time a new quarterly earnings report is released, or when significant company or market developments occur, the FedEx earnings model will need to be updated. Going forward these updates will be tracked in the Appendix section of this textbook, and published in future editions. This represents the earnings preview/review cycle, which demonstrates that a model is a living file, and must be maintained continuously.

Appendix 4: Step 1—Prepare for the F3Q2019 Release (Earnings Preview) Note: This section was updated on March 10, 2019 before the F3Q2019 earnings release on March 19, 2019. The following is a list of steps to consider, and questions to think about, as part of the quarterly earnings preparation: •

Review the relevant leading indicators, and latest news from the investor relations page as a final check to determine if there are additional items to consider in the forecast, ahead of the earnings release.

•

Update the valuation metrics to reflect the latest data including: the current share price, last three-month forward Price-to-Earnings multiple, stage-one beta, volatility, interest rates, mean monthly return and standard deviation.

•

Consider the consensus estimate. Are there any points in the consensus earnings forecast which we may be missing in our model? Look to the high-end, and low-end of the consensus range. How has the consensus revenue, EPS estimate, and share price target changed since the last quarter? How does the change in the consensus estimate compare to the change in macroeconomic conditions over the course of the quarter?

•

How does the share value look relative to recent history? Is the Price-to-Earnings ratio above or below the historic average?

•

Save a copy of the existing model in an archive folder.

•

Update the model headers.

•

Change the forecast equations to values, and insert ratio equations.

•

Copy and paste values on the results comparison table (refer to Exhibit 121).

Review last quarter’s press release, earnings transcript, and presentation slides as a refresher on the important topics to watch for in the next press release. Review the Developments & Leading Indicators Over the Quarter: U.S. economic activity is holding up relatively well, although sentiment is turning negative as the trade war heats up. International markets remain relatively sluggish. To reflect this we can bring down our Average Daily Volume and Yield estimates in the International Segments. Competition with Amazon Air seems to be increasing as well, although Amazon makes up a relatively small portion of FedEx revenue, and it is likely too early for the service to have a broader impact on shipping costs across the industry. One positive is that fuel prices have declined, which may not be fully offset in fuel surge charge changes. To reflect this the fuel expense in the Express segment has been lowered in the next quarter. •

197

Update the Valuation Metrics—All Other Items: The following changes have been incorporated into the model to reflect the latest data available as of March 11, 2019. The comparisons represent changes from the model after the F2Q2019 release reflected in Appendix 2 “Step 3—Recalibrate to the Post-Earnings Consensus” on January 2, 2019: •

The FedEx share price has decreased to $176 from $184 in the previous model update. This has resulted in a decrease in the equity-to-total capital ratio to 73% from 75%.

•

Market volatility (VIX, 1-year trailing average) has increased to 18.67% from 16.38%. Refer to “File 31 Equity Risk Premium Model (Appendix 5-February 2019 Update)” and Appendix 5 for additional details.

•

The long-term Beta coefficient (stage-two), using a regression of the percentage change in FedEx shares versus the S&P500, moved from 1.33 to 1.32, and the short-term Beta (stage-one) moved from 1.79 to 1.59. Refer to “File 28-FDX Beta Calculation (Appendix 4-Step 1 F3Q2019 Preview).”

•

The quarterly average 10-year U.S. Treasury rate has increased to 3.36% from 3.06%. Refer to “File 31 Equity Risk Premium Model (Appendix 5-February 2019 Update)” and Appendix 5 for additional details.

•

The three-month average, Next Twelve Month forward Price-to-Earnings ratio has increased to 16x. Shares are trading near the three-month average heading into the release.

•

The 12-month absolute value of the mean monthly return in FedEx shares stands at 2.7%, and the standard deviation in returns has decreased to 6.6% from 8.9%.

After incorporating each of these items into the model, including the change in beta, the target share valuation has decreased to $173 from $196, and the new target price band is $155 to $200, which is tighter than the previous band due to the lower standard deviation in average monthly returns. Follow Along in the Spreadsheet: Refer to “File 29-FDX Model (Appendix 4-Step 1 F3Q2019 Preview)” for the updated FDX model prior to the F3Q2019 earnings release. Refer to the "How to Use This Textbook" section of Chapter 1 for instructions on how to access the spreadsheet files.

Appendix 4: Step 2—Update After the F3Q2019 Release (Earnings Review) Note: This section was updated on March 19, 2019, after the earnings release. Key Takeaways from the F3Q2019 Conference Call: There were five primary takeaways raised on the conference call: 1) the macroeconomic environment continues to be a headwind for earnings, 2) FedEx management is not overly concerned with competition from Amazon, 3) there is still substantial integration work to be done with TNT, 4) the U.S. Business was impacted by mix, and 5) earnings guidance was lowered again. Takeaway 1—The macroeconomic environment continues to be a headwind for earnings: We see solid economic growth in the U.S., but somewhat below last year's pace. Internationally, performance is mixed across regions as overall growth moderates. The Eurozone and Japan still appears sluggish, while emerging markets growth eases at a gradual pace. A recurring theme in global surveys on economic activity is the negative impact from global trade frictions and heightened uncertainty. World trade is slowing and leading indicators point to positive but ongoing deceleration and trade growth in the near term. Since our last earnings call, we have seen the overall China economy slow down further and this has impacted other Asian economies. Given the size of China, no markets will be able to absorb more than a fraction of what China produces, but customers continue to look to diversify from China. We have also seen some customers evaluate mode optimization. Our network and portfolio lets customers respond quickly and act locally for our customers in China as well as around the world. Brie Carere Chief Marketing and Communications Officer & Executive Vice President, FedEx Corp. Fiscal third quarter 2019 earnings conference call, March 19, 2019. Resulting Model Update: Decrease International yield and volume forecasts. Takeaway 2—FedEx management is not overly concerned with competition from Amazon: Management dedicated a substantial amount of time to explaining that the press may be overestimating the impact from Amazon Air.

198

Appendix 4: FedEx Fiscal 3Q2019 Earnings Release …there continues to be significant media and investor interest on the potential for Amazon to disrupt the transportation industry. We have been clear that this is not a threat to our business because Amazon represents less than 1.3% of our total revenue, which is substantially lower than what our largest competitor carries. Nor is Amazon a threat to our future growth. I want to take this opportunity to provide some additional facts about the market we play in. The size of the U.S. domestic parcel market is roughly 15 million packages per day. In addition, the global market for international parcel shipments is another 10 million packages per day, and we carry around 14.5 million packages per day today. Even if the e-commerce market did not grow one additional package, there's still substantial growth in the industrial sector of the market where we're extremely well-positioned to gain share due to our unmatched network and global portfolio. So now, let's talk about e-commerce. This represents a significant additional opportunity for growth. We believe that we're able to add to our existing expertise and provide a market-leading value proposition here as well. We continue to invest and enhance our capabilities. We've already seen several examples such as Extra Hours, FedEx OnSite, FedEx Delivery Manager, and our enhanced return solutions. We are well-positioned to provide the best service at the appropriate price point, leveraging our current capabilities and targeted additional investments. Now, there's an intense media focus on the "last mile", but very few people think about the first few thousand miles. When you see a FedEx truck on the road, it not only is carrying those local last mile shipments, but also the other shipments that are originated from all parts of the globe, creating density for last mile delivery and higher revenue per stop. This is an inherent advantage for players who have a global network in place. Any new entrant in this space will not have this benefit; and also not have any particular advantage on the input costs. Our core business can be segmented in two basic categories; large customers and small and medium enterprises. Large customers of complex global supply chains with specialized needs. Our international network, expertise and broad portfolio enable us to serve these customers with very sticky, customized solutions. Our value proposition is far more sophisticated than just local last mile and we have bundled pricing across the portfolio. Small and medium businesses seek simplicity and generally desire one-stop shop and they want a transportation provider who can handle their local, national, rural and peak needs. We also have a robust loyalty and earn discount pricing strategy for this segment that again rewards customers for bundling their business with FedEx. So in short, we have a terrific portfolio that'll allow us to grow our core business segment and we are very well placed to take share in the rapidly-growing e-commerce segments. In the U.S. market alone, we expect the parcel market to double in size to more than 100 million packages per day by 2026. When you view the unprecedented growth opportunity in our industry in the years ahead and the very small number of providers that'll be able to address this opportunity, it becomes clear why we are optimistic about growth over the next few years.-Rajesh Subramaniam President & Chief Operating Officer, FedEx Corp. Fiscal third quarter 2019 earnings conference call, March 19, 2019. Takeaway 3—There is still substantial integration work to be done with TNT: We were previously modeling total cumulative TNT integration costs at $1.53B (pre-tax) through the end of fiscal year 2020. Now it seems this will stretch into 2021. Hey. Thanks for the time. Just want to clarify some of the TNT integration-related expenses. Obviously, you raised that target a little bit. You pushed the timeline out to 2021. Just wanted to clarify, is the integration now going to take a little bit longer than expected and cost a little bit more? How should we interpret that data, particularly the greater than $1.5 billion target? And how do we reconcile that in the context of the fact that the integration-related expenses here in fiscal 2019 are coming at a little bit less than expected? -Benjamin J. Hartford, Analyst, Robert W. Baird & Co., Inc. Fiscal third quarter 2019 earnings conference call, March 19, 2019. Well, Ben, I would say that – what we have found is that – what we've been talking to you mostly about were the operational integration costs of putting these networks together. Again as Raj talked to, I mean we're starting to get there and we're going to pick up a lot of momentum in FY20. 199

What we're seeing, however, is that as Rob's team continues to bring those antiquated systems up to current standards and beyond, that we probably have some back-office and other areas where we can do further investment. And so, we'll keep you updated on that. But as far as operationally, we tend to think we're going to be there by the end of FY20 and that's the biggest part. So don't be scared about a lot of additional integration money; there won't be a lot of additional integration money.-Alan B. Graf, Jr. Executive Vice President & CFO, FedEx Corp. Fiscal third quarter 2019 earnings conference call, March 19, 2019. Resulting Model Update: Increase the total TNT integration expense to $1.6B. The incremental increase in total expense will be entered in fiscal year 2021. Takeaway 4—U.S. Business was impacted by mix: In the preview we noted that the U.S. economic conditions did not warrant a revision to our forecast; however, this was a major source of why our forecasted results were higher than the reported results. Management has stated that this had to do with a shift in mix toward more e-commerce business with lower yields. Great. Good afternoon, guys. Thanks for the time. Just wanted to go back to the comments around softening global macro condition and softening trade trends. I know you're specifically calling out Asia and Europe. But when you look through your U.S. Express results, I think it's fair to say you saw volume declines in three of your four U.S. Express package subsegments. You also saw yield deceleration in your U.S. Express business. So if you could just sort of comment on what you're seeing in the U.S. I know you're talking about a fairly strong U.S. backdrop, but it seems like we're seeing deceleration there. We'd just be curious what's driving that. -Jack Atkins, Analyst, Stephens, Inc, Fiscal third quarter 2019 earnings conference call, March 19, 2019. Hey, Jack. This is Raj. I think on the U.S. side, our overall volume continues to increase at roughly 6%. And I think if you look at what's driving that growth, it is e-commerce. What we are surprised by is the fact that the weight per package has been lower than what we anticipated it is going to be, and that has impacted our yields. So, we are doing – first of all, as I said, we are in a position to gain share on e-commerce. We will adjust our cost structure to make sure that we have the right cost to serve. We are very focused on driving overnight volume growth and we are focused on revenue management as well. So there's a lot of activity going on in the U.S. today, and I'm quite optimistic that we're going to get this right. Rajesh Subramaniam, President & Chief Operating Officer, FedEx Corp, -Rajesh Subramaniam President & Chief Operating Officer, FedEx Corp. Fiscal third quarter 2019 earnings conference call, March 19, 2019. Resulting Model Update: Since we expect the shift to greater volume of e-commerce to continue, we should revise our future period U.S. Express yield forecasts lower. Takeaway 5—Earnings guidance was lowered again: Non-GAAP EPS guidance decreased from a range between $15.50 and $16.60, to $15.10 to $15.90. The following comments from the earnings call provide descriptions of the segment results, with details of what is included in the new lower guidance for the fiscal year. FedEx Express Results: Slowing international macroeconomic conditions and weaker global trade growth trends continue. Asia volume weakness, which we experienced during peak season, deepened post Chinese New Year. Reflecting these macro challenges, FedEx Express international revenues declined year-over-year in the third quarter. U.S. volume growth continued to benefit from the expansion of our e-commerce solutions, but yields were pressured by this expansion, lower weight per shipment, and service mix changes. With that, total FedEx Express revenue declined 1% year-overyear in Q3 versus growth of 8% year-over-year in the first half of this fiscal year.-Alan B. Graf, Jr. Executive Vice President & CFO, FedEx Corp. Fiscal third quarter 2019 earnings conference call, March 19, 2019. FedEx Express Outlook: We expect earnings to decrease year-over-year at FedEx Express in the fourth quarter, due to lower yields and continued softness in international package volumes resulting from weakening global economic conditions, particularly in Asia and Europe.-Alan B. Graf, Jr. Executive Vice President & CFO, FedEx Corp. Fiscal third quarter 2019 earnings conference call, March 19, 2019. 200

Appendix 4: FedEx Fiscal 3Q2019 Earnings Release Resulting Model Update: Decrease the estimates of yield in FedEx Express U.S. sub-segments, and decrease yield and volume in Express International. FedEx Ground Results: FedEx Ground operating results were negatively impacted by the inflationary impact of the tight labor market on our purchase transportation rates and employee wages. Higher costs related to staffing and network expansion and the January launch of year-around six-day per week operations also impacted Ground's performance. While the launch of six-day operations was a headwind for the quarter, the use of existing assets more efficiently is a positive for the FedEx Ground business, as it ultimately drives improved performance and enhances our competitive position. While these benefits are not always reflected immediately and may take time to be realized, we believe it is a winning strategy for the long term.Alan B. Graf, Jr. Executive Vice President & CFO, FedEx Corp. Fiscal third quarter 2019 earnings conference call, March 19, 2019. FedEx Ground Outlook: While FedEx Ground year-over-year revenue growth is expected to remain strong in the fourth quarter, higher operating costs are expected to continue to negatively impact results. -Alan B. Graf, Jr. Executive Vice President & CFO, FedEx Corp. Fiscal third quarter 2019 earnings conference call, March 19, 2019. Resulting Model Update: Increase the forecast of operating expenses for the FedEx Ground Segment. FedEx Freight Results: FedEx Freight continued to focus on balanced volume and yield growth and produced another strong quarter of operating results. -Alan B. Graf, Jr. Executive Vice President & CFO, FedEx Corp. Fiscal third quarter 2019 earnings conference call, March 19, 2019. FedEx Freight Outlook: We expect year-over-year revenue and earnings growth at FedEx Freight will continue in the fourth quarter, driven by balance of volume and yield growth. -Alan B. Graf, Jr. Executive Vice President & CFO, FedEx Corp. Fiscal third quarter 2019 earnings conference call, March 19, 2019. Resulting Model Update: No significant model changes required for the FedEx Freight Segment. Follow Along in the Spreadsheet: Refer to “File 30-FDX Model (Appendix 4-Step 2 F3Q2019 Review)” for the updated FedEx model after the F3Q2019 earnings release. Refer to the "How to Use This Textbook" section of Chapter 1 for instructions on how to access the spreadsheet files.

201

APPENDIX 5: EQUITY RISK PREMIUM MODEL UPDATE Appendix 5 Overview: Each time the Federal Reserve’s FOMC meets, the ERP model must be updated to reflect the latest details related to the Federal Funds rate. Going forward these updates will be tracked in the Appendix section of this textbook, and published in future editions. Appendix 5 includes the details of the FOMC’s January 30,2019 meeting and data through February 11, 2019.

Appendix 5: Step 1—FOMC Meeting Update

The Fed believes economic activity has remained consistent with the previous assessment. There was no change in the comments in regards to economic activity and job gains. At the January meeting, the Committee voted unanimously to maintain the current the target Federal Funds rate at a range of 2.25% to 2.50%. Follow Along in the Spreadsheet: Refer to “File 31 Equity Risk Premium Model (Appendix 5 February-2019)” for the updated ERP Model. Refer to the "How to Use This Textbook" section of Chapter 1 for instructions on how to access the spreadsheet files.

Latest FOMC Statement from January 30, 2019: No Change in Target Rate

Information received since the Federal Open Market Committee met in December indicates that the labor market has continued to strengthen and that economic activity has been rising at a solid rate. Job gains have been strong, on average, in recent months, and the unemployment rate has remained low. Household spending has continued to grow strongly, while growth of business fixed investment has moderated from its rapid pace earlier last year. On a 12-month basis, both overall inflation and inflation for items other than food and energy remain near 2 percent. Although market-based measures of inflation compensation have moved lower in recent months, survey-based measures of longer-term inflation expectations are little changed. Consistent with its statutory mandate, the Committee seeks to foster maximum employment and price stability. In support of these goals, the Committee decided to maintain the target range for the federal funds rate at 2-1/4 to 21/2 percent. The Committee continues to view sustained expansion of economic activity, strong labor market conditions, and inflation near the Committee’s symmetric 2 percent objective as the most likely outcomes. In light of global economic and financial developments and muted inflation pressures, the Committee will be patient as it determines what future adjustments to the target range for the federal funds rate may be appropriate to support these outcomes. In determining the timing and size of future adjustments to the target range for the federal funds rate, the Committee will assess realized and expected economic conditions relative to its maximum employment objective and its symmetric 2 percent inflation objective. This assessment will take into account a wide range of information, including measures of labor market conditions, indicators of inflation pressures and inflation expectations, and readings on financial and international developments. Voting for the FOMC monetary policy action were: Jerome H. Powell, Chairman; John C. Williams, Vice Chairman; Michelle W. Bowman; Lael Brainard; James Bullard; Richard H. Clarida; Charles L. Evans; Esther L. George; Randal K. Quarles; and Eric S. Rosengren. Source: FederalReserve.gov, FOMC Statement, January 30, 2019, retrieved February 4, 2019.

202

Appendix 5: Equity Risk Premium Model Update

Appendix 5: Step 2—Data Maintenance

Equity Risk Premium Updates: Each quarter the new interest rate, market return, and volatility results must be input into the Equity Risk Premium (ERP) model. The ERP model shown below in Exhibit 150 includes the information from the January 30, 2019 FOMC meeting, as well as market data through February 11, 2019. The new estimate of the 12month forward average VIX (on a trailing basis) is 18.67% (based on the 4Q2019 estimate, in cell J11). This estimate reflects a return to stability from the 4Q2018 average VIX which reached 21.05% (cell H15). The Fed's median Federal Funds projection for 2019 is 2.9%, and 3.1% for 2020. These estimates are included in the ERP model with just two 25 basis point increases in 2019, which are included in the second and fourth quarters in our model. This assumes the Fed takes a pause for the January 30, 2019 and March 20, 2019 meetings. The ERP model also includes one 25 basis point increase in the second quarter of 2020. The spread forecast moved from ~0.6% spread to ~1.0% by the end of 2020, which includes some effect from the balance sheet normalization program. The average 10-year U.S. Treasury rate (trailing) forecast in the model is now 3.36% for 2019 (cell O11). The resulting ERP estimate for 1Q2019 is 5.5%, with a 2019 forecast of 6.0%.

Exhibit 150—Latest Equity Risk Premium Updates

Follow Along in the Spreadsheet: Refer to “File 31 Equity Risk Premium Model (Appendix 5 February-2019)” for the updated ERP Model. Refer to the "How to Use This Textbook" section of Chapter 1 for instructions on how to access the spreadsheet files.

203

APPENDIX 6: EQUITY RISK PREMIUM MODEL UPDATE Appendix 6 Overview: Each time the Federal Reserve’s FOMC meets, the ERP model must be updated to reflect the latest details related to the Federal Funds rate. Going forward these updates will be tracked in the Appendix section of this textbook, and published in future editions. Appendix 6 includes the details from the March 20, 2019 FOMC meeting, and data through March 30, 2019.

Appendix 6: Step 1—FOMC Meeting Update

The primary change from the February meeting is that the Fed has recognized that economic activity has slowed to some extent. There was no change to the target Federal Funds rate; However, the 2019 target rate now stands at 2.40% (refer to the projection material in Exhibit 151 below). This means the Fed does not anticipate any rate hikes this year. To reflect this point, the ERP model has been updated to remove the previously anticipated 25 basis point increase. The ERP model was also updated to pull the 25 basis point increase in 2020 forward, from the second quarter to the first quarter. Follow Along in the Spreadsheet: Refer to “File 32 Equity Risk Premium Model (Appendix 6 March 2019 Update)” for the updated ERP Model. Refer to the "How to Use This Textbook" section of Chapter 1 for instructions on how to access the spreadsheet files.

Latest FOMC Statement from March 20, 2019: No change to the target Fed Funds rate

Information received since the Federal Open Market Committee met in January indicates that the labor market remains strong but that growth of economic activity has slowed from its solid rate in the fourth quarter. Payroll employment was little changed in February, but job gains have been solid, on average, in recent months, and the unemployment rate has remained low. Recent indicators point to slower growth of household spending and business fixed investment in the first quarter. On a 12-month basis, overall inflation has declined, largely as a result of lower energy prices; inflation for items other than food and energy remains near 2 percent. On balance, market-based measures of inflation compensation have remained low in recent months, and survey-based measures of longer-term inflation expectations are little changed. Consistent with its statutory mandate, the Committee seeks to foster maximum employment and price stability. In support of these goals, the Committee decided to maintain the target range for the federal funds rate at 2-1/4 to 21/2 percent. The Committee continues to view sustained expansion of economic activity, strong labor market conditions, and inflation near the Committee’s symmetric 2 percent objective as the most likely outcomes. In light of global economic and financial developments and muted inflation pressures, the Committee will be patient as it determines what future adjustments to the target range for the federal funds rate may be appropriate to support these outcomes. In determining the timing and size of future adjustments to the target range for the federal funds rate, the Committee will assess realized and expected economic conditions relative to its maximum employment objective and its symmetric 2 percent inflation objective. This assessment will take into account a wide range of information, including measures of labor market conditions, indicators of inflation pressures and inflation expectations, and readings on financial and international developments. Voting for the FOMC monetary policy action were: Jerome H. Powell, Chairman; John C. Williams, Vice Chairman; Michelle W. Bowman; Lael Brainard; James Bullard; Richard H. Clarida; Charles L. Evans; Esther L. George; Randal K. Quarles; and Eric S. Rosengren. Source: FederalReserve.gov, FOMC Statement, March 20, 2019, retrieved March 25, 2019. 204

Appendix 6: Equity Risk Premium Model Update

Exhibit 151—Latest Projection Materials Example (March-2019 Meeting)

Source: FederalReserve.gov, FOMC Projections Materials, March 20, 2019, retrieved March 25, 2019.

Appendix 6: Step 2—Data Maintenance

Equity Risk Premium Updates: In addition to the Federal Funds target updates, the regular ERP data maintenance has been updated in the ERP model to include the latest available equity returns, volatility, and spread between the 10year U.S. Treasury security and the Federal Funds rate. The spread forecast has been updated to ~1.1% by the end of 2020, which includes an estimate of the effect from the Balance Sheet normalization program. Impact on the ERP Estimate: The ERP estimate for 1Q2019E now stands at 5.4% versus the previous estimate of 5.5%. The slight decrease from the last update reflects the impact of lower interest rates (Federal Funds and spread) offset by increased volatility and lower equity market return projections. The 4Q2019E ERP estimate is now 5.6%, and remains below the historic average of 6.0%.

205

Exhibit 152—Latest Equity Risk Premium Updates

Follow Along in the Spreadsheet: Refer to “File 32 Equity Risk Premium Model (Appendix 6 March 2019 Update)” for the updated ERP Model. Refer to the "How to Use This Textbook" section of Chapter 1 for instructions on how to access the spreadsheet files.

206

APPENDIX 7: EQUITY RISK PREMIUM MODEL UPDATE Appendix 7 Overview: Each time the Federal Reserve’s FOMC meets, the ERP model must be updated to reflect the latest details related to the Federal Funds rate. Going forward these updates will be tracked in the Appendix section of this textbook, and published in future editions. Appendix 7 includes the details from the May 1, 2019 FOMC meeting, and data through May 24, 2019.

Appendix 7: Step 1—FOMC Meeting Update

The Fed seems to be in a “wait and see” mode, after noting the decrease in capital investment and household spending in the first quarter. The Committee voted unanimously to maintain the current target Federal Funds rate; however, based on the commentary from Chairman Jerome Powell at the press conference, it now seems unlikely that there will be any rate increases in 2020. The results of the May-2019 Market Participant Survey from the New York Fed reflect the market’s view which is consistent with no changes in the target rate for 2020. To include this point, the ERP model has been updated to remove the 25 basis point increase in rates for 2020. As a result the Federal Funds rate remains flat at 2.40% through the end of 2020 (which is the midpoint of the Fed's 2.25% to 2.50% range). Follow Along in the Spreadsheet: Refer to “File 33 Equity Risk Premium Model (Appendix 7 April 2019 Update)” for the updated ERP Model. Refer to the "How to Use This Textbook" section of Chapter 1 for instructions on how to access the spreadsheet files.

Latest FOMC Statement from May 1, 2019: No change to target range

Information received since the Federal Open Market Committee met in March indicates that the labor market remains strong and that economic activity rose at a solid rate. Job gains have been solid, on average, in recent months, and the unemployment rate has remained low. Growth of household spending and business fixed investment slowed in the first quarter. On a 12-month basis, overall inflation and inflation for items other than food and energy have declined and are running below 2 percent. On balance, market-based measures of inflation compensation have remained low in recent months, and survey-based measures of longer-term inflation expectations are little changed. Consistent with its statutory mandate, the Committee seeks to foster maximum employment and price stability. In support of these goals, the Committee decided to maintain the target range for the federal funds rate at 2-1/4 to 21/2 percent. The Committee continues to view sustained expansion of economic activity, strong labor market conditions, and inflation near the Committee’s symmetric 2 percent objective as the most likely outcomes. In light of global economic and financial developments and muted inflation pressures, the Committee will be patient as it determines what future adjustments to the target range for the federal funds rate may be appropriate to support these outcomes. In determining the timing and size of future adjustments to the target range for the federal funds rate, the Committee will assess realized and expected economic conditions relative to its maximum employment objective and its symmetric 2 percent inflation objective. This assessment will take into account a wide range of information, including measures of labor market conditions, indicators of inflation pressures and inflation expectations, and readings on financial and international developments. Voting for the FOMC monetary policy action were: Jerome H. Powell, Chair; John C. Williams, Vice Chair; Michelle W. Bowman; Lael Brainard; James Bullard; Richard H. Clarida; Charles L. Evans; Esther L. George; Randal K. Quarles; and Eric S. Rosengren. Source: FederalReserve.gov, FOMC Statement, May 1, 2019, retrieved May 5, 2019.

Appendix 7: Step 2—Data Maintenance

Equity Risk Premium Updates: In addition to the Federal Funds target updates, the ERP model has also been updated to include the latest available market equity returns, volatility, and the spread between the 10-year U.S. Treasury Security and the Federal Funds rate. The spread forecast includes a reversion towards the longer-term average of 1.64%, with the spread moving from 0.25% on average in 1Q2019 to 1.30% by the end of 2020. The latest ERP estimate for 2Q2019E has moved to 5.3% versus the previous estimate of 5.5%, while the 12-month forward estimate (1Q2020) has decreased from 5.7% to 5.6%. The model-based ERP forecast remains below the historic average of 6.0%. The estimates used in the DCF-based share valuation of the FedEx model now reflect the following forecasts: • Market Sharpe ratio including 2019 estimates: 0.319 • Expected average market volatility: 17.17% (above the current average) • Expected average risk-free rate of return: 3.03% (above the current average)

Exhibit 153—Latest Equity Risk Premium Updates

Follow Along in the Spreadsheet: Refer to “File 33 Equity Risk Premium Model (Appendix 7 April 2019 Update)” for the updated ERP Model. Refer to the "How to Use This Textbook" section of Chapter 1 for instructions on how to access the spreadsheet files.

208

APPENDIX 8: EQUITY RISK PREMIUM MODEL UPDATE Appendix 8 Overview: Each time the Federal Reserve’s FOMC meets, the ERP model must be updated to reflect the latest details related to the Federal Funds rate. Going forward these updates will be tracked in the Appendix section of this textbook, and published in future editions. Appendix 8 includes the details from the June 19, 2019 FOMC meeting, and market data through July 6, 2019. This Appendix entry will also clarify and standardize the approach used for forming the future forecast for each metric.

Appendix 8: Step 1—FOMC Meeting Update

The primary change in the June FOMC meeting Statement is that household spending has improved. The Projection Material, in Exhibit 154 below shows that the FOMC members now expect the Federal Funds target to be 2.1% in 2019, down from the previous 2.6%, and 2.4% in 2020 down from the previous 2.6%. It is interesting to note that there was a dissenting member for the June vote, James Bullard who was in favor of a 25 basis point decrease. Subsequent to the June FOMC meeting, the U.S. Jobs report for the month of June (released the first week in July) showed strong results with 224,000 jobs added for the month, well ahead of the consensus expectation of 165,000. This is an important data point heading into Jerome Powell’s testimony before the House Financial Services Committee on July 10, 2019 and the Senate Banking Committee on July 11, 2019. The market remains ahead of the FOMC in terms of an expected rate cuts, as shown in the CME Group’s FedWatch Tool, which estimates the market’s expectation of the Federal Funds rate using pricing of Federal Funds futures contracts. The ERP Model has been updated to reflect the market expected Federal Funds rate, using the CME’s probability distribution for each FOMC meeting. The resulting Federal Funds rate by quarter is as follows: 3Q2019E 2.15%, 4Q2019E 1.86%, 1Q2020E 1.70%, and 2Q2020E 1.62%. Follow Along in the Spreadsheet: Refer to “File 34 Equity Risk Premium Model (Appendix 8 June-2019 Update)” for the updated ERP Model. Refer to the "How to Use This Textbook" section of Chapter 1 for instructions on how to access the spreadsheet files.

Latest FOMC Statement from June 19, 2019: No change to the target rate

Information received since the Federal Open Market Committee met in May indicates that the labor market remains strong and that economic activity is rising at a moderate rate. Job gains have been solid, on average, in recent months, and the unemployment rate has remained low. Although growth of household spending appears to have picked up from earlier in the year, indicators of business fixed investment have been soft. On a 12-month basis, overall inflation and inflation for items other than food and energy are running below 2 percent. Market-based measures of inflation compensation have declined; survey-based measures of longer-term inflation expectations are little changed. Consistent with its statutory mandate, the Committee seeks to foster maximum employment and price stability. In support of these goals, the Committee decided to maintain the target range for the federal funds rate at 2-1/4 to 2-1/2 percent. The Committee continues to view sustained expansion of economic activity, strong labor market conditions, and inflation near the Committee’s symmetric 2 percent objective as the most likely outcomes, but uncertainties about this outlook have increased. In light of these uncertainties and muted inflation pressures, the Committee will closely monitor the implications of incoming information for the economic outlook and will act as appropriate to sustain the expansion, with a strong labor market and inflation near its symmetric 2 percent objective. 209

In determining the timing and size of future adjustments to the target range for the federal funds rate, the Committee will assess realized and expected economic conditions relative to its maximum employment objective and its symmetric 2 percent inflation objective. This assessment will take into account a wide range of information, including measures of labor market conditions, indicators of inflation pressures and inflation expectations, and readings on financial and international developments. Voting for the monetary policy action were Jerome H. Powell, Chair; John C. Williams, Vice Chair; Michelle W. Bowman; Lael Brainard; Richard H. Clarida; Charles L. Evans; Esther L. George; Randal K. Quarles; and Eric S. Rosengren. Voting against the action was James Bullard, who preferred at this meeting to lower the target range for the federal funds rate by 25 basis points. Source: FederalReserve.gov, FOMC Statement, June 19, 2019, retrieved June 25, 2019.

Exhibit 154—Latest Projection Materials Example (June-2019 Meeting)

Source: FederalReserve.gov, FOMC Projections Materials, June 19, 2019, retrieved June 25, 2019.

210

Appendix 8: Equity Risk Premium Model Update

Appendix 8: Step 2—Data Maintenance

Summary of Forecast Approach: Each quarter the new interest rate, market return, and volatility results must be input into the Equity Risk Premium (ERP) model, in addition new estimates must be entered for each metric. This section will clarify the approach I use to form the future expectations in the “base-case” version of the ERP model. •

Volatility: I assume volatility will regress toward the long-term historic average. To do this I take the difference between the current quarterly average VIX and the long-term average, and divided it by the remaining number of quarters contained within the model. Then add the incremental change in volatility each quarter to smooth the forecast back to the historic average. o

o

Latest Forecast: The trailing one-year average VIX estimate used in the ERP model is now 16.52%. User Input: If you believe the market is headed for a period of uncertainty, you can change the blue input cells to increase the future volatility assumptions. If you believe the market will stabilize, you can decrease the volatility assumptions.

•

Fed Funds Rate: In the base-case scenario I assume the Federal Funds rate will increase or decrease based on the market's expectations as approximated by the CME's FedWatch tool, which uses Federal Funds futures contracts to asses the probability of future rate changes. As time passes, compare the market's expectations to the FOMC's latest projection material, to determine if the market outlook is dislocated from that of the FOMC members. o Latest Forecast: As described in Appendix 8, Step 1 the forecasted Federal Funds rate by quarter is as follows: 3Q2019E 2.15%, 4Q2019E 1.86%, 1Q2020E 1.70%, and 2Q2020E 1.62%. o User Input: If you believe rates will be higher or lower in the future, then change the Federal Funds input cells.

•

Spread Between 10-year Treasury and Federal Funds Rate: I use 10-year U.S. Treasury futures contracts to approximate the expected spread for the next quarter. Then regress the spread towards the long-term average spread adjusting by 25 basis points per quarter, which reflects the fact that spreads can persist above or below the historic average for long periods of time. o Latest Forecast: The model risk-free rate, which is the projected average 10-year U.S. Treasury rate, based on the Federal Funds forecast and spread projections is now 1.96%. o User Input: In general, if the market expects the economy to expand, the yield curve will steepen and the spread will increase. If the market expects the economy to contract the spread will decrease. If you believe we are headed for an expansion, then you may want to increase the spread assumption in the model. If you believe we are headed for a contraction, then you may want to decrease the spread assumption in the model.

•

Market Return Assumption: The average total return for the S&P500 through 1963 has been 11.4%. The market return assumptions in our base case scenario assume that the average annual return in the future will be approximately equal to the historic average. o Latest Forecast: The 2019 Constant Sharpe ratio based on the market return expectations and riskfree rate forecast is now 0.319. o User Input: If you believe market returns will be higher or lower in the future, then adjust the forecast returns in the model.

The latest estimates from the revised ERP model used in the DCF-based valuation include the following: • Expected average market volatility: 16.52% (below the current average) • Expected average risk-free rate of return: 1.96% (above the current average) • Resulting stage-one ERP estimate: 5.4% • Required return on equity assuming a beta of 1.0: 7.3%

211

Exhibit 155—Latest Equity Risk Premium Updates

Follow Along in the Spreadsheet: Refer to “File 34 Equity Risk Premium Model (Appendix 8 June-2019 Update)” for the updated ERP Model. Refer to the "How to Use This Textbook" section of Chapter 1 for instructions on how to access the spreadsheet files.

212

APPENDIX 9: FEDEX FISCAL 4Q2019 EARNINGS RELEASE Step 1: Prepare for the Release

Step 2: Update After the Release

Appendix 9 Overview: Each time a new quarterly earnings report is released, or when significant company or market developments occur, the FedEx earnings model will need to be updated. Going forward these updates will be tracked in the Appendix section of this textbook, and published in future editions. This represents the earnings preview/review cycle, which demonstrates that a model is a living file, and must be maintained continuously.

Appendix 9: Step 1—Prepare for the F4Q2019 Release (Earnings Preview)

Note: This section was updated on June 20, 2019 before the earnings release on June 25, 2019. The following is a list of steps to consider, and questions to think about, as part of the quarterly earnings preparation: Review the relevant leading indicators, and latest news from the investor relations page as a final check to determine if there are additional items to consider in the forecast, ahead of the earnings release. • Update the valuation metrics to reflect the latest data including: the current share price, last three-month forward Price-to-Earnings multiple, stage-one beta, volatility, interest rates, mean monthly return and standard deviation. • Consider the consensus estimate. Are there any points in the consensus earnings forecast which we may be missing in our model? Look to the high-end, and low-end of the consensus range. How has the consensus revenue, EPS estimate, and share price target changed since the last quarter? How does the change in the consensus estimate compare to the change in macroeconomic conditions over the course of the quarter? • How does the share value look relative to recent history? Is the Price-to-Earnings ratio above or below the historic average? • Save a copy of the existing model in an archive folder. • Update the model headers. • Change the forecast equations to values, and insert ratio equations. • Copy and paste values on the results comparison table (refer to Exhibit 121). • Review last quarter’s press release, earnings transcript, and presentation slides as a refresher on the important topics to watch for in the next press release. Review the Developments & Leading Indicators Over the Quarter: The trade “war” between the U.S. and China is ongoing, and global economic conditions remain relatively consistent with our last model update, so there is little to adjust. One significant development has emerged. On June 7, 2019 FedEx announced that it would not renew its FedEx U.S. Domestic Express contract with Amazon. The company estimates that in 2018 Amazon made up less than 1.3% of revenue. Some of the lost revenue from Amazon will be replaced as FedEx continues to grow its e-commerce business with other retailers. In the mean time we have brought our Express Segment volume estimates down over the course of the next four quarters to reflect this development. In addition to the usual FedEx metrics, we should also review our assumptions for the pension plan for two reasons: 1) The company calculates the mark-to-market adjustment in the fourth fiscal quarter, and 2) Interest rates have decreased significantly with the average 10-year U.S. Treasury rate down about 1% over the course of the year. •

213

Currently we are modeling an 11 basis point decrease in the discount rate, which we can increase to 50 basis points as a rough estimate (enter this forecast in cell W216 of File 36). As a reminder the model adjustments we made after the F3Q2019 earnings release which are currently reflected in our forecast include the following: 1) Decreased yield and volume estimates to reflect weak global economic conditions. 2) Increased TNT integration expense to $1.6B and moved some expense to 2021. 3) To reflect an increased share of e-commerce business (other than Amazon) we decreased our yield forecast with the U.S. Express segments. 4) Increased the operating expense forecast for FedEx Ground to reflect the effects of a tight labor market and higher purchased transportation rates. Update the Valuation Metrics—All Other Items: The following changes have been incorporated into the model to reflect the latest data available as of June 20, 2019. The comparisons represent changes from the model after the F3Q2019 release reflected in Appendix 4 “Step 2—Update After the F3Q2019 Release (Earnings Review)” on March 20, 2019: •

The FedEx share price has decreased to $169 from $176 in the previous model update. This has resulted in a decrease in the equity-to-total capital ratio to approximately 71% from 72%. This ratio is used in the Weighted Average Cost of Capital (WACC) calculation. Since equity has a lower cost relative to debt, a decrease in equity value will decrease the weight of equity and the discount rate (the WACC), which will increase the Net Present Value (NPV) of the future cash flows in the target DCF calculation.

•

Market volatility (VIX, 1-year trailing average) has decreased to 16.5% from 18.7%. Refer to “File 34 Equity Risk Premium Model (Appendix 8-June 2019 Update)” and Appendix 8 for additional details. The 16.5% average market volatility is shown in cell C389 within File 36.

•

The long-term beta coefficient (stage-two). Using a regression of the percentage change in FedEx shares versus the S&P500, beta moved from 1.32 to 1.34 (entered within the CAPM equation in cell C402 of File 36). The short-term beta (stage-one) moved from 1.59 to 1.83 (entered in cell C387 of File 36). For the details of the beta calculations refer to “File 35-FDX Beta Calculation (Appendix 9-Step 1 F4Q2019 Preview).”

•

The quarterly average 10-year U.S. Treasury rate has decreased significantly to 1.96% from 3.36%, as the market is now pricing in expected interest rate cuts from the FOMC. Refer to “File 34 Equity Risk Premium Model (Appendix 8-June 2019 Update)” and Appendix 8 for additional details. The 1.96% rate is included in cell C391 within File 36.

•

The three-month average, Next Twelve Month (NTM) forward Price-to-Earnings ratio has decreased to 14.8x. Shares are trading near the three-month average heading into the release.

•

The 12-month absolute value of the mean monthly return in FedEx shares stands at 2.5%, and the standard deviation in returns has increased to 10.9%.

After incorporating each of these items into the model, including the change in beta, the target share valuation has decreased to $162 from $173, and the new target price band is $131 to $201, which is wider than the previous band due to the higher standard deviation in average monthly returns. Follow Along in the Spreadsheet: Refer to “File 36-FDX Model (Appendix 9-Step 1 F4Q2019 Preview)” for the updated FDX model prior to the F4Q2019 earnings release. Refer to the "How to Use This Textbook" section of Chapter 1 for instructions on how to access the spreadsheet files.

214

Appendix 9: FedEx Fiscal 4Q2019 Earnings Release

Appendix 9: Step 2—Update After the F4Q2019 Release (Earnings Review) Note: This section was updated on June 25, 2019, after the earnings release. Key Takeaways from the F4Q2019 Conference Call: There were three primary points raised on this call. Most important of which was the first look at the fiscal 2020 guidance. Global trade issues and macroeconomic headwinds remain a key topic for FedEx. Also, after multiple conference calls discussing how important (or un-important) Amazon.com is to FedEx, management spent time discussing the plan for e-commerce transformation having dropped Amazon as a customer. Takeaway 1—First look at the Fiscal 2020 Guidance: Overall EPS growth guidance fell short of the consensus estimate. Capital expenditures are expected to be up approximately 8% year-over-year to $5.9B as fleet updates and hub upgrades continue in Memphis and Indianapolis. Projected TNT integration costs increased again. The expected tax rate also increased from fiscal 2019 to a range of 23% to 25%. Model Update: For the fiscal 2020 forecast the TNT integration expense and tax rate estimates were increased, growth rates and yield in the Express Segment were decreased to get back to the EPS growth guidance. Takeaway 2—Global trade and macroeconomic environment: Overall, U.S. economic growth is holding up well with real GDP growing 3.1% in the first quarter of calendar year 2019. However, the industrial sector has suffered from an inventory build up and increased trade tensions. Through May, manufacturing output was 1.5% off its December peak. For calendar year 2019, we expect global economic growth to moderate, as the developed world sees slower growth, and both domestic and external factors weigh on emerging markets. We expect services to continue to underpin global GDP growth. Global trade has slowed as trade frictions have exerted a negative impact on sentiments and of course the manufacturing sector. As the Chinese economy has continued to decelerate, this has also impacted other Asian markets’ export performance. China’s exports, which grew almost 10% in 2018, have barely grown this year amid heightened trade tensions with the United States. Outlook for the European economies remains slow due to a number of sector and country-specific factors, such as disruptions in the auto manufacturing sector, social tensions, policy uncertainty, as well as uncertainty related to Brexit. In Germany and Italy, which are two important markets for us, manufacturing output, which turned negative in November, has continued to decline with April manufacturing output down 3.5% and 1.9% year-over-year in Germany and Italy, respectively. -Brie Carere, FedEx Corp, Chief Marketing and Communications Officer & Executive Vice President, Fiscal fourth quarter 2019 earnings conference call, June 25, 2019. Model Update: Decreased the volume growth rates in the Express International businesses. Takeaway 3—E-Commerce Transformation: E-commerce continues to be a driving force of total U.S. domestic market growth. We are building our portfolio, network and capacity to best serve thousands of retailers in this space. And we continue to differentiate, for example, with the launch of the FedEx 7-Day service. In 2020, FedEx Ground will deliver seven days a week, year-round for 80% of U.S. GDP. This is truly a transformational move that builds upon the largest global commerce transportation network in the world to further serve the growing e-commerce market. We are already faster than the competition by at least one day in 25% of the lanes. This move will further speed up our network and allow us to continue to gain market share…. While we see a rational pricing market, e-commerce will continue to put pressure on yields with lighter packages moving shorter distances. It is important to note that contrary to the erroneous and misinformed reporting in The Wall Street Journal on June 23, FedEx has made no recent pricing changes, from no pricing changes to our strategy, and we have certainly made no changes related to any one customer. -Brie Carere, FedEx Corp, Chief Marketing and

215

Communications Officer & Executive Vice President, Fiscal fourth quarter 2019 earnings conference call, June 25, 2019 Let me also caution observers who follow FedEx and this industry to be very careful extrapolating past assumptions and trends into the future. For instance, we’ve noted repeatedly, short haul package delivery will become increasingly important as retailers ship e-commerce orders from store or local fulfillment. Hence average yields have to be matched with operational changes, not visible to most, to assure potential future profitability. Also, future developments in speeding up e-commerce deliveries and postal reform, which by the way we have supported, will likely be discontinuities in the next several years. -Frederick W. Smith, FedEx Corp, Chairman, President & CEO, Fiscal fourth quarter 2019 earnings conference call, June 25, 2019. Model Update: Decreased the yield for the Express business to recognize the impact of the e-commerce transformation, and note that comparisons to historic results will become increasingly less reliable as the nature of the business continues to changes.

Results and Outlook Comments by Segment FedEx Express FedEx Express fourth quarter operating income declined, as weakness in global trade and industrial production drove a decline in our International Priority revenue. Year-over-year comparisons were also impacted by an $85 million gain on the sale of a non-core business of TNT Express last year. To mitigate the weakness, we have undertaken several immediate cost-containment actions, including significant reductions of variable incentive compensation, limiting hiring and discretionary spending, and completing our U.S. voluntary employee buyout program. FedEx Express Outlook: At FedEx Express, we expect earnings to be down in FY 2020, due to weakness in international priority revenue and ongoing shifts to lower-yielding services. Our strategic decision to not renew the FedEx Express U.S. domestic contract with Amazon will also be a near-term headwind, which we expect reverse to a positive in FY 2021 as we replace the lost volume and optimize the network. Additionally, we do not expect a significant benefit from the fuel surcharge table changes in FY 2019 to repeat in FY 2020. FedEx Express will continue to implement actions to reduce cost to serve, improve efficiencies and adjust its global network to match anticipated demand. While we expect to make significant progress on TNT integration activities in FY 2020, integration work will continue in FY 2021. -Alan Graf Jr., FedEx Corp, Executive Vice President & CFO, Fiscal fourth quarter 2019 earnings conference call, June 25, 2019. Global trade disputes and low global growth rates create significant uncertainty for the Express business, leading us to be cautious in projecting FY 2020 earnings for this segment. The integration of TNT is now progressing at a good clip and we will see significant benefits by this time in summer 2021 -Frederick W. Smith, FedEx Corp, Chairman, President & CEO, Fiscal fourth quarter 2019 earnings conference call, June 25, 2019. Model Update: Decreased FedEx Express International volumes, and U.S. yield. Model overall FedEx Express earnings down year-over-year. At FedEx Ground, we continued to see strong e-commerce volume growth in the fourth quarter. However, FedEx Ground income and margins were negatively impacted by increased purchase transportation rates and the January launch of year-round six-day-per-week operations. FedEx Ground Outlook: At FedEx Ground, we expect volume and revenue growth to remain very strong in FY 2020. However, operating margins will face headwinds from higher operating cost associated with expanding FedEx’s Ground’s delivery schedule, improving our capabilities for large packages, and other investments to significantly improve efficiency and safety. -Alan Graf Jr., FedEx Corp, Executive Vice President & CFO, Fiscal fourth quarter 2019 earnings conference call, June 25, 2019. We believe FedEx Ground will increase earnings for the fiscal year with modest, if any, margin compression from current levels despite the investments we’ve announced such as six and seven-day per week delivery, large package capabilities, and insourcing of SmartPost. -Frederick W. Smith, FedEx Corp, Chairman, President & CEO, Fiscal fourth quarter 2019 earnings conference call, June 25, 2019. 216

Appendix 9: FedEx Fiscal 4Q2019 Earnings Release Model Update: Model overall FedEx Ground earnings growth driven by volume growth, and stable operating margin. FedEx Freight closed the year with another strong quarter, despite weakening industrial production. Revenue per shipment increased 4%, operating income increased 15%, and operating margin improved to 9.9%. FedEx Freight Outlook: At FedEx Freight, we expect continued improved performance as we remain focused on improving revenue quality by implementing technology solutions that will drive efficiency and further differentiate us in the LTL market. -Alan Graf Jr., FedEx Corp, Executive Vice President & CFO, Fiscal fourth quarter 2019 earnings conference call, June 25, 2019. Let me emphasize, however, that based on our current forecast of U.S. GDP growth for FY 2020, we anticipate FedEx Freight will increase earnings and margin over the period. -Frederick W. Smith, FedEx Corp, Chairman, President & CEO, Fiscal fourth quarter 2019 earnings conference call, June 25, 2019. Model Update: Model overall FedEx Freight earnings growth, and an improvement in operating margin. Follow Along in the Spreadsheet: Refer to “File 37-FDX Model (Appendix 9-Step 2 & 3 F4Q2019 Review)” for the updated FedEx model after the F4Q2019 earnings release. Refer to the "How to Use This Textbook" section of Chapter 1 for instructions on how to access the spreadsheet files.

Appendix 9: Step 3—Recalibrate to the Post-Earnings Consensus Note: This section was updated on July 8, 2019 after the Equity Risk Premium close for the second calendar quarter. Recalibration Logic: At Gutenberg Research we provide consensus-based earnings models to act as a “base-case” scenario which our community members can input their own assumptions into. About a week after the earnings release we update the model to include the latest consensus estimates and any additional updates. To recalibrate the model to the latest consensus estimates I have adjusted the international sub-segment growth rates, and the operating expense assumptions. To recalibrate the model to meet the latest consensus estimates for the next four quarters, I applied adjustments to the Express U.S. yield estimates. The logic for booking the adjustments here is that we had already applied declines in the international businesses to recognized global macro headwinds, and the remaining differences between our model and the consensus estimates were relatively small, which could be explained by slight differences in the expected impact of the e-commerce transformation. For the recalibration to meet the annual consensus estimates I adjusted the ADV and yield growth rates for the International Express sub-segments. Other Model Updates: The model has been changed slightly to conform to the Gutenberg Research consensus-based model format. As a result the Price-to-Earnings multiple is now calculated including the value of net cash/(debt), and the notes within the valuation section have been moved to embedded comments to simplify the file. Follow Along in the Spreadsheet: Refer to “File 37-FDX Model (Appendix 9-Step 2 & 3 F4Q2019 Review)” for the updated FedEx model after the F4Q2019 earnings release. Refer to the "How to Use This Textbook" section of Chapter 1 for instructions on how to access the spreadsheet files.

217

APPENDIX 10: EQUITY RISK PREMIUM MODEL UPDATE Appendix 10 Overview: Each time the Federal Reserve’s FOMC meets, the ERP model must be updated to reflect the latest details related to the Federal Funds rate. Going forward these updates will be tracked in the Appendix section of this textbook, and published in future editions. Appendix 10 includes the details from the July FOMC meeting, and data through July 31, 2019. Note: Refer to Appendix 8 for a description of the ERP Model methodology.

Appendix 10: Step 1—FOMC Meeting Update

The latest FOMC meeting resulted in a 25 basis point decrease in the target Federal Funds rate. This is the first rate cut since 2008. The decision was in-line with expectations (refer to Appendix 8 for prior market expectations); however, the market reacted negatively to Chairman Powell’s explanation of the cut representing what he referred to as a “midcycle adjustment”. This was initially interpreted to mean that the July cut did not signal the start of a rate cutting cycle. Later in the press conference the Chairman explained that additional future cuts are possible, which stabilized the market, but left spectators someone confused. Ultimately, leaving the door open to future cuts without backing the FOMC into a situation where the market prices in a 100% probability of another cut, which is probably a favorable position for the FOMC. The market’s expectations for future interest rates remain nearly unchanged from our previous check in June (refer to Appendix 8). The CME Group’s FedWatch Tool, which estimates the market’s expectation of the Federal Funds rate using pricing of Federal Funds futures contracts, currently implies Federal Funds rates of 2.00% for September, 1.88% for October, and 1.80% for December. This compares to 1.95%, 1.87%, and 1.76% for September, October, and December respectively prior to the latest FOMC meeting. Overall the market continues to price in approximately 30 basis points of further rate cuts for the year. We will see how this estimate changes with the next round of economic data. Follow Along in the Spreadsheet: Refer to “File 38 Equity Risk Premium Model (Appendix 10 July-2019 Update)” for the updated ERP Model. Refer to the "How to Use This Textbook" section of Chapter 1 for instructions on how to access the spreadsheet files.

Latest FOMC Statement from July 31, 2019: 25 basis point rate cut. First Cut Since 2008!

Information received since the Federal Open Market Committee met in June indicates that the labor market remains strong and that economic activity has been rising at a moderate rate. Job gains have been solid, on average, in recent months, and the unemployment rate has remained low. Although growth of household spending has picked up from earlier in the year, growth of business fixed investment has been soft. On a 12-month basis, overall inflation and inflation for items other than food and energy are running below 2 percent. Market-based measures of inflation compensation remain low; survey-based measures of longer-term inflation expectations are little changed. Consistent with its statutory mandate, the Committee seeks to foster maximum employment and price stability. In light of the implications of global developments for the economic outlook as well as muted inflation pressures, the Committee decided to lower the target range for the federal funds rate to 2 to 2-1/4 percent. This action supports the Committee’s view that sustained expansion of economic activity, strong labor market conditions, and inflation near the Committee’s symmetric 2 percent objective are the most likely outcomes, but uncertainties about this outlook remain. As the Committee contemplates the future path of the target range for the federal funds rate, it will continue to monitor the

218

Appendix 10: Equity Risk Premium Model Update implications of incoming information for the economic outlook and will act as appropriate to sustain the expansion, with a strong labor market and inflation near its symmetric 2 percent objective. In determining the timing and size of future adjustments to the target range for the federal funds rate, the Committee will assess realized and expected economic conditions relative to its maximum employment objective and its symmetric 2 percent inflation objective. This assessment will take into account a wide range of information, including measures of labor market conditions, indicators of inflation pressures and inflation expectations, and readings on financial and international developments. The Committee will conclude the reduction of its aggregate securities holdings in the System Open Market Account in August, two months earlier than previously indicated. Voting for the monetary policy action were Jerome H. Powell, Chair; John C. Williams, Vice Chair; Michelle W. Bowman; Lael Brainard; James Bullard; Richard H. Clarida; Charles L. Evans; and Randal K. Quarles. Voting against the action were Esther L. George and Eric S. Rosengren, who preferred at this meeting to maintain the target range for the federal funds rate at 2-1/4 to 2-1/2 percent. Source: FederalReserve.gov, FOMC Statement, July 31, 2019, retrieved July 31, 2019.

Appendix 10: Step 2—Data Maintenance

Summary of Forecast Approach: Each quarter the new interest rate, market return, and volatility results must be input into the Equity Risk Premium (ERP) model, in addition new estimates must be entered for each metric. This section will describe the changes made to each metric. Refer to Appendix 8 for a description of the approach used to form the future expectations in the “base-case” version of the ERP model. •

Volatility—Latest Forecast: As a result of the recent stabilization in volatility, the trailing one-year average VIX estimate used in the ERP model has decreased to 14.89% down from the previous 16.52%. o User Input: If you believe the market is headed for a period of uncertainty, you can change the blue input cells to increase the future volatility assumptions. If you believe the market will stabilize, you can decrease the volatility assumptions.

•

Fed Funds Rate—Latest Forecast: Based on the month-end market expected Federal Funds rates described in Appendix 10 Step 1, the forecasted average Federal Funds rate by quarter has been updated as follows: 3Q2019E 2.17%, 4Q2019E 1.89%, 1Q2020E 1.74%, and 2Q2020E 1.66%. o

•

User Input: If you believe rates will be higher or lower, then change the Federal Funds input cells.

Spread Between 10-year Treasury and Federal Funds Rate—Latest Forecast: The model risk-free rate, which is the average 10-year U.S. Treasury rate based on the Federal Funds forecast and spread projections is now 1.99%, up just 3 basis points from the previous 1.96%. o User Input: In general, if the market expects the economy to expand, the yield curve will steepen and the spread will increase. If the market expects the economy to contract the spread will decrease. If you believe we are headed for an expansion, then you may want to increase the spread assumption in the model. If you believe we are headed for a contraction, then you may want to decrease the spread assumption in the model.

Market Return Assumption—Latest Forecast: The 2019 Constant Sharpe ratio based on the market return expectations and risk-free rate forecast is now 0.320, based on a total market return of 12.1%, slightly above the historic average. o User Input: If you believe market returns will be higher or lower in the future, then adjust the forecast returns in the model. The latest estimates from the revised ERP model used in the FedEx DCF-based valuation include the following: • Expected average market volatility: 14.89% (below the historic average, and current quarterly average) • Expected average risk-free rate of return: 1.99% (above the current 10-year U.S. Treasury rate of 1.88%) • Resulting stage-one ERP estimate: 5.3% • Required return on equity assuming a beta of 1.0: 7.8% •

219

APPENDIX 11: EQUITY RISK PREMIUM MODEL UPDATE Appendix 11 Overview: Each time the Federal Reserve’s FOMC meets, the ERP model must be updated to reflect the latest details related to the Federal Funds rate. Going forward these updates will be tracked in the Appendix section of this textbook, and published in future editions.

Appendix 11: Step 1—FOMC Meeting Update Coming Soon: The ERP Model is updated multiple times per quarter. Below is a list of the planned meeting updates: • September 17 & 18, 2019: FOMC meeting. Target Fed Funds rate update (Summary of Economic Projections). •

October 29 & 30, 2019: FOMC meeting. Target Fed Funds rate update.

•

December 10 & 11, 2019: FOMC meeting. Target Fed Funds rate update (Summary of Economic Projections).

Appendix 11: Step 2—Data Maintenance Coming Soon: Below is a list of the planned quarterly maintenance updates: •

October 1, 2019: Quarterly data maintenance (update equity returns, volatility, and interest rates).

•

January 1, 2020: Quarterly data maintenance (update equity returns, volatility, and interest rates).

220

APPENDIX 12: FEDEX FISCAL 1Q2020 EARNINGS RELEASE Step 1: Prepare for the Release

Step 2: Update After the Release

Appendix 11 Overview: Each time a new quarterly earnings report is released, or when significant company or market developments occur, the FedEx earnings model will need to be updated. Going forward these updates will be tracked in the Appendix section of this textbook, and published in future editions. This represents the earnings preview/review cycle, which demonstrates that a model is a living file, and must be maintained continuously.

Appendix 11: Step 1—Prepare for the F1Q2020 Release (Earnings Preview) Coming Soon: FedEx will release F1Q2020 results on September 17, 2019…stay tuned!

Appendix 11: Step 2—Update After the F1Q2020 Release (Earnings Review) Coming Soon: FedEx will release F1Q2020 results on September 17, 2019…stay tuned!

221