Economic Policy and Household Welfare During Crisis and Adjustment in Tanzania 9780814788691

Citation preview

Economic Policy and Household Welfare during Crisis and Adjustment in Tanzania

Economic Policy and Household Welfare during Crisis and Adjustment in Tanzania Alexander H . Sarris Rogier van den Brink

Published fo r Cornell Universit y Foo d an d Nutrition Polic y Progra m by New Yor k University Pres s New Yor k an d Londo n

NEW YORK UNIVERSITY PRESS New York and London Copyright © 1993 by New York University All rights reserved Library of Congress Cataloging-in-Publication Dat a Sarris, Alexander. Economic policy and household welfare during crisis and adjustmen t in Tanzania / Alexander H. Sarris, Rogier van den Brink, p. cm . The result of a collaborative research program between the Cornell University Food and Nutrition Policy Program (CFNPP) and the Economic Research Bureau (ERP) of the University of Dar es Salaam. ISBN 0-8147-7982-4 1. Tanzania—Economi c policy. 2 . Households—Tanzania . I. Brink , Rogerius Johannes Eugenius van den, 1958 - II . Cornel l University Food and Nutrition Policy Program. III . Title . HC885.S27 199 3 338.9678—dc20 93-1075 6 CIP New York University Press books are printed on acid-free paper , and their binding materials are chosen for strength and durability. Manufactured i n the United States of America 10 9 8 7 6 5 4 3 2 1

I CONTENTS

List o f Table s vii

i

List o f Figure s xi

i

Foreword xi

v

Abbreviations x

v

1 Introductio n 1 2 Backgroun d 5 Population, Natural Resources, and Agroclimatic Zones 5 Postindependence Economic Developments 1 0 General Structure of the Macroeconomy 1 7 3 Macroeconomi c Polic y an d Performanc e 2

7

Interpretation o f the Crisis and the Adjustment Debat e 2 Macroeconomic Performance unde r Adjustment 3 2 The Devaluation Debate in the Context of Adjustment 4

8 1 v

vi Contents Financing o f Publi c an d Externa l Deficit s an d Monetar y Adjustments 4 3 The Hidden Econom y i n Tanzania 4 8 Conclusions 5 6 4 Profil e o f Income s an d Povert y i n Tanzani a 5

8

Population an d Som e Househol d Characteristic s 5 Structure o f Incom e 6 4 Distribution o f Income 7 6 Income Differentiatio n 8 7 Consumption Pattern s 8 8 Poverty Lin e an d Extent o f Povert y 10 3 Conclusions 11 4

9

5 Performanc e o f Agricultur e 11

6

The Institutional Settin g 11 7 Official Source s o f Agricultura l Dat a 12 0 A First Assessment Base d o n Officia l Statistic s 12 An Alternative Scenari o 12 4 Prices 12 5 Survey Dat a 13 0 Imports 13 1 Parallel Export s 13 3 Malnutrition 13 6 Agricultural Structur e an d Technolog y 13 7 Conclusion 14 4

2

6 Trend s i n Income s an d Welfar e o f Variou s Income Group s 14 6 How Rea l i s the Decline i n Rural an d Urba n Income s durin g the "Nyerere Experiment? " 14 7 A Method fo r Analyzin g Househol d Income s 15 9 Classification o f Household s 16 4 Composition o f Agricultural Incom e 16 6 Consumption Pattern s 17 1 Prices 17 2 Other Dat a 17 6 Results 17 9

Contents vi

7 Evaluatio n and Conclusions 18

6

Appendix A — Computation of the Real Exchange Rate for Tanzania 19 3 Appendix B — Variable Shares in the Index of Real Household Incomes 19 6 References 20 Index 21

1

3

i

List of Tables

1 — Selecte d Farmin g Syste m 9 2 — Averag e Annua l Growt h Rat e o f GD P by Secto r (percentages ) 1 8 3 — Laspeyre s Volum e Indice s fo r Domesti c Export s an d Direc t Imports, 1974-198 6 (1976= 100 ) 2 1 4 — Laspeyre s Uni t Valu e Indice s fo r Domesti c Export s an d Import s by Commodit y Grou p (1976= 100 ) 2 2 5 — Summar y o f Centra l Governmen t Operation , 1976-198 7 2 4 6 — Compositio n o f Revenu e an d Expenditure , 1975/1976-1986/198 7 (as percentag e o f tota l revenu e an d percentag e o f publi c expenditure) 2 5 7 — Balanc e o f Payments , 1970-198 8 3 0 8 — Officia l Paralle l and Real Exchange Rates, 1965-198 9 (Tsh/US$ ) 3 3 9 — Recen t Macroeconomi c Development s 3 6 10 — Estimate s o f Parit y Exchang e Rates , 1979-198 4 (Tsh/US$ ) 4 3 11 — Financin g o f Public Defici t 4 4 12 — Change s i n Monetar y Aggregate s durin g th e ER P (absolute changes fro m previou s year ) 4 5 13 — Contributio n o f Lendin g fo r Agricultura l Marketin g t o Tota l Domestic Lendin g Change s (i n million Tsh ) 4 6 viii

List of Tables 14 — Financin g o f the Curren t Accoun t Defici t (i n millio n Tsh, unless noted) 4 7 15 — Estimate s o f th e Secon d Econom y b y th e Missin g Income Approach 5 2 16 — Estimate s o f Secon d Econom y GD P 1967-198 8 (in million Tshs ) 5 5 17 — Siz e of Secon d Econom y (i n millio n 197 6 Tsh) 5 6 18 — Rura l an d Urban Households , 1976-197 8 6 0 19 — Distributio n o f Rural an d Urba n Household s b y Siz e (percentages) 6 1 20 — Househol d Distributio n Accordin g t o Educational Leve l and Industr y o f Hea d o f Household , 1976/7 7 6 2 21 — Structur e o f Househol d Incom e i n Rural an d Urba n Tanzania, 1976/7 7 6 5 22 — Structur e o f Averag e Cas h Incom e pe r Household , Rural an d Urban Far m an d Nonfarm Households , 1969 an d 1976/7 7 (percentages ) 6 8 23 — Nomina l an d Real Per Capit a Cas h Income s i n 196 9 and 1976/7 7 (Tsh/capita ) 6 9 24 — Compositio n o f Rura l Househol d Income , 198 0 and 198 3 (percentages ) 6 9 25 — Annua l Cas h Income i n Private Household s b y Sourc e of Income an d Secto r o f Hea d o f Household , 1976/7 7 (Tsh pe r household ) 7 2 26 — Annua l Cas h Income b y Educationa l Leve l o f Hea d of Household , 1976/7 7 (Ts h per household) 7 4 27 — Annua l Cas h Incom e i n Private Household s b y Economi c Activity an d Secto r o f Hea d o f Household , 1976/7 7 (Tsh pe r household) 7 5 28 — Distributio n o f Income i n Tanzania Mainland , 1976/7 7 (Tsh ) 7 29 — Regiona l Incom e Distributio n Statistics , 1976/7 7 (Tsh ) 7 9 30 — Distributio n o f Pe r Capit a Income , 1976/7 7 8 0 31 — Distributio n o f Rura l an d Urba n Household s Accordin g t o Cas h Expenditures (Tsh ) 8 2 32 — Source s o f Cas h Incom e o f Household s Accordin g t o Cas h Expenditure Categories , 1976/7 7 (Tsh/household ) 8 5 33 — Expenditur e Share s o n Variou s Consumptio n Categorie s by Differen t Incom e Groups , Rural (percentages ) 8 9 34 — Expenditur e Share s o n Various Consumptio n Categorie s by Differen t Incom e Groups , Urban (percentages ) 9 1

7

X List

of Tables

35 — Annua l Consumptio n i n Private Households , 1976/77 , by Foo d Item s b y Househol d Expenditur e Group , Rura l (quantities i n kilogram/capita/year) 9 4 36 — Annua l Consumptio n i n Private Households , 1976/77 , by Foo d Item s b y Househol d Expenditur e Group , Urba n (quantities i n kilogram/capita/year) 9 5 37 — Dail y Pe r Capit a Calori e Intake o f Rural Households , 1976/77 , by Househol d Cas h Expenditur e Clas s 9 6 38 — Dail y Pe r Capit a Calori e Intake o f Urba n Households , 1976/77 , by Househol d Cas h Expenditur e Clas s 9 8 39 — Fractiona l Share s o f Househol d Tota l an d Food Consumptio n Purchased fo r Mone y b y Househol d Expenditur e Class , Rural, 1976/7 7 10 1 40 — Fractiona l Share s o f Househol d Tota l an d Foo d Consumptio n Purchased fo r Mone y b y Househol d Expenditur e Class , Urban, 1976/7 7 10 2 41 — Dail y Pe r Capit a Calori e Intake o f Variou s Food s i n Rura l Households fro m Subsistenc e an d Purchases, 1976/77 , by Expenditur e Clas s (kilocalories/capita/year ) 10 4 42 — Dail y Pe r Capit a Calori e Intak e o f Variou s Food s i n Urba n Households fro m Subsistenc e an d Purchases, 1976/77 , by Expenditur e Clas s (kilocalories/capita/day ) 10 6 43 — Econometri c Estimate s o f Calori e Expenditur e Cross-Sectio n Relations, 1976/7 7 10 9 44 — Pe r Capit a Food an d Tota l Expenditures i n Urban Area s Consistent wit h a Given Pe r Capit a Calori e Intak e in 1976/7 7 10 9 45 — Econometri c Estimate s o f Relation s Betwee n Tota l an d Foo d Expenditures 11 1 46 — Povert y Level s Estimate d Accordin g t o the Ratio o f Foo d to Total Expenditur e (Tsh/capita/annum ) 11 2 47 — Percentag e o f People an d Households i n Poverty i n 1976/7 7 Using Povert y Line s Accordin g t o the Ratio o f Foo d to Total Expenditur e 11 3 48 — Povert y Line s i n 1989 , Household Cas h Expenditur e Categor y in 1976/7 7 (Tsh/capita/annum ) 11 5 49 — Pric e Analysis , Arabica an d Robust a Coffe e 12 8 50 — Structur e o f Rural Income s i n 1982/8 3 by Pe r Capit a Income Quintile s 14 8

List of Tables x

51 — Share s o f Rura l Cas h Nonlivestoc k Income s i n 1982/83 , by Pe r Capit a Incom e Quintile s 14 9 52 — Compositio n o f Rura l Cas h an d Cas h Nonlivestoc k Incomes , 1969 to 198 3 15 1 53 — Trend s i n Per Capit a Rura l Rea l Incomes , 196 9 to 198 3 (in 1976/7 7 prices ) 15 2 54 — Ope n Marke t Produce r an d Urban Consume r Price s fo r Severa l Staples, 1986/8 7 an d 1987/8 8 (al l prices i n Tsh/kilogram) 15 3 55 — Structur e an d Levels o f Urba n Incomes , 1969 , 1976/77 , and 198 4 15 5 56 — Pe r Capit a Househol d Income s i n 198 4 as Percentage o f Thei r Values i n 1977 : Da r e s Salaam , by Quintile s (compariso n is between income s a t 1982/8 3 prices) 15 6 57 — Structur e o f Rural an d Urban Househol d Incom e and Consumptio n i n 198 6 15 7 58 — Source s o f Incom e o f Representativ e Poor , Middle , and Rich Rura l an d Urban Households , 1976/7 7 16 7 59 — Share s o f Cas h Cro p Incom e fro m Foo d an d Cas h (Export ) Crops 16 8 60 — Structur e o f Per Capit a Agricultura l Incom e o f Representativ e Poor , Middle, an d Rich Rural an d Urban Households , 1976/197 7 170 61 — Expenditur e Share s o f Representativ e Household s (percentages ) 17 62 — Term s o f Trad e (Ratios ) betwee n Severa l Pric e Indice s 17 4 63 — Ratio s o f Ope n Marke t Urba n Price s t o Officia l Consume r and Producer Price s 17 5 64 — Deman d an d Suppl y Parameter s fo r th e Si x Representativ e Households 17 8 65 — Value s o f Othe r Mode l Parameters, Circ a 1976-7 7 17 9 66 — Impac t o f Changin g Price s o n Househol d Rea l Income s 18 0 67 — Evolutio n o f Severa l Ke y Pric e Indices an d Wag e Employment , 1975 to 198 9 18 1 68 — Evolutio n o f Aggregat e an d Househol d Share s Parameters , 1976 to 1989 , under th e Influenc e o f Price an d Forma l Employment Trend s (percentages ) 18 2 69 — Evolutio n o f Househol d Rea l Income s unde r Alternativ e Parameters Assumption s i n the Variable-Shar e Cas e 18 3 70 — Impac t o n Househol d Welfar e Whe n Al l Prices, Includin g Those fo r Expor t Crops , Are Ope n Marke t Price s 18 4

i

1

List of Figures

1 — Interna l Term s o f Trad e 3 8 2 — Rea l Pe r Capit a Currenc y an d M 2 Holdings 4 0 3 — Yearl y Change s i n NCP I and th e Parallel Exchang e Rat e 4 1 4 — Pe r Capit a Production o f Foo d Crop s an d Purchase s of Expor t Crop s 12 2 5 — Expor t Crops , Price an d Quantit y Index , 1965-198 9 12 7 6 — Foo d Prices , 1969-1989 , Official an d Parallel Rea l Price s 12 7 — Calori c Compariso n betwee n Officia l Serie s an d Surve y Estimates, Major Cereals , Cassava, an d Bean s 13 2 8 — Ne t Cerea l Imports , 1966-199 0 13 3 9 — Rea l Coffe e Prices , 1970-198 8 (NCPI) ; Paralle l Marke t and Officia l Produce r Pric e 13 5 10 — Change s i n Farm Siz e Distribution, 1971/72-1986/8 7 13 8 11 — Change s i n Farm Size , 1971/72-1986/8 7 13 8 12 — Cro p Land pe r Holding , 1971/72-1986/8 7 13 9 13 — Change s i n On-Far m Labo r Availability pe r Holding , 1971/72-1986/87 13 9

XII

9

List of Figures XJi

14 — Change s i n On-Far m Labo r Availabilit y pe r Hectare , 1971/72-1986/87 14 0 15 — Change s i n Cultivate d Are a unde r Foo d Crops , 1971/72-1986/87 14 0 16 — Change s i n Pure o r Mixtures o f Permanen t Crops , 1971/72-1986/87 14 2 17 — Change s i n Area unde r Cassava , 1971/72-1986/8 7 14

2

i

Abbreviations

AGSASU Agricultura l Sampl e Surve y o f Tanzani a Mainlan d 1986/8 7 Bis Basi c Industrie s Strateg y BNI Basi c Need s Incom e ERP Economi c Recover y Progra m ESAP Economi c an d Socia l Actio n Progra m EWCMB Earl y Warnin g an d Cro p Monitorin g Burea u HBS Househol d Budge t Surve y MDB Marketin g Developmen t Burea u NAPB Nationa l Agricultura l Product s Boar d NESP Nationa l Economi c Surviva l Progra m SAP Structura l Adjustmen t Progra m TAG Technica l Advisor y Grou p

xiv

I FOREWORD

l A # o r k o n thi s book starte d i n earl y 199 0 with th e initiation o f th e collabora tive researc h progra m concernin g th e impac t o f adjustmen t policie s i n households i n Tanzania between Cornel l Universit y Foo d an d Nutrition Polic y Program (CFNPP ) and the Economic Researc h Bureau (ERP ) of the University o f Dar e s Salaam , an d supporte d b y th e Afric a Burea u o f th e U.S . Agenc y fo r International Developmen t (USAID) . Th e originall y envisione d outpu t wa s t o include mostly descriptive and background informatio n o n the Tanzanian econ omy a s a prerequisit e t o subsequen t surve y an d mode l buildin g work . Given , however, th e massiv e earl y literatur e o n Tanzania , th e lack o f recen t analyses , and th e considerabl e controversie s concernin g bot h th e functionin g o f th e economy a s well a s the need fo r an d nature o f adjustment , ou r interest quickl y turned mor e analytical . Th e attemp t wa s therefor e mad e no t onl y t o giv e a background o n th e Tanzania n econom y bu t als o t o analyticall y describ e bot h the structur e o f household s a s wel l a s the functioning o f th e true, as contraste d to the observed , economy . The bulk of the work was completed in early 199 1 and revised in late summe r of 1991 . It was subsequently revised in early 1992 , taking into account comment s and suggestion s fro m severa l colleague s an d external reviewers .

xv

XVI" Foreword

Among th e authors , A . Sarri s i s responsibl e fo r Chapter s 3 , 4 , an d 6 , an d Rogier van den Brin k fo r Chapte r 5 . The other chapter s wer e written jointly . Several peopl e hav e contribute d a t variou s stage s o f th e work . Th e author s would lik e t o singl e ou t ou r Tanzania n collaborators , M.S.D . Bagachwa , A . Mbelle, R. Mabele, and A. Tibaijuka fo r variou s comment s an d suggestions ; K . Budwar, E . Lugusha , L . Merid , A . Naho , E . Stephenson , an d S . Zografakis fo r research assistance during the various stages of the research; and R. Christiansen, C. del Ninno, P. Fleuret, P. Pinstrup-Andersen, and two other anonymous reviewers fo r helpfu l suggestion s fo r th e revision. Davi d Sah n wa s th e directo r o f th e project unde r which this activity wa s a part, and is acknowledged fo r his overall support. The document wa s word processed by Elizabeth Vakalopoulou , typese t by Gaudenci o Dizon , an d produced b y Ne w Yor k University Press . Finally, w e would lik e to thank ou r wives , Maria an d Natasha, respectively , for their patience and support during the long period it has taken to complete this book.

ALEXANDER H . SARRIS ROGIER VAN DEN BRINK

I 1 INTRODUCTION

TPhe economi c developmen t o f Tanzania , fro m independenc e i n 196 1 unti l now, has been characterized by a series of internal and external shock s that have teste d th e resilienc e o f th e economy , th e stabilit y o f it s institutions , an d the tolerance an d inventiveness o f it s people. Despit e th e fact that Tanzania i s one of th e world's poorest countries (th e fourth-poorest, accordin g to the 199 0 World Bank Development Report) , it has managed to weather all storms with a remarkable degree of political stability , an d without extreme hardships such as the famines tha t hit other, more-developed countries . The beginning of the decade of the eighties found the Tanzanian economy i n deep economic crisis, resulting from a sequence of policy responses and a series of externa l shock s that , accordin g t o man y observers , worsene d a n alread y critical situation. Recognizing th e impossibility o f dealing with the crisis alone, the government sough t external assistance from multilateral donors, such as the International Monetar y Fun d (IMF). The conditions impose d by the IMF were at first strongl y resiste d o n th e ground s tha t the y woul d und o al l th e socia l an d distribution gain s tha t ha d bee n achieve d b y Tanzani a sinc e independence . However, attempt s b y th e governmen t t o implemen t it s ow n stabilizatio n an d adjustment programs largely failed , with the result that in 198 6 the governmen t of Tanzania agreed to a three-year package of measures and policy reforms as a 1

2 Alexander

H. Sarris and Rogier van den Brink

condition fo r multilatera l an d othe r dono r assistance . Th e initia l progra m ha s been succeede d b y anothe r progra m tha t starte d i n 198 9 an d i s stil l bein g implemented. There are deep controversies surrounding the acceptance and implementation of externally or internally imposed policy reforms in Tanzania, which date from the en d o f th e seventies . The y concer n th e necessit y o f reforms , th e typ e o f policies adopted, the effectiveness o f various measures in reviving the economy, and the impac t o f reform s o n various segment s o f th e population. Th e purpos e of th e presen t repor t is , first , t o outlin e th e mai n structura l feature s o f th e Tanzanian economy, especiall y thos e relevant t o the crisis, and second, to deal with the last of the above-mentioned aspects of the reform debate, especially the impact of previous policies and current stabilization and adjustment measures on the poorer segments o f the Tanzanian population. Underlyin g al l policy refor m programs tha t ar e recommende d a s remedie s t o a n economi c crisis—whic h i s usually manifeste d i n a persistent , larg e balance-of-payment s deficit—i s th e belief tha t the major reasons for the crisis lie with "wrong" policies previousl y adopted, whose modification wil l correc t the fundamental structura l disequilibria. Whethe r thi s wil l happe n depend s o n th e structur e an d respons e o f th e economy t o variou s signals , o n th e degre e o f implementatio n o f th e state d reforms, and on the effectiveness o f the new policies in changing the established structures o f politica l an d economi c institutions . Muc h o f th e debat e o n th e effectiveness an d impac t o f adjustmen t programs—an d Tanzani a i s n o excep tion—has bee n drive n b y ideologica l biase s rathe r than detaile d analysis . Ou r effort i n this report will be to contribute to the understanding of the functionin g of th e Tanzania n econom y an d it s peopl e durin g th e crisi s an d unde r polic y reforms, without invoking ideologica l arguments . Our emphasis throughout will be on economic structur e and behavior. Give n the short period over which reforms have been in place in Tanzania, it is difficul t to evaluat e impacts , unless on e understand s how variou s parts of th e econom y are interrelate d an d function . I t i s i n th e sam e manne r tha t w e wil l approac h understanding th e impac t o n households . Anothe r reaso n fo r ou r emphasi s o n economic an d institutional structur e concern s th e difficulty o f isolatin g effect s of polic y changes . Ou r effort wil l b e to isolat e externa l an d internal cause s o f various economic changes, so as to pinpoint problems as well as positive aspects. While th e issu e o f th e impact o f adjustmen t programs on poorer sections o f the population, i n Africa an d elsewhere, ha s received muc h political attention , especially afte r UNlCEF' s Adjustment with a Human Face (Cornia , Jolly , an d Stewart 1987) , there have been ver y few empirica l studie s examinin g th e issu e analytically. Recently , Sah n and Sarris (1991) , i n a comparative stud y o f rura l smallholders i n fiv e Africa n countries , includin g Tanzania , showe d tha t th e

Introduction 3

economic signal s afte r the onset o f adjustmen t program s do not appear to hav e led to a deterioration of real rural smallholder incomes. In the case of Tanzania, the only detaile d study that has presented an examination of income trends over a period of almost two decades is the one by Bevan and his co-authors (1988), which is also reproduced in Bevan et al. (1989). That study show s tha t durin g th e seventie s an d earl y eighties , rura l an d urba n rea l incomes decline d significantly . A recen t stud y b y Collie r an d Gunning (1989 ) showed tha t th e adjustmen t effort s sinc e 198 4 hav e stoppe d th e continuou s earlier incom e decline , bu t a s o f 198 9 no significan t recover y i n income s i s apparent. These conclusions seem to form the basis for current, informed opinion in Tanzania. Our analysis i n this report shows, to the contrary, that the earlier conclusio n that serious income declines occurred throughout the seventies and early eighties must be questioned, and especially so as far as the rural, and even the urban, poor are concerned. There were indeed some real income declines, but these, it seems, were mostly concentrated among the urban middle class and not among the poor. Our analysis shows that during the period of adjustment, the rural and urban poor were hardl y affected . Th e declin e i n urban , middle-clas s rea l income s wa s stopped but not reversed. The only groups that seem to have suffered real income decline durin g the reform period are the rural middle-income an d richer households, an d the urba n rich. Thes e preliminar y result s (whic h nee d t o b e furthe r substantiated by detailed empirical research involving th e use of counterfactua l models), have severa l implication s fo r economic polic y i n Tanzania, whic h are explored i n this report. To arrive at our result, we employ a series of analyses that emphasize not only the structure and economic institutions of the economy, but also the observability of th e economy i n Tanzania. In fact, on e o f th e major conclusions tha t emerge from the analysis is that official economi c statistic s have failed, at an increasing rate sinc e th e mid-1970s , t o giv e a n even moderatel y accurat e pictur e o f eco nomic developments. I n such a setting, turnin g points in the economy migh t be misjudged, and one has to rely on a very different se t of dat a for analysis. Our investigation start s i n Chapter 2 with a quick overvie w o f th e structur e of th e economy an d economic development s sinc e independence , emphasizin g the recent years. The years of crisis and concurrent economic events and politics are outlined as a background for the rest of the report. In Chapte r 3 , w e examin e macroeconomi c development s an d policie s an d present an analysis of the size of the "second economy," the largely unobserve d and uncontrolled, but apparently very substantial, part of the Tanzanian economy that we feel is the key to understanding the evolution of incomes. After an initial, descriptive revie w o f macro developments an d the reform program, the chapter

4 Alexander

H. Sarris and Rogier van den Brink

estimates th e siz e o f th e secon d econom y an d show s tha t th e evolutio n o f th e total econom y i s differen t tha n wha t i s apparen t fro m th e officia l statistics . I n that chapter, we also analyze the various means of financing public expenditures, and w e sho w tha t gradually , foreig n source s ende d u p financin g mos t o f th e domestic deficits . In Chapter 4, we present a detailed structural profile of households in Tanzania, based on analysis of several household surveys. In that chapter, we also estimate poverty lines and the proportion of rural and urban poor in Tanzania. The results suggest a degree of poverty much larger than previously thought, but also a very equitable pattern of incomes. Tanzania, although it has some substantial incom e inequalities, appear s t o hav e bee n spare d th e enormou s maldistributio n o f in comes observed in several other developing countries . In Chapte r 5 , w e presen t a n analysi s o f development s i n th e agricultura l sector, whic h i s b y fa r th e mos t importan t economi c secto r i n Tanzania . W e discover that the trends in official productio n statistic s mus t be seriously ques tioned and that more recent information seems to suggest that the official report s of growth in food production over the last decade are not warranted. Analysis of relative price trends suggests that the official, relative prices of food versus those of expor t product s hav e not , o n aggregate , change d ver y muc h ove r th e las t decade, while real official price s declined throughout the crisis, only to recover after the onset of adjustment reforms. Real prices in the open market, however, seem to be telling a different story ; prices at least for food products, seem to have increased throughout the crisis, only t o fall afte r the onset of reforms. In Chapter 6, w e presen t ou r analysis o f househol d incomes . W e first sho w that th e allege d declin e i n househol d rea l incom e betwee n th e mid-1970 s an d mid-1980s i s no t supporte d b y existin g information . The n w e develo p a ne w methodology for analysis of household incomes based on an explicit recognition of th e informa l sector . W e deriv e incom e an d consumption pattern s fo r typical members of six household income classes (poor, middle, and rich households i n both rural and urban areas), and we show that the evolution of economic signal s since th e mid-1970 s warrant s th e conclusion s alread y mentioned . I n th e fina l chapter, w e summariz e an d synthesiz e ou r conclusion s an d deriv e polic y implications.

I 2 BACKGROUND

T h e purpos e of this chapter is to describe the background against which recent developments mus t be examined. Th e first sectio n give s a brief descriptio n of the resource characteristics of Tanzania. The next section reviews briefly th e postindependence economic developments culminating in the recent adjustment efforts. The third section outlines the macroeconomic structure, while the fourth section briefly review s the agricultural an d industrial sectors .

POPULATION, NATURAL RESOURCES, AND AGROCLIMATIC ZONES A low population density and a varied agricultural resource base give Tanzania a particularly robust agricultural sector. Its mineral resources include coal, iron ore deposits, phosphates, and smaller deposits o f copper , lead, tin, nickel, an d sulphur. Sod a comes fro m Lak e Natron, an d salt ca n be obtaine d fro m certai n springs and the sea. Gold and diamonds ar e present in significant quantities . Tanzania has access to the Great Lakes region of East Africa (see Map 1) . In the north, Lake Victoria border s both Uganda an d Kenya. Lak e Tanganyika i n the west is the border with Zaire, and Lake Nyasa in the southwest borders with Malawi. Thes e lakes , togethe r with Tanzania's marin e resource s i n the coasta l areas, constitute major, but relatively undeveloped, fishery resources. Addition5

MAPI Map of Tanzania

Source: World Bank (1991).

Background 7

ally, Tanzania' s rivers hav e significan t potentia l fo r hydroelectri c powe r an d irrigation. Tanzania is a country of low population density. In 1988,22.5 millio n people inhabited 881,28 9 squar e kilometers , whic h implie s a densit y o f abou t 25. 5 persons per square kilometer. Relativel y fe w town s exist, the major ones bein g Dar es Salaam, with approximately 1. 4 million people, and Mwanza, with about 180,000 people . Onl y 5 percen t o f th e tota l lan d are a i s cultivated , althoug h nearly 9 0 percent o f th e country coul d a t least theoreticall y suppor t som e typ e of farming (Lan d Resources Developmen t Centr e 1987) . Large areas, however, are tsets e fl y infested , whic h preclude s thei r settlemen t b y eithe r peopl e o r animals. The incidence of the tsetse fly produce s a typical pattern of population settlement characterize d by pockets of densely populated , tsetse-free areas . According t o thre e demographi c censuse s (take n i n 1967 , 1978 , an d 1988 , respectively) the population growth rate declined significantly, fro m 3.3 percent for the 1967-1978 period, to 2.8 percent during the 1978-1988 period. This trend has been accompanied by a slowing of the growth of the urban population, which grew at an average annual rate of 10. 7 percent during 1967-197 8 but at only 5. 4 percent for the 1978-198 8 period . Tanzania has been described as the country wit h the most varied ecology o f any in Africa, enabling a wide variety of agricultural activities to be undertaken. These rang e fro m highlan d te a an d coffe e productio n t o arid-are a nomadi c pastoralism. Mos t o f Tanzania' s semiari d centra l zon e i s suite d fo r extensiv e farming systems and livestock production, due to the high rainfall variability and fragile soils . Small-scale , rain-fe d cultivatio n o f cereal s (sorghu m an d millet), tubers, and cotton is undertaken by farming communitie s that are often area s of locally high population densities due to the presence of the tsetse fly elsewhere . Livestock i s produce d b y th e nomadi c Masa i an d a numbe r o f seminomadi c groups who are responsible fo r most of the country's production. A tropica l fores t cove r i s foun d alon g th e coas t an d o n Zanzibar . Her e coconuts, cashew nuts, rubber, cocoa, cloves, sisal, various spices, and fruits are grown. Nex t t o maiz e an d rice, cassav a i s a n importan t foo d crop . Th e mai n growing seaso n i s fro m Marc h to July. Th e northern coastal are a has a second, but short, rainy seaso n in November-December. Soil s ar e mainly sandy , excep t for the clayish bottomlands i n the river valleys. The western part of the country is dominated by a plateau about 1,00 0 meters high, which has an average rainfall of over 750 millimeters per year. The growing season i s betwee n Novembe r an d April . I t i s her e tha t th e country' s mai n potential fo r surplu s foo d cro p productio n ca n b e realized . Maize , sorghum , millet, cassava , rice, cotton , an d tobacc o ar e th e mai n crop s grown . I n som e regions, the area under cultivation coul d be readil y expanded . Oxe n traction i s

8 Alexander

H. Sarris and Rogier van den Brink

widely practice d i n th e norther n par t o f th e region , wher e cotto n an d ric e production dominate the cropping pattern. The region also includes larg e area s of forest reserves, and its lakes constitute important fishery resources. The highland areas possess some agriculturally attractiv e features—high and generally reliable rainfall over a five-to-nine-month period, fertile volcanic soils, and a relativel y coo l climate . Th e soi l type s foun d i n th e highland s reac t relatively wel l to fertilizer application and permit intensive agricultural production systems . Fertilize r us e i s therefor e widespread . Wheat , maize , cassava , sorghum, coffee, tea , potatoes, bananas, beans, and pyrethrum ar e some o f th e main crops grown in the highlands. Scope for the expansion of the cultivated area exists i n th e souther n highland s only . Th e norther n an d wester n highland s ar e under considerabl e lan d pressure . Th e mai n farmin g system s ar e th e banana coffee intercro p syste m i n th e norther n highland s an d the pure-croppe d maiz e system in the south. The diversit y o f farmin g system s i s illustrate d i n Table 1 , which present s a selected number of zones an d some main farm characteristics. From this tabl e i t becomes clea r that it is no t possible t o spea k o f a single , average Tanzanian farming system. Average farm size, for instance, varies from 0.78 hectar e i n Kager a t o 4.3 5 hectare s i n Singida . Croppin g pattern s sho w equally marked differences acros s regions. First, the share of farm area devoted to expor t crop s range s fro m 8 0 percen t i n Kagera an d 76 percen t i n th e Coas t regions to virtually nil in areas such as Iringa, Singida, Kondoa, and Mara. Note, however, tha t perennial expor t crops always allo w fo r intercropping with foo d crops—coffee i s nearly alway s intercropped with bananas, and cashew nut s are often intercroppe d with cassava . Th e share s devote d t o annua l expor t crops , which ar e mor e difficul t t o intercro p wit h foo d crops , ar e i n genera l muc h lower—cotton i s cultivate d o n 2 5 percen t o f tota l far m are a i n Mwanza , 1 3 percent in Shinyanga, and 6 percent in Morogoro. Food crops cultivate d cente r around maize mixed-croppin g system s i n most of th e areas. Rice i s particularly importan t i n the Coast, Morogoro, Tanga, and Mwanza areas . Th e mor e drough t resistan t crop s (sorghu m an d millet ) ar e typically found in Morogoro, Shinyanga, Iringa, Ruvuma, Singida, Kondoa, and Mwanza. The importanc e o f cattl e a s a sourc e o f far m incom e fo r th e mixe d far m enterprise als o varie s quit e prominently b y region. Tabora, Iringa, an d Singid a record a n averag e o f mor e tha n fiv e hea d o f cattl e pe r farm . I n the Coas t an d Morogoro areas , no larg e livestoc k i s reported . Man y o f th e othe r area s hav e relatively lo w numbers of cattl e kept on the farm. Off-farm income , as reported by thi s survey , varie s fro m 12-1 3 percen t o f tota l incom e i n Kilimanjaro , Ruvuma, an d Singida, t o 45-56 percen t i n Lushoto and Morogoro. Subsequen t

Food Crops

1.00 3.65 2.32

Maize/beans/vegetables (0.29 ) Coffee/banana (0.56 ) Cashew/coconut (0.76 ) Rice/maize (0.15) ; cassava (0.06 ) Maize/beans/vegetables (0.52) ; "maize/millet (0.36) 0.78 Coffee/banana (0.80 ) Roots/pulses (0.17 ) 1.18 Coffee/banana (0.66 ) Maize/beans (0.31 ) 2.88 Maize/millet (0.29) ; maize/sorghum/ millet (0.20); maize (0.16) 1.85 Cereals/pulses (0.51) ; sweet potatoes (0.43 ) 1.68 Coffee/banana (0.18 ) Maize (0.40 ) 0.99 Cotton (0.06 ) Maize (0.39); rice (0.22); sorghum 0.14 ) 1.38 Cashew nuts (0.29) Cereals/pulses (0.31) ; pulses/roots (0.23) ; cereals (0.17 ) 3.24 Tobacco (0.05 ) Maize/beans/cassava (0.24) ; maize (0.23); maize/millet (0.20 ) 2.66 Cotton (0.13 ) Maize (0.37); maize/sorghum (0.14 ) 4.35 Millet/maize/sorghum (0.73) ; maize/millet/groundnuts/(0.08); pulses (0.17 ) 9 Tobacc o (0.06) Maize/groundnuts/cassav a (0.50) ; 6. maize (0.14 ) 4 Cashews/coconut / Maize , rice, maize/cassava (0.50 ) n.a cassava (0.46 )

Export Crops

Share of Farm Area

Source: FAO/IBR D Cooperative Research Project (1975) . Notes: n.a . denote s dat a not available; .. . denote s less than 0.05.

Tanga 1.2

Tabora 1.2

Shinyanga Singida

Ruvuma

Mara Mbeya Morogoro Mtwara

Kagera Kilimanjaro Kondoa

Arusha Coast Iringa

Region

Average Farm Size (Hectares)

TABLE 1 Tanzania: Selected Farming System .

…

. n.a

2 0.2

1.2 5.7

1.3

n.a

2.9 0.7

2.0 1.5 n.a.

n.a. ... 7.1

.

9

0.28 0.13

0.12

n.a. 0.22 0.45 n.a.

n.a. 0.12 n.a.

n.a 0.20 0.27

Cattle Off-Farm (Head Income (Share per Farm) of Total)

10 Alexander

H. Sarris and Rogier van den Brink

surveys (Collie r e t al . 1986 ) attes t t o thi s wide-rangin g diversit y (Tanzania , Agricultural Sampl e Survey 1989) . POSTINDEPENDENCE ECONOMIC DEVELOPMENTS At independenc e i n 1961 , Tanzani a inherite d a n economi c syste m base d o n private enterprise unde r some government regulation . Th e country wa s mainl y agricultural—producing foo d fo r domestic consumption , and sisal, coffee, cot ton, an d a fe w other , mino r crop s fo r export . Althoug h sisa l productio n wa s concentrated on European-owned plantations, most of the agricultural products were produce d b y Africa n farms . Industr y wa s undeveloped , wit h cotto n gin ning the major activity. The government guaranteed prices on major marketable crops, and parastatals were involved in export-crop marketing. In the first three years after independence, no major changes were made in the system, and policy was guided by a three-year plan that had been developed before independence . The first five-year plan, 1964-1969, adopted in 1964, did not change the basic structure of the free-enterprise economy, although the state was supposed to have a mor e direc t rol e i n development . Agricultur e wa s t o b e th e leadin g sector , evolving by "improvement"—namely, b y promoting government-controlled ex tension, marketing, and credit—and by "transformation"—trying ou t some pilot, large-scale villag e settlemen t scheme s designe d t o promot e modern , capital intensive cultivatio n methods. Th e interna l deman d generate d b y agricultura l growth, i t wa s hoped , woul d facilitat e expansio n o f import-substitutin g manu facturing, and some import restrictions were imposed. The plan also emphasized Africanization, b y promoting investment s i n education, an d economic indepen dence, by cutting th e rigid link of the money suppl y to the sterling balance and establishing th e Central Bank (1966) . The country's economic performance i n these early, postindependence year s was quit e adequate , wit h a n averag e annua l Gros s Domesti c Produc t (GDP ) growth of more than 6 percent and agricultural growth of about 4 percent. Exports grew rapidly, at about 8 percent annually, and manufacturing also expanded very fast. N o majo r institutiona l change s occurre d unti l 1967 , an d the pre-indepen dence, farmer-controlle d cooperativ e movemen t i n agricultur e continued t o evolve (Coulso n 1982) . The period 1967-1973 produced drastic change in economic and social policy in Tanzania. The Arusha declaration on socialism and self-reliance, i n February 1967, establishe d a s nationa l goal s th e creatio n o f a socialis t stat e an d self reliance i n nationa l development . Socialis m wa s viewe d a s necessar y fo r th e achievement o f economic justice and equality. It was deemed important that the state b e responsibl e fo r meetin g basi c need s suc h a s health , education , an d

Background 1

1

nutrition. T o promot e thes e goals , responsibilit y fo r contro l an d expansio n o f economic an d social asset s and activities wa s to be transferred t o the state. Following the declaration, all major firms in production, marketing, distribution, and finance wer e nationalized. A large number of parastatals were created. Traditional agricultural cooperatives, which had been encouraged since independence, were given rights to enable them to replace private traders. However, the cooperatives were drawn under state control, and in 1975 , they were replaced by crop-marketing parastatals. In 1967 , the State Trading Corporation was created to monopoliz e internationa l commerce , an d late r i n th e 1970s , privat e villag e shops were closed, to be replaced by communal ones. In rura l areas , ujamaa (familyhood ) village s wer e promote d a s th e basi c economic an d socia l unit . Th e ide a o f thes e village s wa s tha t scattere d rura l households woul d b e brough t close r together , s o tha t socia l service s coul d b e made available , communa l activitie s coul d b e promoted , an d farmer s woul d become mor e socially , politically , an d economicall y active . Thes e idea s als o motivated wide r nationa l policy . I t was , however , th e slo w pac e o f voluntar y villagization, tha t b y 197 3 le d t o compulsor y villagization , an d ove r th e nex t three years large numbers of rural households were forced to relocate in villages. Those villages were not just the physical locations of households. They acquired legal status , a s incorporate d entities , an d wer e governe d b y electe d council s along wit h government-appointe d civi l servants . A s lega l entitie s the y coul d manage local trading store s and communal farms. During the period 1967-1973 , decentralization o f public administration took place. Responsibility fo r planning an d running public service s wa s given to the regions, an d civi l servant s wer e transferre d fro m centra l headquarter s t o th e regions. Export crops were encouraged through credit facilities and projects, and several stat e farms were established. Economic performanc e slowe d durin g thi s period . Accordin g t o th e bes t available figures, GDP growth slowed to about 3 percent annually, less than half its pre-1967 rate . Manufacturin g gre w a t 7.6 percen t annually , compare d wit h 13.2 percent between 196 4 and 1967 . Agricultur e gre w onl y i n line with population growth, at 2.6 percent, for a stagnant per capita output (Stewart 1986) , but it continue d t o provid e th e subsistenc e need s o f th e country . Investmen t ros e quite fast, reaching about 20 percent of GDP by 1973 . Public investment , mainly by parastatals, grew to 66 percent of total gross fixed capital formation by 1973 . The volume of export s gre w by a n average o f onl y 1. 1 percen t annually , whil e imports increased substantially , a t 7.1 percen t annually . Th e result was that by 1970 a previously balance d external trad e account showed , for the first time , a significant deficit . B y 1973 , the external trade deficit ha d reached 35 percent of total export earnings, or almost 7 percent of GDP. However, the external defici t

12 Alexander

H. Sarris and Rogier van den Brink

was easil y finance d b y long-ter m capita l inflows , an d th e overal l curren t an d capital accoun t balances remained positive. Durin g the period, public spendin g grew by about 1 6 percent annually, an d revenue grew similarly a t 15. 2 percent, both a t rate s highe r tha n tha t o f nomina l GDP , whic h gre w b y 10. 7 percen t annually. Th e result was tha t by 197 3 public expenditur e (recurren t an d development) accounted for 23 percent of GDP, compared with 1 8 percent in 1968 . Of total publi c expenditure s durin g th e period , abou t 3 0 percent , o n average , wa s spent fo r basi c need s (education , health , housing , wate r an d electricity , an d social security) . During the period the public secto r experienced smal l deficits , averaging abou t 4—5 percent of GDP, which were financed initiall y b y domesti c borrowing. However , b y 197 3 foreig n borrowin g mad e u p mor e tha n hal f o f the financing of the government deficit. Inflation during this period, as measured by th e Nationa l Consume r Pric e Inde x (NCPl ) average d a meage r 6. 6 percen t annually. The yea r 197 4 marke d th e firs t o f a serie s o f crise s tha t wer e t o shoc k th e Tanzanian econom y throughou t th e res t o f th e decade . Yea r 1973/7 4 brough t drought and a major drop in domestic food supplies, necessitating cereal imports. The first oil shock, in 1973, with the quadrupling of world oil prices, necessitated sharp increases in import expenditures. The government responded by trying to curtail consume r goo d import s an d drawing o n th e first o f th e two tranche s o f the IMF quota. It also sharply increased its domestic borrowing. Net claims of the domestic bankin g syste m o n governmen t increase d b y 13 7 percen t i n 1974 , compared with an average annua l growth of 2 9 percent i n the three years prior to 1974 . A s a consequence , th e mone y suppl y increase d b y 2 2 percent , an d inflation, a s measure d b y th e NCPl , jumped t o 19. 5 percen t i n 1974 , compare d with an annual average of 6.5 percent for the four years before that. Import expenditures in 1974, despite the government's pledge to reduce them, increased b y 53. 5 percen t over 1973 , whil e expor t earning s increase d b y onl y 12.6 percent . Th e curren t accoun t defici t almos t tripled , fro m 75 5 millio n Tanzanian shilling s (Tsh ) in 197 3 to 2,037 millio n Tsh in 1974 . Cereal import s accounted for 22.1 percen t of total imports in 1974 . The government' s respons e t o thi s firs t crisi s wa s t o adop t a stabilizatio n policy packag e tha t include d th e followin g measure s (Weave r an d Anderso n 1981): 1. Majo r increases in producer food crop prices 2. A 4 0 percen t increas e i n the minimum wage , an d progressively smalle r wage increases for higher-echelon civi l servant s 3. Shar p increases in the retail prices of basic food s 4. Pric e increases i n the main consumer products to cover costs

Background 1 3 5. Credi t restrictions 6. Impor t allocation to essential good s 7. Maintenanc e of public service levels an d expansion o f education, extension, and rural health programs 8. Moderat e curtailment of public infrastructure project s 9. Mobilizatio n of external finance t o cover external deficit s 10. Ta x increase s Good weather returned in the 1974/7 5 year, but cereal imports continued at a high leve l i n 197 5 becaus e o f a continued shortfal l i n domestic markete d sup plies. Expor t earning s wer e slightl y lowe r i n 1975 , bu t impor t expenditure s increased by 5 percent. The government increased long-term external borrowing and also received help from the IMF oil facility and the IMF compensatory financ e facility, bot h relatively lo w conditionalit y loans . Nevertheless, public expendi tures sharply increased, by 46 percent, althoug h revenues increase d by onl y 3 1 percent. Financin g fo r th e sharpl y increase d domesti c defici t (1,862. 5 millio n Tsh i n 197 5 vs . 85 1 millio n Ts h i n 1974 ) cam e fro m a 99 percen t increas e i n domestic borrowin g an d a 16 9 percent increas e i n foreign borrowing . Inflatio n continued a t a 26 percen t pac e i n 1975 , an d i n Octobe r 197 5 th e governmen t devalued the shilling by 1 1 percent. During 197 4 and 1975, besides the aid from the IMF , th e governmen t receive d progra m loan s fro m th e Worl d Ban k an d external ai d fro m a variet y o f othe r donor s wh o ha d bee n supportiv e o f th e government durin g th e firs t decad e o f independence . I n th e 1970 s th e majo r bilateral donor s wer e th e Scandinavia n countrie s (Sweden , Norway , Denmark , and Finland), the Federal Republic of Germany , and the Netherlands. In th e year s 197 6 an d 1977 , th e externa l constrain t ease d significantly , a s world price increases for coffee an d tea led to sharp increases in export earnings, and the weather was normal in Tanzania. Thos e factors, couple d wit h producer food pric e increases , brough t greate r supplies o f domesti c cereal s ont o officia l markets an d reduced th e nee d fo r cerea l imports . Import expenditures i n 197 6 came down by 22 percent, while export earnings increased by 49 percent, leading to an almost-balanced trade account. The favorable export situation continued in 1977, but imports agai n sharpl y increased . Becaus e o f th e continue d inflo w o f long-term capital , however, the external account was quite healthy, and in fact, by the end of 1977 , the government ha d accumulated large foreign reserves . Major institutional policy changes during the period included the abolition of village primar y cooperatives , whic h unti l tha t tim e ha d bee n responsibl e fo r collecting an d delivering food and export crops to the crop parastatals, and their replacement b y parastata l cro p authorities . Anothe r majo r initiativ e wa s th e implementation o f th e Basi c Industrie s Strateg y (Bis) , whic h sough t t o foste r

14 Alexander

H. Sarris and Rogier van den Brink

import substitutio n an d self-reliance b y building capacit y i n several key indus tries, such as steel. In 1977, the breakup of the East African Community (Kenya , Tanzania, an d Zambia) adde d furthe r impetu s t o a policy o f self-relianc e tha t necessitated heav y investmen t i n infrastructure , suc h a s transportatio n an d communications. In 1978 , at the recommendation o f th e IMF and the World Bank th e govern ment use d muc h o f it s accumulate d foreig n exchang e reserve s t o liberaliz e imports, which indeed soared by 43 percent from their already high 197 7 level . Exports, however, fell by 1 8 percent, as the coffee crisi s eased and world prices declined. Th e result wa s a sharply increase d trad e deficit tha t depleted foreig n exchange reserves , and external account s that could not be balanced by capita l inflows. Man y o f th e officia l loan s wer e transforme d int o grant s i n 1978 . Th e situation became much worse in 1979. Due to the war with Uganda, which started at the en d o f 197 8 an d lasted fo r mos t o f 1979 , th e governmen t ha d to impor t substantial amount s of arm s and war-related supplie s that are estimated to have cost abou t US $ 30 0 million , equivalen t t o mor e tha n half o f Tanzania' s 197 7 export earnings. In early 1979 , the government again resorted to the IMF and drew on its firs t tranche for balance-of-payment support . It also appealed for external aid, which was no t easily forthcomin g thi s time i n the aftermat h o f th e war. It was a t this stage tha t mor e credi t wa s sough t fro m th e IMF , and fo r th e firs t time , majo r internal controversie s aros e regardin g specifi c polic y reform s tha t th e IM F requested (Bierman n and Campbell 1989) . The opposition stemme d from anxiet y about th e impac t o f devaluation , requeste d cut s i n socia l services , an d abou t external examinatio n int o som e type s o f persona l appropriatio n expenditure . Negotiations broke off in November 1979 . In that year the government expanded the money supply (b y 47 percent), as public expenditure rose by 34 percent and revenues declined . Negotiations with the IMF resumed in early 1980 , at the time that the secon d oil shock of 197 9 started taking its toll. Despite its hard conditions, the IMF loan was the only alternative , sinc e 198 0 official marketing s of foo d wer e poor, and significant cereal imports had to be obtained. The agreement was not conditional on a devaluation, however, and that made it politically more palatable. The Fund agreement release d othe r externa l fund s fro m th e Worl d Bank an d elsewhere . But by December 1980 , Tanzania failed to comply with the limitations on public borrowing an d imports , an d th e IM F agreemen t wa s suspended , cuttin g of f Tanzania from other sources of external finance . The foreig n exchang e crisis , whic h becam e mor e acut e i n 1981 , meant tha t the government had no other way but to cut imports, which it did, but not until 1982. That action intensified shortage s of consumer goods and at the same time

Background 1

5

curtailed domestic production of substitutes that relied on imported intermediate inputs. Large excess demand forced many Tanzanians into the parallel markets, where prices of scarce goods soared. Simultaneously, the supply of export crops and foo d crop s t o th e officia l parastatal s declined , worsenin g th e foreig n ex change shortage . The government obtained external commercial credit s at high interest rate s an d quickly accumulate d substantia l arrears . It use d inflationar y financing t o keep up recurrent public expenditures . The seriousnes s o f th e situatio n led, i n 1981 , to the first attemp t a t internal adjustment, the National Economic Survival Program (NESP). The basic thrust of the NESP was redistributionary. I t entailed a cut in salaries an d social services , an increase i n taxe s fo r salarie d workers , an d a n increase i n officia l produce r prices. Prices did not, however, reach the levels o f those in the parallel market, and marketed production did not increase. Inflation ros e strongl y du e to monetary expansion. The short-live d NES P gave ris e i n 198 2 t o a Structural Adjustmen t Progra m (SAP). This program attempted to implement the recommendations o f an impartial technical advisory group of experts funded by the World Bank. The aims of the progra m were t o limi t th e publi c defici t b y cuttin g recurren t expenditures , without cutting socia l and economic services , and to implement efficiency- an d productivity-enhancing measure s fo r parastatals , whil e maintainin g employ ment. Simultaneousl y th e governmen t undertoo k som e smal l devaluation s (1 0 percent in March 1982 ; 20 percent in June 1983 ) but kept tight import controls. The government anticipated increased IMF and World Bank financing with its program, bu t thi s di d no t com e becaus e som e o f th e precondition s fo r IM F agreement wer e no t met . I n th e absenc e o f a n IM F agreement, mos t bilatera l donors, with the exception of the Scandinavian countries, curtailed aid, and that made the situation worse. In 198 3 and 1984 , the budgets were based on the deflationary measure s that the IM F had demande d durin g earlie r negotiations . Fundamenta l change s i n orientation als o too k plac e i n 1984 , with th e liberalizatio n o f domesti c foo d markets and the abandonment of state monopolies and marketing boards. In June 1983, a further devaluation of 23 percent was announced, but that was still much lower than demanded by the IMF, which regarded external trade as the key growth constraint, while the government regarded domestic demand as the constraint. In June 1984, food subsidies were abolished. But although the measures were steps in the right direction, they did not seem to alter any of the fundamental problems; extensive shortage s an d paralle l market s prevailed . A n officia l crackdow n o n economic saboteur s and racketeers in 198 3 only worsene d the situation. The crisis and the lack of foreign exchange finally force d the government to abandon its efforts t o deal with the crisis alone, and it again started discussion s

16 Alexander

H. Sarris and Rogier van den Brink

with the IMF. In August 1986 , an agreement was reached based on the adoption of a three-year Economi c Recover y Progra m (ERP) , which include d severa l o f the earlier IMF conditions, the main elements of which were the following (Worl d Bank 1987a) : 1. Trade

policies

(a) Exchang e rate devaluation, with the goal of an equilibrium rate by 198 8 (b) A n improved foreign exchange allocation syste m (c) Tarif f refor m (d) Expansio n of the "own funds" import scheme first established in 198 4 (e) Improvement s i n the export retention schem e throug h expansion of th e list of good s allowed to be imported and the narrowing of differentials betwee n retention rates (f) Ai d coordination 2. Fiscal and monetary policies (a) Expenditur e ceiling s (b) Parastata l reforms , suc h a s subsid y reductio n an d commercial orienta tion (c) Credi t ceilings (d) Interes t rate adjustments toward a positive rea l rate 3. Price policies (a) Reductio n of the number items under price control (b) Larg e price increases for items controlle d 4. Agricultural policies (a) Produce r price increases for export crops, aiming at prices close to 60-70 percent of f.o.b. level s (b) Marketin g reforms , suc h a s foo d trad e liberalizatio n throug h abandon ment o f th e parastata l monopol y an d allowin g cooperativ e union s an d larg e producers to participate in export marketing (c) Liberalizatio n o f input supply and distribution; cooperatives and private organizations to be allowed to carry out these function s Adoption of the program and agreement with the IMF led to substantial inflow s of financin g fro m th e World Bank an d other donors. The main reforms under taken unde r the ER P were (1 ) a very larg e exchang e rat e devaluatio n (fro m 1 7 Tsh/US$ i n early 198 6 to 19 0 Tsh/US$ in March 1990) ; (2) establishment of an Open Genera l Licensin g Facilit y (OGL ) fo r foreig n exchang e allocation ;

Background 1

7

(3) positive real interest rates; (4) removal of price controls (reducin g from 40 0 to 1 2 the number of categories of goods controlled); (5) real increases in officia l producer price s fo r expor t crops ; (6 ) agricultura l marketin g reforms ; an d (7) establishment of fiscal an d monetary targets. In 1989 , th e governmen t adopte d a new , three-yea r Economi c an d Socia l Action Program (ESAP) that was to run until 1992 , and which stresses the socia l dimensions o f adjustment an d poverty alleviation , whil e continuing th e pace of economic reforms . Th e mai n element s o f thi s progra m ar e (1 ) continue d ex change rat e adjustment ; (2 ) continue d trad e polic y reforms ; (3 ) publi c secto r management reform, including parastatal reform; (4) financial secto r restructuring; (5) reform of agricultural pricing and marketing; (6) industrial restructuring; (7) rehabilitatio n o f infrastructure ; (8 ) rehabilitatio n o f socia l servic e deliver y capacity; an d (9 ) alleviatin g impac t of environmenta l degradation . Clearl y th e government has made substantial change s sinc e the late 1970 s and early 1980s , when i t strongl y resiste d reform s implemente d unde r th e ER P and envisione d under ESAP. GENERAL STRUCTURE OF THE MACROECONOMY According t o th e lates t officia l figures , th e Gros s Domesti c Produc t (GDP ) at factor cos t o f Tanzani a mainlan d i n 198 9 stoo d a t 351. 2 billio n Tsh . Wit h a mainland populatio n estimate d by th e 198 8 censu s a t 22.5 millio n people , th e per capita GDP in the same year (assuming 2. 8 percen t population growth) wa s 15,184 Tsh , or , a t th e averag e officia l exchang e rat e fo r th e yea r o f 143.3 8 Tsh/US$, i t wa s equa l t o US $ 105.90 . Thi s woul d pu t Tanzani a amon g th e poorest countrie s i n th e world . I n fact , th e 199 0 Worl d Ban k Developmen t Report classifie s Tanzani a a s the world's fourth-poores t country . B y compari son, an d again usin g officia l figures , th e current-pric e GD P in 197 8 wa s 28.5 8 billion Tsh, and the per capita GDP was 1,67 8 Tsh. At the then official exchang e rate o f 7.7 1 Tsh/US$ , tha t translates t o US$ 217.60 . Thi s apparent , enormou s decline i n pe r capit a incom e an d it s cause s wil l b e on e o f th e object s o f ou r investigation. Before we proceed, a major caveat is in order. The quality of national accounts statistics i n Tanzani a appear s t o hav e deteriorate d durin g th e las t decade . I n Chapter 5, we present ample evidence to support the view that official statistic s are seriously flawed , an d hence any conclusions based on them must be viewe d as tentative. In 198 8 an d 1989 , agriculture accounte d for 46.6 percent o f GDP in constant prices, bu t fo r 6 0 percen t o f GD P i n curren t prices . Nonagricultural-goods producing sectors (manufacturing, mining and quarrying, electricity, water , and

18 Alexander

H. Sarris and Rogier van den Brink

construction) accounte d for 13. 4 percent of GDP in constant prices, but only 7. 8 percent i n current prices. The remaining sectors , comprising private and public services (trade, transport, communications, finance, insurance, real estate, public administration, an d other services) , accounte d fo r the remainin g 4 0 percen t o f GDP in constant prices, but only 24. 1 percen t in current prices. Agriculture ha s grown i n importance i n the recent period, while public administration , a sector that had grown substantially i n the 1970s , seems to have declined in importance under the economic pressures of the 1980s . Table 2 exhibits aggregate and sectoral average annual growth rates over the period 196 6 t o 1989 . Durin g th e perio d 1976-1989 , th e aggregat e GD P in real terms gre w b y 2. 1 percen t annually , wit h th e mos t recen t four-yea r perio d exhibiting the fastest growth , and the five-year period 1980-198 5 th e smallest , compared wit h a n averag e growt h rat e o f 3. 8 percen t fo r th e ten-yea r perio d before 1976 . Althoug h nationa l account s dat a befor e an d afte r 197 6 ar e no t comparable becaus e o f a majo r revisio n i n method s an d coverag e base d o n primary 197 6 information , i t appear s fro m thes e officia l statistic s tha t th e TABLE 2 Tanzania: Averag e Annual Growt h Rate of GD P by Secto r (percentages) . Economic Activity 1971 Agriculture, forestry , fishing an d hunting 1. Mining and quarrying 0. Manufacturing 8. Electricity an d water 9. Construction 12. Trade, hotels, and restaurants 3. Transportation and communication 11. Finance, insurance, real estate, and business services 5. Public administration and other services 6.

1966-

4 1 4 4 2

1971- 19761976 1980 3.7 -7.1 6.5 7.8 -1.0

19801985

1.0 -2.5 -1.1 16.4 1.8

19851989

3.0 -1.5 -4.9 2.9 -5.9

19761989

4.8 -5.5 2.6 6.3 8.1

2.9 -3.3 1.5 7.9 -0.6

4

2.4 0.

0 -1.

3 6.

1 1.

4

1

4.9 2.

1 -3.

4 3.

4 0.

2

4

3.7 5.

1 4.

2 4.

5 4.

6

8

13.5 8.

3 2.

6 1.

2 3.

0

Total industries 4.

0

4.3 2.

1 0.

9 3.

9 2.

2

GDP at factor cost 3.

8

4.3 2.

0 0.

7 3.

9 2.

1

Sources: Tanzania Bureau of Statistics (1987); TET (1989, 1990). Notes: Dat a for 1966-197 5 calculate d fro m Tanzania , Burea u of Statistic s (1987)1 ; 1976-198 8 calculated from information in TET (October 1989, January 1990).

Background 1

9

post-1976 perio d has been one of more-or-less continuous crisis . Public administration was the fastest growing sector throughout the 1970s, while construction has been the fastest growin g secto r since 1985 . The overall annua l growt h rate of official GD P in the 1 3 years since 197 6 (2.1 percent) has been smaller than the 1978-1988 (intercensal ) annua l population growt h rate, which was 2.8 percent . The major factor affecting th e economy i n the 1970 s and 1980 s has been the adverse external position. Although during the 1960s and mid-1970s the external accounts wer e i n balance , wit h export s closel y trackin g imports , a n externa l imbalance started in 1978, which grew throughout the next decade and continues today. In 1976 , the deficit i n the trade balance was 2.2 percent of GNP. In 1978 , it jumped to 15.2 percent of GNP. During the crisis period 1980-1984, it declined to about 8 percent of GNP, but by 1988 it had surged to 26 percent of GNP. Another significant aspec t o f Tanzania's developmen t ha s bee n th e consistentl y larg e share o f GD P (around 2 0 percent ) tha t i s absorbe d b y fixe d capita l formation , excepting the period 1983-1985. A large proportion of that has been financed by aid from various donors. In the external accounts, merchandise exports comprised about 60 percent of receipts fro m export s o f good s an d service s unti l 1984 . Merchandis e import s constituted a n eve n large r proportion o f import s o f good s an d services , large r than 90 percent . A sizabl e proportio n o f tota l externa l receipt s al l throug h the period, especially afte r 1985 , is accounted for by current transfers from abroad. Before 1985 , thes e wer e mainl y official . Sinc e 1985 , however , th e shar e o f transfers i n external receipt s has increased substantiall y accountin g fo r 57 percent and 60 percent of total export receipts in 1987 and 1988. This compares with a range of 20-2 5 percen t in the period before 1985 . The rise reflects increasin g imports under own-funds sinc e 1984 , for which a balancing credit item is added in the external inflow s categor y unde r transfers. In the pre-198 5 period , mos t o f th e financin g o f th e curren t accoun t defici t was throug h capita l transfers . Sinc e 1985 , however , a significan t amoun t ha s been financed b y net external borrowing . Thi s i s bound to create seriou s problems of debt servicing. In fact, since 198 4 the outflows o n the capital account of the balance o f payments have been clos e t o o r larger than inflows, althoug h i n the pre-1984 period the inflows wer e substantially higher . Tanzania's si x top export earners are agricultural primary products—coffee , cotton, sisal, tea, tobacco, and cashew nuts . Together they account for about 60 percent of total export earnings. Since th e mid-seventies, th e volumes of all si x have declined significantly, with declines ranging from 1 0 percent (coffee) t o as much as 80 percent (sisal). The bulk o f importe d item s consists o f capital an d intermediate goods , wit h consumer good s accountin g fo r onl y abou t 2 0 percen t o f tota l imports . Foo d

20 Alexander

H. Sarris and Rogier van den Brink

imports accoun t fo r 40-6 0 percen t o f consume r imports . Ver y fe w import s o f nonessential, consumer-incentiv e good s see m t o hav e bee n officiall y recorde d before 1988 . Table 3 indicates Laspeyres volume indices for exports and imports, with base weights for the year 1976 . It is quite clear from the table that compared with the early an d late 1970s , the decade o f the 1980 s saw a significant reductio n in the officially recorde d quantity of exports. The reduction affected al l export categories. A slightly differen t patter n seems to have been followed by imports. There seems t o have been a massive decline i n the early 1980s , particularly o f capita l and consumer goods. Imports of intermediat e good s (amon g which oil product s feature prominently) increased substantially until 1981 and then declined precipitously unti l 1986 . It was only i n 198 8 tha t imports of these good s increased i n value t o a level simila r t o tha t o f 1980 . Th e trend s obviou s i n Tabl e 3 do no t change significantly i f one changes the base year for the calculations of the index weights. In Table 4 we exhibit unit value indices for exports and imports and the barter and income terms of trade from 197 4 to 1986. With the exception of 1977 , which was the year of the world coffee boom , the barter terms of trade (defined a s the ratio o f th e uni t valu e o f export s t o th e uni t valu e o f imports ) see m t o hav e declined, despite the fact that oil prices have declined since the second oil shock of 1979 . Given the decline in the volume of exports evident in Table 3, it is only natural that the income terms of trade (defined as the ratio of the value of exports divided by the unit import value) woul d exhibit a n even steepe r decline; this i s evident from the final ro w of Table 4. In 1987 , 55 percent of all exports were destined for western Europe, a larger share than in 198 0 (48 percent) . Germany, UK, Netherlands , Italy, an d Finland currently absor b th e bulk o f export s t o wester n Europe. Countrie s i n Asia an d Oceania absor b anothe r 2 5 percen t o f Tanzania' s exports , wit h India , Japan , Singapore, Taiwan , and Hong Kong a s the largest markets. Another 1 2 percent of export s g o to African countries , with Kenya and Algeria accounting i n 198 7 for abou t hal f o f that . Th e Unite d State s account s fo r a meage r 2 percen t o f Tanzania's exports , a share tha t is hal f o f wha t i t was i n th e earl y 1980s . Th e geographical patter n o f import s resemble s tha t o f exports , with 5 7 percen t originating in western Europe, 1 9 percent in Asia and Oceania, 1 2 percent in the Middle East (largely oil products), and a meager 5 percent from the United States. The public sector is a dominant aspect of the Tanzanian economy, accounting for a large portio n o f th e monetary economy . I n the earl y year s afte r indepen dence the public secto r was kept small, and the private secto r functioned muc h as it did before independence . I n 1967 , wit h the Arusha declaration, the public sector starte d becoming heavil y involve d i n all aspect s o f th e Tanzanian econ -

9 1980

7957 1982

1983

74.81 78.8 1 75.9 1 73.47 68.3 8 62.6 6

93.72 100.0 0 81.8 1

90.96 100.0 0 73.4 8

Source: Tanzania , Burea u of Statistics , "Foreign Trade Statistics, 1987 " (March 1989) .

64.97 80.5 2 84.6 9 2 101.73 100.0 0 88.8 2 113.39 85.-9 6 83.6 4 1 158.08 100.0 0 91.3 0 76.91 52.7 8 84.9 8

73.17 76.1 5 73.2 2

90.13 100.0 0 80.5 0

47.21 38.7 4 39.1 0 58.44 55.8 3 36.1 6

67.26 62.2 4 59.0 2 59.02 62.6 2 40.7 4

77.18 88.3 6 72.1 4 59.85 49.8 5 46.4 7

77.73 58.1 9 46.8 3 71.08 72.8 7 66.4 2 66.20 62.9 4 65.1 6 100.83 116.5 3 105.3 0

64.24 70.7 0 66.2 7

72.70 74.1 3 56.5 4

1986

58.00 58.9 5 56.2 3

1984 1985

69.45 68.9 7 55.46

81.65 100.0 0 88.9 7 158.45 140.9 4 126.6 8 181.61 63.3 0 53.8 4 87.71 100.0 0 87.5 7 119.47 81.8 8 72.8 4 57.71 55.26 37.2 8

7975 797

Imports (all) 132.8 Capital goods 249.4 Intermediate goods 90.6 0 Consumer goods 130.6 0

1977

75.79 100.0 0 91.9 2

7975 1976

Exports (all) 86.1 8 Food, drink, and tobacco 75.7 1 Raw materials, including fuel 99.76 Manufactured products 81.8 1

1974

TABLE 3 Tanzania: Laspeyre s Volum e Indices fo r Domestic Export s an d Direct Imports, 1974-198 6 (1976= 100) .

1977

1978 1979

1980

1981

1982

1983

1984

1985

1986

5 74.2 1 100.0 0 115.6 5 94.0 4 76.2 2 67.7 7 66.7 8 45.3 7 57.7 3 76.5 1 56.2 3 60.0 3

0 62.5 0 100.0 0 91.2 5 69.8 2 60.3 4 49.4 1 45.9 6 30.6 8 31.6 2 44.3 7 34.5 6 35.3 2

Barter terms of trade 97.4

Income terms of trade 87.6

Source: Tanzania , Burea u of Statistics , "Foreig n Trade Statistics, 1987 " (March 1989) .

7 91.7 5 100.0 0 126.3 5 117.1 1 144.6 5 140.6 6 126.0 8 289.9 3 221.1 5 241.0 9 336.0 4 748.5 2 7 97.8 0 100.0 0 109.1 4 95.2 3 150.1 4 215.1 4 246.1 1 276.8 0 337.3 2 405.7 5 493.3 7 809.7 8

4 99.2 8 100.0 0 120.2 1 128.2 9 177.4 9 234.6 0 252.5 0 331.0 9 326.5 9 335.0 7 431.5 6 761.5 0 7 109.2 6 100.0 0 131.8 1 194.2 4 255.7 9 359.8 8 390.2 1 458.9 9 401.2 2 306.1 3 419.1 1 685.2 3

9 113.9 2 100.0 0 113.1 0 166.0 1 201.7 7 245.25 202.95 174.6 8 236.8 7 277.2 5 286.9 2 572.75

9 81.5 4 100.0 0 121.1 8 99.5 6 130.4 7 166.0 0 176.0 7 163.7 0 204.7 1 291.7 9 267.8 6 333.8 0

0 63.5 4 100.0 0 151.6 2 123.4 4 127.0 3 141.9 8 159.6 2 139.5 9 173.5 4 236.2 9 228.8 8 498.6 7

4 73.6 8 100.0 0 139.0 2 120.6 4 135.2 9 159.0 0 168.6 1 150.2 1 188.5 3 256.3 5 245.7 0 457.1 4

1976

Imports (all) 78.4 Capital goods 66.8 Intermediate goods 73.0 Consumer goods 86.2

Exports (all) 76.4 Food, drink, and tobacco 58.1 Raw materials, including fuel 104.2 Manufactured products 109.2

1974 1975

TABLE 4 Tanzania: Laspeyre s Uni t Value Indices fo r Domestic Export s an d Imports by Commodity Grou p (1976= 100) .

Background 2

3

omy. Through nationalization, the state acquired control of a significant proportion o f th e productiv e asset s o f th e nonagricultura l sector , a s wel l a s th e large-scale agricultura l sector . I n subsequen t years , parastatal s proliferate d t o the point that by the mid-1980s ther e wer e 425 o f the m i n Tanzania, a number comparable onl y t o countries suc h a s Brazi l an d Mexico wit h GDP s 50 an d 3 5 times, respectively, that of Tanzani a (World Bank 1988) . Table 5 presents a summary of central government revenues and expenditure for the period 1975-197 6 to the most recent available year. It can be noticed that ever since 197 6 there has been a public-sector deficit, amountin g to as much as 14.5 percent , an d neve r les s tha n 7. 5 percent , o f GD P at marke t prices . I n th e composition of public expenditure, current expenditure has been kept quite high, while developmen t expenditur e sinc e 198 2 ha s decline d t o abou t hal f o f it s pre-1980 shar e of GDP. A large share of total public expenditures i s for internal transfers, suc h as interest on public deb t and subsidies to parastatals. Table 6 exhibit s th e compositio n o f publi c revenue s an d expenditur e fro m 1975-1976 t o 1986-1987 . Trad e taxe s (impor t an d expor t duties ) constitut e a small portion of total public revenue, declining fro m a high of about 21 percent in 1978-197 9 t o onl y 6 percen t o f revenue s i n 1982-1983 , t o recove r t o 1 3 percent i n 1986-198 7 an d a n estimate d 1 5 percen t i n 1988-1989 . I t i s quit e noticeable tha t i n contras t t o man y primar y exportin g countries , expor t ta x revenues since 198 0 are almost negligible. However, export producers are heavily taxed implicitly, as will be seen later, with the implicit revenue being largely absorbed as parastatal marketing costs. The bulk of public revenu e come s fro m consumption an d excise taxes , as well a s personal an d income taxes. Turning to the composition of public expenditure, the most noticeable trends in the eightie s ar e the increas e i n th e shar e o f genera l publi c service s (mainl y administration), th e shar p declin e i n th e share s o f education , health , an d eco nomic service s to about half their 1975-197 6 shares , and the quadrupling of the share of servicin g th e public debt. The importance of the public sector is underestimated by the data in Table 5. Parastatals productio n accounte d fo r abou t 1 3 percent o f GD P by 1985 , a share that ha d grow n rapidl y fro m 7 percen t i n 196 7 an d 9 percen t i n 1972 . I n manufacturing, parastatals accounted for about 47 percent of value added and 47 percent o f wag e employment . I n al l sector s excep t agriculture , parastatal s ac count for more than 20 percent of sector value added and an equal or larger share of wag e employment . Whe n tha t i s adde d t o th e 1 3 percen t shar e o f publi c administration i n the GDP , we obtai n a combined shar e o f th e publi c secto r i n reported GDP of about 26 percent. The importance of the public sector is reflected in the structure of formal wage employment. "Forma l employment " refers t o peopl e workin g i n forma l estab -

1981

1982

1983

0 10.1

4 8.2

(7.54)

9 11.8

(11.05) (10.36 ) (13.16 ) (14.86 ) (13.06 ) (14.48 ) (12.52 )

12.56 13.6

6.45

11.39

11.26

1980

1984

1985 1986

1987

(7.64)

6.00

5 6.3

8 (8.55) (10.78 ) (11.93 )

4.88 5.4

19.34 19.0 1 18.8 6 18.6 4 18.9 7 19.70 18.97 16.12 15.9 8 18.0 5 19.94 20.1 8 20.1 2 22.9 8 23.2 7 20.80 20.61 19.79 21.3 6 22.5 7 (0.60) (1.17 ) (1.26 ) (4.34 ) (4.30 ) (1.10) (1.64) (3.68) (5.39 ) (5.55 )

1978 1979

2

19.41 18.38 1.03

18.53 18.32 0.21

7977

Sources: Worl d Bank, "Tanzania Public Expenditure Review," Report No. 7559-TA (Ma y 22 , 1989) ; and authors' computations . Note: GD P data are reported in calendar year while publi c expenditur e dat a are in fisca l years . To compute th e ratios to GDP indicated in the table fo r each calendar year, the average o f th e two adjacent fiscal year s wa s used in the numerator o f th e relevant fractions .

Percent of GD P at market prices Current revenue Current expenditure Surplus/(deficit) Development expenditure Total surplus/ (deficit)

1976

Tanzania: Summar y of Central Government Operation, 1976-1987 .

TABLE 5

0.2 1.2

2.3 38.1 7.9 5.9

0.4

1.8

2.4 36.9 9.2 7.3 1.7 34.1 13.6 7.0

0.9

0.2

16.0 13.5 13.3 6.7

21.6

17.2

17.4 12.3 13.6 7.1

36.4 23.1

33.6 26.8

15.8 12.2 14.1 7.1

10.1 8.9

6.2 16.1

1.8 35.1 4.8 3.9

0.8

0.3

16.0 24.6 11.3 5.3

9.0

40.6 29.5

13.9 7.0

1.9 36.1 22.0 7.5

1.0

0.4

2.1 34.8 16.9 11.4

1.2

0.3

16.4 11.0 11.8 5.4

7.3

6.2 14.7 7.7 11.2 5.0

51.3 32.2

1.1 2.5

40.2 33.0

11.0 6.3

2.1 29.8 18.5 17.8

1.0

0.3

17.9 12.5 12.5 5.4

13.7

52.5 33.2

6.9 0.2

2.0 27.0 21.0 20.2

1.1

0.3

17.1 13.3 13.2 5.1

15.6

49.5 30.8

5.9 0.1

2.0 26.0 18.8 18.1

1.0

0.3

22.0 12.8 11.7 5.5

15.6

51.6 26.5

6.2 0.1

7957- 7952- 19837952 1983 1984

Sources: Worl d Bank, "Tanzani a Public Expenditur e Review, " Report No. 7559-TA (Ma y 22, 1989) .

Expenditures General public servic e Defense Education Health Social security and welfare Housing and community amenities Other community socia l services Economic service s Other purposes Public debt

Revenues Import duties 11.7 Export duties 4.1 Consumption and excise duties 42.4 Income and personal taxes 28.4 Other taxes and income sources 13.4

7975- 7976- 7977- 7975- 7979- 19807976 1977 1978 7979 1980 1981

2.2 24.2 16.1 15.4

1.0

0.5

29.9 13.9 7.3 5.0

9.9

55.9 25.8

8.4 0.1

19841985

0.5 22.8 25.0 24.2

0.6

0.2

28.9 10.4 7.3 4.3

9.5

52.1 30.9

7.5 0.0

0.5 16.5 32.1 31.4

0.4

0.3

25.5 14.6 6.4 3.7

10.6

53.4 23.0

13.0 0.0

7955- 19867956 1987

TABLE 6 Tanzania: Compositio n of Revenue and Expenditure, 1975/1976/-1986/198 7 (a s percentage of total revenue and percentage of public expenditure).

26 Alexander

H. Sarris and Rogier van den Brink

lishments, an d henc e doe s no t coun t smallholde r agricultur e o r small-scale , unincorporated, "informal" sector activities, which account for the great bulk of employment i n the country. In 1965 , government employment accounte d for 27 percent o f the total forma l wag e employment o f abou t 250,000, and parastatals for 5 percent, for a combined total of 32 percent. In 1970 , the joint share rose to 59 percen t (ou t o f a total o f 376,000) ; i n 197 6 t o 6 6 percen t (ou t o f a total o f 481,000); and in 1984 it stood at a dominating 77 percent (of a total of 633,000). Almost all the growth in formal employmen t in the country since 197 0 occurred in the publi c sector , wit h th e numbe r employed i n governmen t an d parastatals more tha n doublin g ove r th e period , whil e forma l employmen t i n th e privat e sector grew very little. Of the total of 622,000 formal-sector employees in 1981, 120,000 were listed by the Bureau of Statistics as casual. Agriculture (plantations and larg e stat e an d privat e farms) , occupie d 119,00 0 employees , industr y 135,000, constructio n 46,000 , transportatio n 55,000 , an d all th e othe r service s remaining 266,000 . The formal financia l secto r in Tanzania is not large. Apart from the Bank o f Tanzania, which is the central bank, the commercial banking system is composed of one major mainland bank (the National Bank of Commerce) and a few smaller, specialized one s (Cooperativ e Rura l Developmen t Bank , Tanzani a Investmen t Bank, Tanzania Housing Bank, Tanganyika Post Office Saving s Bank, Tanganyika Developmen t Financ e Compan y Ltd.) . The major function o f th e bankin g sector ha s bee n t o len d t o parastatal s an d t o th e government , wit h ver y littl e lending extende d t o the private sector . Th e bulk o f th e lending tha t is reporte d as goin g t o th e privat e secto r i n fac t goe s t o cooperatives . S o almos t al l ban k lending i n Tanzania i s directed to the public sector . Thi s has consequences fo r the causes of inflation , a s will be seen later.

I 3 MACROECONOMIC POLICY AND PERFORMANCE

TTo understan d th e natur e o f curren t adjustmen t effort s i n Tanzania , i t i s necessary t o review analyticall y th e evolution o f macroeconomic an d other key policies and government responses over the last two decades. There are four periods that seem to characterize economi c development s an d policies i n Tanzania. Th e firs t cover s th e perio d fro m independenc e i n 196 1 t o th e Arush a declaration i n 1967 . Th e secon d extend s fro m 196 7 t o 1973 . The thir d cover s the period from 197 4 to 1982 , and the final on e concerns the recent years, from 1982 to now. The last period could be further subdivided into 1982-1985 , when Tanzania attempte d to adjust withou t externa l help , an d the period afte r 1986 , since the agreement with the IMF. Each subperiod is characterized by a different set of policies an d external influences an d hence must be examined separately . Given ou r interest in structural adjustmen t an d the recovery, ou r natural focu s will b e o n th e las t tw o o f th e fou r period s mentioned , particularl y th e mos t recent, since the earlier ones have been covered extensively i n other literature. Our effort wil l be to understand the extent to which economic performance ha s been relate d t o policies an d external shock s an d how th e governmen t an d the economy o f Tanzania have responded . Our analysis in this chapter starts with a review o f debat e on the origins and causes of the economic crisis. The second section reviews recent macroeconomic 27

28 Alexander

H. Sarris and Rogier van den Brink

developments and existing interpretations of the trends. The third section reviews the debate concerning devaluation, the policy that has been the source of the most controversy i n th e contex t o f stabilizatio n an d adjustment . I n th e subsequen t section we examine the financing of the domestic and external deficit. In the final section we make new estimates of the unobserved, "second" economy in Tanzania and provide a picture of the total economy, observed and unobserved. INTERPRETATION OF THE CRISIS AND THE ADJUSTMENT DEBATE There has been wide debate concerning both the causes of the crisis in Tanzania and the appropriate responses. In this section we shall review some of the major arguments, providing whereve r possible som e new elements . Our earlier review of the structure of the economy and the evolution of policy from independence until the late 1970 s shows a continuing effort by the government to control the economic force s an d processes i n the country through direct control o f al l sphere s o f economi c activity . Th e underlyin g rational e fo r thi s attitude wa s tha t th e bes t wa y t o achiev e a fas t pac e o f economi c an d socia l development wa s to rely o n a top-down syste m of economi c an d social contro l and modernization through adoption of foreign technology. These attitudes were manifested i n th e nationalizatio n o f privat e firms , th e abolitio n o f produce r cooperatives, the villagization campaign, the substitution of state production and marketing of both agricultural and nonagricultural products, and the adoption of the largely foreign-exchange-intensiv e Basi c Industrie s Strategy . Th e result o f these policies , b y th e lat e 1970s , wa s a n economic structur e i n whic h activit y was suppl y constraine d an d depende d quit e heavil y o n import s fo r it s prope r performance. Despit e effort s a t diversification , agricultur e an d even u p to th e present, remains dominant as a generator of income and, more crucially, foreig n exchange. Ndulu (1988) has aptly summarized the main macro features of the Tanzanian economy. Beside s the overwhelming importanc e of agricultur e and the supply constrained, import-dependen t natur e of production , th e links amon g the fisca l deficit, th e balance o f payments , and the money suppl y process ar e key. Whil e in th e lat e 1960 s an d earl y 1970s , th e monetar y bas e wa s backe d mainl y b y foreign asset holdings, by the late 1970s , almost 90 percent of the expansion of the monetary base was due to net claims on government. This latter development was the result of high domestic credit demand from the public sector , includin g the parastatals. In 1978, the 196 5 Bank of Tanzania Act was amended to lift the limit on government borrowing, which had been 25 percent of recurrent revenue. This amendment allowed shar p and inflationary increase s in the monetary base, in order to accommodate the mounting deficits .

Macroeconomic Policy and Performance 2

9

The fina l majo r featur e identifie d b y Ndul u i s th e contras t betwee n th e flex-price natur e o f th e foo d markets , an d the fix-pric e natur e o f man y o f th e nonfood markets , especiall y thos e fo r domesti c manufacture s an d services . Prices i n the latter sectors wer e largely administere d an d insulated from worl d market developments. In terms of political economy, besides the ruling party (CCM) and the government, Ndul u identifie s thre e majo r interest groups , namely , th e largel y urban based wage/salar y earners , includin g governmen t employees ; th e commercia l entrepreneurs; an d th e rura l peasantry . Polic y i n th e 1960 s an d 1970s , whil e emphasizing modern, urban-centered development, tried explicitly to benefit the peasants throug h provisio n o f socia l service s a t th e expens e o f commercia l entrepreneurs. Durin g th e lat e 1970s , however , budgetar y pressure s le d t o formal-sector wag e restraint s an d shar p decline s i n th e rea l income s o f urba n wage earners, while peasants were better insulated through the growth of parallel markets. Th e commercia l entrepreneurs , mostl y traders , gaine d significantly , however, as the erosion of government controls led to massive parallel and black markets (some legal, but mostly not ) both domestic an d external. Bienenfeld (1989 ) ha s summarize d th e proces s tha t le d t o th e crisi s o f th e early 1980 s a s the gradua l constructio n o f a geographically dispersed , import intensive, urbanize d economy , criticall y dependen t o n expandin g agricultura l surpluses t o fee d th e urba n populatio n an d ear n foreig n exchange . Th e crisi s arose because the agricultural surplus did not grow fast enough. There has been wide debate concerning the causes of the crisis that led to the first IMF negotiations in 1979-1980 and subsequent adjustment efforts. There are those tha t argu e tha t th e majo r cause s wer e mainl y external , suc h a s th e oi l shocks, th e drought , an d the Uganda wa r (Green, Rwegasira, an d van Arkadi e 1980), whil e other s sugges t tha t th e cause s wer e mostl y interna l an d du e t o economic mismanagemen t (Sharple y 1985 ; Lofchie 1989) . In Table 7, we display the Tanzanian balance of payments in U.S. dollars for 1970-1988 a s reported by the IMF. The number that gives a clear picture o f the evolution of the crisis is the basic balance, that is, the sum of the balance of trade in good s an d service s an d net long-term capita l inflows . I t can be clearl y see n from Table 7 that 197 4 and 1978 produced the first major external crises. The 197 4 crisis appears to have been caused by a sudden, 50 percent increase in impor t expenditure s (i n U.S . dollars) , whil e export s increase d b y onl y 1 0 percent. In the subsequent three years, 1975-1977, export earnings increased due to the coffee boom , an d import expenditure declined . I n 1978 , a sharp drop in merchandise expor t earnings, simultaneous with a 53 percent increase in import expenditure, created the new crisis. While the high import expenditures in 197 8 and 197 9 ca n b e rationalize d i n term s o f th e spendin g o f foreig n exchang e

1971 7972

1973 1974 1975 1976

1977

1981 1982 1983 1984 1985

1986

1979

1987

1988

625.2 697.2 164.0 174.8 29.5 23.1 140.9 145.3 -1,262.6- 1,218.5 -473.4 -346.5 136.0 225.4 -337.4 -121.1 60.7 -75.1 63.5 98.9

1978

Source: IM F International Financia l Statistic s (variou s issues). Inflow s ar e described b y positive numbers , while outflow s b y negative .

1. Exports of goods and services 686.7 884.8 530.2 486.3 480.3 432.8 456.6 441A 499.4 2. Total transfers 128.7 130.3 119.2 103.4 158.6 474.1 707.0 722.0 456.3 Private unrequited transfers (net) 21.8 22.6 25.4 18.8 62.1 250.6 230.0 233.3 232.0 Official unrequite d transfers (net ) 106.9 107.7 93.8 84.6 96.5 223.0 223.5 477.0 490.0 3. Imports of goods and services -1,249.0 -1,187.0 -1,030.4 -785.7 -853.4 • -1,036.5 -1,105.0 -1,419.4 -1,479.8 4. Current account surplus (1+2+3 ) -433.6 -171.9 -381.0 -196.0 -214.5 -147.4 -174.3 -265.3 -258.4 5. Long-term capital inflows (net ) 166.3 204.5 167.9 177.7 89.7 -39.5 -24.7 -36.5 31.8 6. Basic balance (4+5) -267.3 32.6 -213.1 -18.3 -124.8 -186.9 -199.0 -301.8 -226.6 7. Short-term capital inflows (net ) -85.8 57.0 157.6 -136.6 -108.1 -410.9 895.6 4.4 -5.9 8. Exceptional financin g 270.1 89.1 45.6 321.3 305.8 469.4 -657.3 124.2 163.9

1980

1. Exports of goods and services 321.8 349.7 411.7 455.8 488.4 391.3 633.2 656.4 2. Total transfers 12.8 5.8 -4.1 49.2 102.3 54.6 114.7 4.9 Private unrequited transfers (net ) 11.1 3.5 -14.5 -14.4 -11.4 11.5 11.5 19.4 Official unrequite d transfers (net ) 1.7 2.3 10.4 19.3 60.6 90.8 43.1 95.3 3. Imports of goods and services -370.2 -455.2 -473.3 -568.2 -822.9 -823.6 -722.3 -843.5 4. Current account surplus (1+2+3) -35.6 -99.7 -65.7 -107.5 -285.3 -330.0 -34.5 -72.4 5. Long-term capital inflows (net ) 71.6 137.7 108.3 155.3 117.6 170.5 102.4 100.7 6. Basic balance (4+5) 36.0 38.0 42.6 47.8 -167.7 -159.5 67.9 28.3 7. Short-term capital inflows (net ) -7.8 -37.7 -8.0 -18.0 -4.6 20.9 11.3 -18.4 8. Exceptional financin g 7.1 0.8 21.6 35.9

1970

Tanzania: Balanc e of Payments, 1970-1988.

TABLE 7

Macroeconomic Policy and Performance 3

1

reserves accumulated in 1977, the Uganda war, and the 197 9 oil price hike, it is hard to see why impor t expenditure staye d at a continued high level until 1981 . In fact , apar t fro m a smal l declin e i n 1981 , i t wa s no t unti l 198 2 tha t impor t expenditures were seriously curtailed. Meanwhile export earnings seemed to be doing quite well until 1981 , but from 1982 on, they took a sharp dive from which they have not as yet recovered. When we examine the volume and value indice s of merchandis e trad e exhibite d earlie r i n Table s 3 an d 4 , i t ca n b e see n tha t although th e barte r term s o f trad e hav e turne d agains t Tanzani a betwee n th e "non-crisis," "non-boom" year 197 5 an d the earl y 1980s , i t wa s th e declin e i n the volume o f export s tha t seem s t o have bee n a key elemen t i n the decline i n export earnings. The well-documente d declin e i n tota l an d certainly pe r capita, expor t cro p production contemporaneous with an increase in total food crop production (se e Chapter 5 ) hav e bee n rationalize d b y severa l analyst s (Odegaar d 1985 ; Elli s 1982) in terms of declining official produce r prices for export crops, both in real terms and relative to those of food crops (especially in the parallel, uncontrolled markets), the declining efficiency an d resultant rising marketing margins of the official marketin g system, which resulted in large transfers from peasants to the state (Ellis 1983) , and by the unavailability of incentive consumer goods in rural areas due to official rationin g (Bevan 1989) . The response of the government of Tanzania to the major shocks of 197 8 and 1979 wa s slo w i n coming . Curren t spendin g continue d a t a rapid pace . Whil e during the period 1976/7 7 t o 1979/8 0 curren t expenditures gre w by a n average of 13. 9 percent annually, in the next three years they grew by an average of 23.7 percent annually, despit e the foreign exchang e crisi s and the fact that recurrent revenues gre w b y onl y 18. 7 percen t annually . Developmen t expenditure s als o continued at a high rate as a share of GDP, and it was only in 1982 that they were severely curtailed. Thus the current deficit gre w from a small 1. 2 percent of GDP in 197 9 t o 4. 3 percen t o f GD P in 1982 . Th e tota l governmen t deficit , whic h reached a high of 14. 9 percent of GDP during the crisis year 1979 , declined only marginally t o 12. 5 percen t b y 198 2 an d wa s no t seriousl y cu t unti l 1983 . Th e portion of the fiscal defici t tha t was financed through bank borrowing increase d from 8 percent in 1977-197 8 t o 50 percent in 1979-1980 . The net claims of the banking syste m o n th e governmen t gre w b y a n averag e 19. 5 percen t annuall y between 197 6 and 1978 . They jumped by 73 percent in 197 9 and by a n average of clos e t o 3 0 percent annuall y fo r the next three years. It thus appears that the government attempted to keep up demand in the face of a supply shock . The resul t wa s a shar p increas e i n inflation . A s measure d b y th e Nationa l Consumer Pric e Inde x (NCPI) , prices tha t ha d increase d b y 6.9 , 11.6 , 6.6 , an d 12.9 percent in 1976,1977,1978, and 1979, respectively, jumped by 30.3 percent

32 Alexander

H. Sarris and Rogier van den Brink

in 198 0 an d kept growin g a t a rate higher than 25 percen t fo r the nex t severa l years. At the same time, the exchange rate kept appreciating because the government steadfastl y refuse d t o mak e mor e tha n token adjustment s i n th e nomina l exchange rate . The result wa s a sharp increase i n the parallel marke t premiu m on foreign exchang e and hence further evasion of the official markets . In Table 8 we exhibit the official an d parallel market exchange rates, as well as estimates o f the real exchange rat e (column 9) computed by a simplificatio n of the IMF method, as explained in Appendix A. It is evident from that table that the paralle l marke t premiu m (colum n 6)—tha t is , th e percen t b y whic h th e parallel market rate exceeds the official rate—jumpe d to more than 15 0 percent in 1980 and continued at levels above 200 percent until 1987 . The computations of th e rea l exchang e rat e (colum n 9 ) sho w a significan t appreciatio n o f th e Tanzanian shillin g startin g i n 198 1 an d continuing unti l 1986 , the first yea r of the ERP. It thus appears that while external shocks were instrumental i n initiating the crisis, i t wa s largel y wea k interna l adjustmen t efforts , couple d perhap s wit h optimism concernin g th e continuatio n o f foreig n capita l an d aid inflows , tha t eventually led to an uncontrollable situation . Underlying causes were the structural weaknesse s o f th e economy—reliance o n agricultura l export s fo r foreig n exchange generation, import dependence by the industrial sector, and the ineffi ciencies of the parastatal marketing sector . MACROECONOMIC PERFORMANCE UNDER ADJUSTMENT

The history of adjustment efforts in Tanzania over the past decade reveals strong efforts a t expenditure control, but consistently excessive optimism in economic targets. Ndul u (1988 ) presente d table s contrastin g SA P targets an d th e actua l performance o f severa l macr o indicators over the years of th e first adjustmen t effort, 1982-1985 . I n term s o f fisca l targets , performanc e durin g tha t perio d was bette r tha n planned . However , increase s i n externa l resources , i n expor t earnings as well as external loans and grants, fell far short of what was expected. The seriou s externa l ga p tha t resulte d pushe d th e governmen t t o resor t t o inflationary domesti c finance , an d money suppl y gre w faste r tha n planned i n all but the first yea r of the SAP. The target s se t fo r th e 1986/8 7 t o 1988/8 9 ER P were quit e ambitious : a n average rate of economic growth of 4-5 percent annually, a progressive reduction in the rate of inflation to less than 10 percent in 1988/89 from over 30 percent in 1985, improvement s i n th e externa l positio n throug h faste r expor t growth , increases i n utilizatio n rate s i n manufacturin g fro m 20-3 0 percen t i n 198 5 t o 60-70 percent by 1989 , and increases in export earnings of 1 6 percent annually.

1965 1966 1967 1968 1969 1970 1971 1972 1973 1974 1975 1976 1977 1978 1979 1980 1981 1982

7.143 7.143 7.143 7.143 7.143 7.143 7.143 7.143 6.900 7.143 8.264 8.324 7.960 7.415 8.221 8.182 8.322 9.567

7.143 7.143 7.143 7.143 7.143 7.143 7.143 7.143 7.021 7.135 7.367 8.377 8.289 7.712 8.217 8.197 8.284 9.283

8.52 8.64 8.68 8.25 9.10 10.45 15.00 15.40 13.45 14.00 25.00 20.40 15.05 11.75 13.50 26.50 24.35 29.15

n.a. 8.80 8.70 8.50 8.70 10.10 11.60 15.20 14.53 13.46 20.58 21.93 21.47 13.07 11.98 21.02 26.57 32.60 1.193 1.210 1.215 1.155 1.274 1.463 2.100 2.156 1.949 1.960 3.025 2.451 1.891 1.585 1.642 3.239 2.926 3.047

Ratio Official Official Parallel Parallel Rate Rate Rate Rate (3)1(1) End of Period End of Period End of Period * Average Period Average Period (2) (5) (3) (4) (1) n.a. 1.232 1.218 1.190 1.218 1.414 1.624 2.128 2.070 1.886 2.794 2.618 2.590 1.695 1.458 2.564 3.207 3.512

143.21 134.77 122.38 110.51 100.00 102.92 101.56 97.38 90.99 87.05 87.31 87.27 79.50 71.34 77.41 67.21 60.15 56.82 1.04 1.05 1.00 0.99 1.00 1.00 1.11 1.13 1.22 1.29 1.25 1.27 1.47 1.63 1.57 1.57 1.33 1.22

Real Other Exchange Currency Ratio Rate* vis-a- Correct. (4)1(2) vis $ Factor Period (index (index Average 1969= 100) 1969= 100 (6) (7) (8)

TABLE 8 Tanzania: Officia l Paralle l an d Real Exchange Rates, 1965-198 9 (Tsh/US$) .

149.41 141.65 122.30 109.93 99.99 102.93 112.91 109.79 111.26 112.28 109.24 111.10 116.67 116.31 121.50 105.59 80.21 69.15

0.67 0.71 0.82 0.91 1.00 0.97 0.89 0.91 0.88 0.89 1.03 0.89 0.82 0.83 0.82 0.95 1.25 1.49

n.a. 0.57 0.67 0.76 0.82 0.69 0.55 0.43 0.43 0.47 0.37 0.34 0.32 0.49 0.57 0.37 0.39 0.42 (Table continues on the following page.)

4.78 5.04 5.84 6.50 7.14 6.94 6.33 6.51 6.20 6.36 7.56 7.49 6.82 6.38 6.77 7.75 10.38 13.83

Real Exchange Nominal Ratio Rate* Equivalent Ratio (Index Exchange (10)1(2) (10)1(4) Period Period Rate 1969=100) (7)x(8) (Tsh/US$) Average Average (10) (12) (9) (11)

3 2 2 8 0 2 7

50.00 70.00 150.00 180.00 190.00 230.00 300.00

39.62 57.08 100.80 165.00 180.00 210.00 250.00 4.014 3.866 9.091 3.480 2.270 1.840 1.560 3.556 3.733 5.769 5.046 2.801 2.115 1.744

60.16 67.38 47.42 114.39 147.76 174.84 227.29 1.13 0.98 1.20 1.40 1.69 1.58 1.48

67.87 66.06 57.08 160.07 250.39 276.43 337.01

18.35 27.41 28.90 32.31 33.43 45.22 57.06

1.65 1.79 1.65 0.99 0.52 0.46 0.40

0.46 0.48 0.29 0.20 0.19 0.22 0.23

Other Real Real Exchange Currency Exchange Nominal Ratio Rate* Equivalent Ratio Ratio Rate* vis-a- Correct. Factor (Index Exchange (10)1(2) (10)1(4) (4)1(2) vis $ Rate Period Period (index 1969=100) Period (index (7)x(8) (Tsh/US$) Average Average Average 1969=100) 1969= 100) (10) (12) (8) (7) (9) (6) (11)

Sources: P . P . Cowit t (ed.) , World Currency Yearbook, International Currency Analysis (Brooklyn , N.Y. , 1985) ; Pick's Currency Yearbook (Pic k Publishing Corp , N.Y . variou s issues) ; IM F International Financia l Statistic s (variou s issues) ; Bagachw a e t al. , 1990 , "Tanzania : A Stud y o f Non Traditional Exports, " University o f Dar es Salaam , January, and authors' computations. a A decline implies appreciation .

1983 12.457 11.14 1984 18.105 15.29 1985 16.499 17.47 1986 51.719 32.69 1987 83.717 64.26 1988 125.000 99.29 1989 192.300 143.37

Official Official Parallel Parallel Ratio Rate Rate Rate Rate (3)1(1) End of Period End of Period End of Period Average Period Average Period (3) (4) (5) (1) (2 )

(continued)

TABLE 8

Macroeconomic Policy and Performance 3

5

In terms of official GD P growth, the performance sinc e 198 5 has been one of revival, wit h agriculture leadin g th e way. Tabl e 2 exhibits growt h rates of rea l GDP in the various sectors from which it is clear that since 198 5 all sectors, with the exceptio n o f minin g an d quarrying , hav e improve d o n thei r 1980-198 5 performance. Fo r 1989 , firs t estimate s indicate d a real GD P growth rat e o f 4. 4 percent, with all sectors growing, agriculture growing at 4.6 percent, and manufacturing a t 5.1 percent . Despite this growth, however, in 198 9 only agriculture, electricity an d water , commerce , finance , an d publi c administratio n achieve d real product levels higher than those of 1980 . On the external accoun t front , a s can be see n fro m Tabl e 7, th e target of 1 6 percent growt h in export earnings was not attained. Total export earnings fro m the six major agricultural exports were at their lowest level of the decade in 1989, at US$ 180. 4 million, afte r a brief reviva l i n 198 6 du e to a small coffe e boom , compared wit h US$ 26 4 million i n 198 0 and US$ 196. 1 millio n i n 1985 , at the depth of th e crisis. Among th e major export crops, only cotto n appears to have exhibited a stron g expor t volum e increase , wit h th e othe r crop s no t showin g increasing trends. This outcome might be related to the fact that very little price marketing liberalization i n agricultural export s took place until 1990 . Nontraditional exports, however, especially manufactured products, have staged a strong recovery. Fro m a continuous declin e betwee n 198 0 an d 1985 , from US $ 241. 5 million to US$ 90.5 million, this category has recovered to US$ 214.8 million in 1989. This recovery was no doubt aided by the generous export retention scheme instituted under the ERP. Imports, on the other hand, surpassed their 198 0 leve l in 1982 , and because o f dono r support as well a s the own-fund impor t scheme , they have stayed quite close to their targeted levels. On the inflation front, performance has not lived up to the targets. The changes in the NCPI for the years 1986 , 1987 , 1988 , and 198 9 were 32.4, 30.0, 31.2 , and 25.9 percent , respectively . Thi s i s n o differen t fro m th e performanc e durin g 1980-1985, whe n annual inflation range d between 26 an d 36 percent. The ambitious targets for industrial production als o did not materialize. Th e doubling o r tripling o f capacit y utilizatio n envisione d i n 198 6 implie d a corresponding increas e i n output . Bu t o f 3 1 industrie s whos e outpu t fo r th e perio d 1980-1988 is reported in the 1989 Bank of Tanzania's Economic and Operations Report, onl y 1 5 exhibited an increase in production between 198 5 and 1988, and of thes e onl y 1 0 experienced a total increas e o f mor e than 50 percen t ove r the three-year period, despite the fact that the bulk of imports during that time were intermediate good s and machinery. In Table 9 we summarize some of the recent sectoral and macro developments. The significan t growt h i n rea l GD P observed sinc e 198 3 i s du e mostl y t o th e growth in agricultural production, as has already been mentioned, while manu-

GDP real index 100. GDP agriculture total index 100. GDP manufacturing rea l index 100. GDP construction real index 100. GDP trade real index 100. GDP public administration real index 100. Gross fixed capital formation real index 100. GFCF public real index 100. GFCF private real index 100. GFCF buildings real index 100. GFCF other works real index 100. GFCF machinery and equipment real index 100. GDP deflators (1976= 100 ) GDP at factor cost total 100. Agriculture 100. Manufacturing 100. Construction 100. Trade 100. Public administration 100. Terms of trade (TOT) agriculture/manufacturing 100. TOT agriculture/construction 100. TOT agriculture/trade 100. TOT manufacturing/construction 100. Per capita real currency holdings in 1976 Tsh 144. Index of real per capita currency holdings 100. Currency/dem deposit ratio (percent) 63. Currency/TOT deposit ratio (percent) 42.

TABLE 9 Tanzania: Recen t Macroeconomic Developments . 5 5 1 6 5 0 7 2 4 0 7 3 7 0 4 4 0 9 3 4 0 2 5 9 5 0

118.2 128. 121.7 139. 124.5 141. 121.4 134. 122.5 139. 104.0 95. 97.7 98. 100.2 103. 99.3 100. 102.5 105. 144.3 160. 99.7 110. 59.4 74. 39.9 45. 0 0 0 0 0 0 0 0 0 0 7 0 5 5

5

100.4 102. 101.1 99. 94.0 97. 103.5 88. 98.0 98. 106.6 128. 110.9 110. 114.0 104. 106.7 119. 115.3 102. 96.7 79. 118.1 133.

7977 797

0 0 0 0 0 0 0 0 0 0 0 0

1976 2 1 4 4 0 1 8 2 0 1 0 4 9 6 7 7 0 2 7 9 4 0 8 4 8 7

105.5 108. 100.2 104. 100.4 95. 99.4 105. 100.0 100. 139.0 136. 124.1 108. 110.9 107. 141.7 111. 117.1 119. 88.5 97. 148.9 112. 141.4 159. 162.5 176. 137.1 152. 139.8 160. 153.0 166. 102.7 124. 118.5 115. 116.2 109. 106.2 106. 98.1 95. 192.4 185. 133.0 128. 63.6 64. 41.6 42.

7979 1980 3 6 0 2 0 8 3 9 2 3 8 3 2 4 3 3 4 2 0 0 4 5 2 1 3 7

107.6 108. 105.1 106. 84.7 82. 100.7 105. 96.0 94. 151.6 151. 112.5 117. 108.0 112. 118.7 123. 141.2 125. 76.2 91. 124.7 130. 188.4 224. 213.8 274. 189.0 189. 181.4 200. 201.1 255. 133.3 153. 113.2 145. 117.9 137. 106.4 107. 104.2 94. 181.4 165. 125.3 114. 75.3 77. 47.0 47.

7987 1982

115.5 127.8 70.8 79.8 104.2 137.7 135.8 94.1 191.9 105.3 62.3 193.1 563.0 728.2 429.5 444.1 658.4 314.4 169.5 164.0 110.6 96.7 110.9 76.6 104,6 57.1

112.1 120.8 73.8 68.0 93.8 154.4 140.0 98.1 196.0 95.7 48.8 213.3 445.2 560.2 321.2 342.9 533.2 296.9 174.4 163.3 105.0 93.7 104.9 72.5 101.3 48.5

330.3 400.5 274.8 251.7 395.7 242.7 145.8 159.1 101.2 109.2 118.4 81.8 104.1 53.0

273.6 330.2 231.5 228.1 311.9 208.1 142.6 144.8 105.9 101.5 129.6 89.6 66.2 39.1

1986

109.3 114.0 76.8 74.7 93.0 151.5 114.2 76.5 164.8 98.8 52.2 158.6

1985

105.7 109.6 74.8 62.1 92.0 151.3 78.3 66.0 95.0 84.3 45.8 96.5

1984

Source: Compute d from the national account s and Bank o f Tanzania yearbooks.

GDP real index GDP agriculture total index GDP manufacturing rea l index GDP construction real index GDP trade real index GDP public administration real index Gross fixed capital formation real index GFCF public real index GFCF private real index GFCF buildings real index GFCF other works real index GFCF machinery and equipment real index GDP deflators (1976= 100 ) GDP at factor cost total Agriculture Manufacturing Construction Trade Public administration Terms of trade (TOT ) agriculture/manufacturin g TOT agriculture/construction TOT agriculture/trade TOT manufacturing/constructio n Per capita real currency holdings in 197 6 Tsh Index of real per capita currency holdings Currency/dem deposit ratio (percent) Currency/TOT deposit ratio (percent)

1983

743.0 1002.3 712.9 480.9 834.3 398.0 140.6 208.4 120.1 148.2 111.2 76.9 108.7 58.6

120.0 133.4 73.8 81.6 109.6 138.5 134.3 66.6 225.1 137.9 49.2 186.2

1987

967.7 1,413.7 1,118.1 601.1 1,353.5 481.2 126.4 235.2 104.4 186.0 106.5 73.6 94.1 54.6

124.9 139.4 77.8 85.6 114.0 142.7 129.8 60.7 222.4 126.4 78.6 163.0

1988

1,242.0 1,571.0 1,320.3 719.0 1,585.9 651.9 119.0 218.5 99.1 183.6

130.6 145.7 81.8 92.9 119.0 147.0

1989

38 Alexander

H. Sarris and Rogier van den Brink

facturing output has stagnated. The output of the trade sector has not kept up with that of agriculture, while the output of public administration, after a brief decline in the first two years after the onset of the ERP, grew in 1989 at a rate close to its pre-ERP level. The mos t interestin g developmen t seem s t o b e occurrin g i n gros s fixe d capital formatio n (GFCF) . Th e privat e secto r GFC F has grow n ver y fas t sinc e 1983, whil e publi c secto r GFC F declined an d remained stagnan t a t it s alread y low, pre-ERP levels. Examining the implicit deflators for the various sectors, it can be noticed that the deflator fo r agriculture ha s grow n faster than all others , except th e one fo r trade. Notice the very smal l rise in the deflator for public administration, whic h reflects th e small rise in public secto r wages in the face of inflation . In terms of relative prices (see Figure 1) , it can be clearly see n that the terms of trad e of agricultur e afte r 198 3 (remember , thes e ar e official prices ) initiall y grew vis-a-vis manufacturing, but since 1986 , they have declined. The terms of trade o f agricultur e relativ e t o construction , a largely nontrade d activity , hav e FIGURE 1 Internal Term s of Trade. 9/in ,

o 1? 220-

/*

VO i> ON

ZX 200 o>

"2 £ 180 -

t

g

*• — ^ » 1

/

'/ \

*I

(D T3

g 140 -

/+

o

/ •

° 120 #o M

A* *•

-

80 1976

/

/

^^ t

| 160 -

£ IOO

i

/

/1

1978

*• /

*• ^ •" *>* ..^** 1980

1982

Agriculture/Manufacturing Manufacturing/Construction Source: Computed by authors.

\

*

*£~~' Y / / •*

/

01 #1

/

•^

1988 19 1984 198 6 Year Agriculture/Construction

$

Macroeconomic Policy and Performance 3

9

increased substantially . Thi s also obtain s fo r the terms of trad e between manufacturing (whic h is largely a n import substitute) and construction. It is not clear to wha t exten t thes e trend s ar e influence d b y th e us e o f officia l price s i n th e national accoun t statistics . In th e las t row s o f Tabl e 9 , th e development s i n rea l pe r capit a currenc y holding ar e examined . Afte r a n initia l ris e betwee n 197 8 an d 1981 , holding s declined significantly unti l 1985 , to levels much below those of 1976-1977 , and have remained relatively constant since then. The decline of per capita real cash balances canno t be attribute d t o th e expande d us e o f deman d or time deposits . In fact, as shown by the last two rows in Table 9, the ratio of currency held outside the banks to deman d or total deposit s betwee n 198 2 an d 198 5 increase d rather than declined , whic h woul d b e compatibl e wit h a substitutio n towar d ban k deposits during the years 1982-1985. This indicates the preference of the private sector for cash. Collier and Gunning (1989) used the developments in currency holdings as a key indicato r supportin g thei r theor y o f incom e declin e an d peasan t suppl y response unde r rationing and shortages during the period 1980-1985 . In Figure 2 we exhibit the indices of real per capita currency and M2 holdings in Tanzania from 1967 to 1989 (deflated by the NCPI). It is clear first that currency and M 2 moved i n uniso n throughou t th e period , especiall y durin g 1975-1985 . Also notice that there are two peaks in real currency holdings per capita, in 197 2 and 1979, the latter lasting until 1981. Finally, notice that real per capita currency holdings in the post-SAP period, 1986-1988 , are similar to those held during the "normal" period 1968-1969 . According to Collier and Gunning, the increase in cash holdings in the years 1978-1981 wa s due to shortages of consumer goods that made their availability random, whic h require d consumers t o hol d mor e cas h s o a s to be read y t o buy when good s becam e available . Afte r 1981 , shortage s becam e s o endemi c tha t cash needs declined and producers, especially peasants , reduced their marketed supply. This presumably was associated also with a decline in their real incomes. After 1985 , liberalization o f marketin g an d improved availabilit y o f consume r goods le d t o positiv e suppl y response , bu t rea l cash need s di d no t increas e significantly a s consume r good s wer e no w mor e readil y availabl e an d precautionary demand for cash declined. Collier an d Gunnin g use d th e apparen t declin e i n rea l pe r capit a currenc y holdings betwee n 198 2 an d 198 5 a s furthe r evidenc e tha t tota l rea l incom e declined durin g th e period . Ou r analysi s o f secon d econom y GDP , late r i n this chapter , however , suggest s tha t apar t from 1979 , 1982 , an d 198 3 (a s wel l as 1986 , whic h i s a post-SA P year) , rea l GD P grew. Betwee n 197 7 an d 198 5 official GD P grew annuall y a t a n averag e rat e o f 1. 4 percent , whil e tota l GDP

40 Alexander

H. Sarris and Rogier van den Brink

FIGURE 2 Real per Capita Currency an d M2 Holdings. 140

xrx

1

-r 1970 197

1990

5

Year Per capita currency

Per capita M2

Source: Computed by authors.

(official an d second-economy ) seem s t o hav e grow n somewha t faster , a t 2. 1 percent annually . The apparen t declin e i n real per capita currenc y holding s durin g th e 1982 1985 period could have occurred because of an increase in expected inflation, in addition to the shortages that were investigated by Collier and Gunning. In Figure 3, we plot the year-to-year percentage changes in the parallel exchange rate and the NCPI . It ca n b e notice d tha t durin g th e perio d 1980-198 5 bot h th e rat e o f inflation, a s measured by the rate of chang e of the NCPI, and the rate of chang e of th e paralle l exchang e rat e wer e a t highe r averag e level s tha n durin g an y previous four - o r five-year period . This , o f course , o n th e basi s o f traditiona l theory, should have led to a decrease in real per capita cash holdings, and indeed this appears to have happened. After 198 5 th e rate of inflatio n di d not change muc h from th e crisis period, so tha t thi s facto r i s no t expecte d t o hav e le d t o a chang e i n rea l pe r capit a cash holdings . T o th e exten t tha t bot h officia l an d a s tota l rea l GD P hav e increased a t rate s highe r tha n populatio n growth , th e deman d fo r rea l cas h balances should have increased somewhat. The lack of a perceptible increase has been attributed by Collier and Gunning to improved consumer good availability.

Macroeconomic Policy and Performance 4 1 FIGURE 3 Yearly Change s in NCPI and the Parallel Exchange Rate. 80

1975 198 Year Parallel exchange rate Source: Compute d by authors.

T

1990

0 NCPI

THE DEVALUATION DEBATE IN THE CONTEXT OF ADJUSTMENT One of the subjects of major debate in Tanzania in the early and mid-1980s wa s over currency depreciation as a tool of balance-of-payments adjustment . In fact, it was mostly over the reluctance of the government to devalue that negotiations with the IMF broke off several times. There is extensive literature on the reasons for the government of Tanzania's resistance to devaluation, even though it had accepted several other orthodox stabilization measures (e.g., van Arkadie 1983; Jamal 1986 ; Loxle y 1989 ; Ndul u 1988 ; Sing h 1986) . I n fact , almos t al l th e measures recommende d i n the report of th e mostly neutra l Tanzania Advisor y Group (TAG ) wer e implemente d i n th e firs t Structura l Adjustmen t Program , except the suggestion o f a mild devaluation . The basi c argument s o f thos e opposin g devaluatio n wer e tha t devaluatio n would no t lea d t o a n improvemen t i n th e balanc e o f payments , tha t i t woul d generate inflatio n an d hence tensio n ove r income shares , an d that it would no t lead to an improvement in the real exchange rate (the ratio of traded to nontraded goods prices) because of inflation, thus necessitating further devaluations and an inflationary spiral .

42 Alexander

H. Sards and Rogier van den Brink

The reason that the critics thought devaluation would not improve the balance of payments wa s elasticity pessimism . Suppl y response of expor t crops to price increases wa s though t t o b e ver y small , give n tha t man y suc h crop s wer e perennial ones . Smugglin g woul d not decline, accordin g t o the critics, because the main motivation for smuggling was the exchange with consumer and luxury goods not available in Tanzania and capital flight. O n the demand side, most of the rationed imports were intermediate and capital goods, and they were already quite compressed. Hence very little reduction of imports was possible. Inflatio n was seen as inevitable, as those whose incomes would be most affected (mainl y urban salary earners) would resist an d demand wage adjustments , thus creating an inflationar y spira l an d frustratin g effort s t o chang e relativ e prices . Th e conflicting claim s on income could generate a contraction along the lines of the well-known argument by Krugman and Taylor (1978). The basic counterproposal pu t forth by critics of devaluation was that, given the severe foreig n exchang e constraint , a n initial injectio n o f foreig n exchang e was a prerequisite for any recovery. Such an increase in foreign exchange inflow s would improve the availability of incentive consumer goods and motivate export crop suppl y respons e i n additio n t o mobilizin g idl e domesti c manufacturin g capacity. Thi s wa s als o on e o f th e suggestion s o f th e TAG . However, th e TAG believed that the currency had become so overvalued that some devaluation was necessary. Our calculations shown in Table 8 indeed indicate that devaluation of the order of 25-50 percen t was necessary i n 1981-1982 . The TAG believed that the argument s o f th e devaluatio n critic s wer e exaggerated . Th e inflationar y impact woul d no t b e tha t sever e becaus e officia l pric e controls ha d alread y broken down, and most prices, especially those of food, followed parallel market rates. Furthermore, the strength of urban workers was overestimated, an d those favoring devaluatio n argued was that lower-paid workers could be protected by increases i n the minimum wage. Perusing th e debate , i t appear s tha t wha t le d t o th e acrimoniou s argument s between the government and the IMF was not so much the principle of devaluation as the magnitude and pace of devaluation an d adjustment. The IMF argued for a shock treatment, with more than 100 percent devaluation, abolition of the subsidy on sembe (maiz e flour , a n urban staple ) an d hence a sevenfol d increas e i n it s price, substantial increases in producer prices of export crops, and liberalization of all price controls. Apart from the speed of adjustment, the other major concern of th e Tanzania n governmen t wa s abou t equitabl e sharin g o f th e burde n o f adjustment, somethin g the IMF was not so concerned with. The evidenc e o n th e pac e o f devaluatio n t o som e exten t bear s ou t the criti cisms of a sudden devaluation. Jamal (1986) computed the parity exchange rates (in Tsh/US$) tha t would have been required in the period 1979-198 4 t o restore

Macroeconomic Policy and Performance 4 3 real far m price s t o 1972-197 3 level s unde r the assumption s o f maintainin g o r abolishing farm input subsidies. We exhibit his estimates in Table 10 and contrast them wit h ou r own calculation s o f th e nominal equivalen t exchang e rat e fro m Table 8 , column 11 , which i s computed a s the rate that would have maintaine d a constant real exchange rate (based on relative CPls, as explained i n Appendix B) at the 196 9 level. It can be seen that under any of the scenarios suggested, the required devaluation is nowhere near the nearly 10 0 percent required by the IMF in 1980-1981 . It appears that the principal fram e o f referenc e fo r the IMF must have bee n th e paralle l marke t rate , whic h wa s indee d mor e tha n 10 0 percen t above the official rate throughout the period 1980-1984. To argue in those terms, however, presupposes an understanding of the behavior of the parallel economy, a topic whic h we examine later in this chapter. FINANCING OF PUBLIC AND EXTERNAL DEFICITS AND MONETARY ADJUSTMENTS One of the major objects of reform under the SAP and the ERP has been the public budget. Th e fisca l deficit—tha t is , th e amoun t tha t require s financing—ha s increased slightl y a s a share of GD P since 1985 , but it has staye d muc h belo w the very high levels o f the late 1970 s an d early 1980s . Table 11 presents the public financing requirements on a fiscal year basis from 1976 t o 198 9 an d th e method s o f financin g them . Throughou t th e period , th e

TABLE 10 Tanzania: Estimate s of Parit y Exchange Rates , 1979-198 4 (Tsh/US$) . Exchange Rate to Restore Real Farm Prices to 1972--1973 Levels* Official Nominal Exchange Rate* (period average) (1) 1979 1980 1981 1982 1983 1984 a b

8.217 8.197 8.284 9.283 11.143 15.292

With Subsidy (2)

Without Subsidy (3)

8.24 8.24 n.a. 13.23 13.23 15.51

8.24 8.24 n.a. 14.10 14.10 17.00

Derive d from Table 8. Derive d from Jamal (1986), Table 2.

Nominal Equivalent Exchange Rate* (4) 6.77 7.75 10.38 13.83 18.35 27.41

0.00 0.00 0.00 0.00 12.03 13.71 25.44 15.53 7.64 15.58 15.34 42.76 46.57 70.24

i(percentages) 41.97 87.41 53.91 49.60 46.04 42.63 71.18 53.94 31.34 49.10 36.40 76.08 82.22 105.95

57.78 12.59 46.54 49.86 52.65 57.77 27.68 45.79 68.62 50.64 63.67 24.12 17.81 -5.97

n.a. n.a. n.a. n.a. n.a. 36.24 64.98 71.24 44.69 23.60 50.73 8.85 9.32 -15.00

(3)1(1) (2+3)1(1) (4)1(1) (4-5)1(1) (10) (8) (9) (7)

Source: Compute d from data in Bank of Tanzania Economic and Operations Report, June 1989 and June 1982. Note: A year refers to a fiscal year starting the previous year (e.g., 1986 is the 1985/86 fiscal year).

n.a. n.a. n.a. n.a. n.a. 1,374 -1,499 -1,231 1,913 2,144 1,257 2,858 2,355 2,686 41.97 87.41 53.91 49.60 34.01 28.92 45.74 38.41 23.70 33.51 21.06 33.32 35.65 35.71

1,422 202 1,320 1,940 3,511 3,686 1,112 2,215 5,486 4,016 6,182 4,514 4,938 -1,774

l'million Tsh) 0 0 0 0 802 875 1,022 751 611 1,236 1,489 8,001 12,909 20,889

1976 1977 1978 1979 1980 1981 1982 1983 1984 1985 1986 1987 1988 1989

1,033 1,402 1,529 1,930 2,268 1,845 1,838 1,858 1,895 2,658 2,045 6,235 9,881 10,619

Fiscal Year

2,461 1,604 2,836 3,891 6,668 6,380 4,018 4,837 7,995 7,931 9,709 18,712 27,719 29,739

(2)1(1) (6)

Import Public Foreign Support and Financing Grants and Counterpart Domestic Nonbank Requirements Borrowing Borrowing Loans Funds (3) (2) (5) (4) (1)

TABLE 11 Tanzania: Financin g o f Publi c Deficit .

Macroeconomic Policy and Performance 4

5

major sourc e o f financ e ha s bee n foreig n grant s an d loans , a s wel l a s impor t support an d counterpar t fund s (column s 2 , 3 , an d 8) . Durin g th e crisi s year s 1980-1986, thi s shar e decline d considerably , an d the mai n metho d o f financ e became domestic borrowing (see columns 8 and 9), with bank borrowing becoming quit e importan t (se e colum n 10) . Sinc e 1986 , ban k borrowin g (tha t is , financing through money creation) appears to have declined dramatically, and in fact, i t appear s tha t i n 1988/8 9 som e o f th e publi c deb t hel d b y bank s wa s redeemed. If thi s picture truly reflected th e mode o f financing , the n the money suppl y (Mi or M2) should not have expanded very much since 1986 , in accordance with the ERP guidelines. However , a s shown i n Table 12 , the volume o f currenc y i n circulation and the money supply have grown by very large amounts since 1986 , much larger than what is justified b y reported domestic bank borrowing. The explanation of this apparent paradox, as Collier and Gunning (1991) have argued, i s tha t i n Tanzani a ther e i s no commercia l bankin g syste m i n th e traditional sense. The main function of the existing commercial banking system, which is basically compose d of only one bank, the National Bank of Commerc e (NBC), is to extend loans to the parastatals. In fact, 70-9 0 percent of all lending is to official entities . A large part of tha t is to cover current parastatal deficits , especially those of official agricultura l marketing authorities, the chief being the National Milling Corporation (NMC), and cannot be repaid. Such lending should, therefore, be treated very much like money creation . TABLE 12 Tanzania: Change s in Monetary Aggregate s durin g the ER P (absolute change s from previous year). Change in Currency Change in Circulation (M2)

1983 1984 1985 1986 1987 1988

Million Tsh (1)

Percent (2)

205 2,278 2,247 5,591 6,241 7,236

2.6 27.8 21.5 44.0 34.1 29.5

Million Tsh (3) 4,398 1,091 8,753 11,382 16,090 23,450

Percent (4)

Domestic Bank Borrowing Million Tsh (5)a

17.8 3.7 29.0 29.2 32.0 35.3

3,510 2,723 3,399 3,291 2,220 -939

in Money Supply

Source: Computed from Bank of Tanzania, Economic and Operations Report for the year ended June 30, 1989. a Compute d by averaging fiscal year figures.

46 Alexander

H. Sards and Rogier van den Brink

In Table 1 3 we exhibi t th e changes i n the domestic lendin g t o marketing o f agricultural produce (almost all going to parastatals). It can be clearly see n that between 198 2 and 1988 , 50 percent of the increase i n total commercial lendin g is accounted for by increases in lending to agricultural marketing parastatals. For 1985-1988, th e figur e i s 4 9 percent . I f w e conside r th e change s i n th e mone y supply (M2 ) fro m 198 5 t o 198 8 an d total commercia l ban k lendin g t o officia l entities over the sam e period, then the latter makes up 80.1 percen t of the total change i n M2 . It therefor e appear s tha t indee d th e chang e i n domesti c mone y supply i s closely associate d with bank credit to official, an d especially agricul tural marketing, parastatals. This obviously makes agricultural marketing reform an item with important macroeconomic consequences . Table 1 4 shows th e modes o f financing o f the external defici t durin g 1983— 1988. I t is obvious that the accumulation o f arrear s and debt rescheduling hav e been the main methods of financing a growing curren t account deficit sinc e the onset of the ERP. The current account deficit has grown from -5.4 percent of GDP at market prices i n 1985 , to -12.0 percen t i n 1988 , an d capital flow s hav e no t ameliorated the situation by much. It must be recalled that own-funded import s are counterbalanced by an equivalent current transfer item on the credit side, so that they do not affect the overall current account balance. The figures show the unsustainability o f the external deficit an d the reliance on debt rescheduling. TABLE 13 Tanzania: Contributio n of Lending fo r Agricultural Marketin g t o Total Domestic Lendin g Change s (i n million Tsh). Change in Total Change in Total Domestic Lending by Lending to Contribution of Commercial Banks Marketing of Agricultural from Previous Year Agricultural Produce Marketing (1) (2 ) (2)/(l ) x 10 0 1983 1984 1985 1986 1987 1988 Change 1982-1988 Change 1985-1988

1,064 83 2,312 1,05 4,574 2,02 10,323 -4,63 27,220 30,07 18,386 2,10

4 78. 4 45. 9 44. 9 -44. 1 110. 3 11.

4 6 4 9 5 4

62,815 31,45

2 50.

1

55,779 27,21

0 48.

8

Source: Bank of Tanzania, Economic and Operation Report for the year ended June 30, 1989.

713 0 -1

-333 -168

-211 -4.8 -0.3

-664

675

3,394 2,038 1,134

0 0

6,967

0 0

2,800

175

-248

-6,894 -5.4 -5.7

-6,500 -323 -559 1,048 -560

1985

47

-421

-2,426 -6.2 -2.7

1,768

751

-5,491 -1,133 1,679

1984

Source: Bank of Tanzania, Economic and Operations Report for the year ended June 30, 1989.

Reserve decrease (-increase ) Arrears (+ increase) Debt rescheduling Others

IMF (net )

Financing

As a percent of GDPMP As a percent of GDPMP

Overall balance

Current account Capital account (MLT ) net Supplier's credit (net) Imprt. sup. and exc. financ e Errors and omissions

1983

TABLE 14 Tanzania: Financin g o f the Current Account Defici t (i n million Tsh, unless noted) .

3 8 9 0 1

-27,133 -12.0 -8.7

9,998 -4,217

477

1988 -37,323 3,932

6 -5,56 3 4,44 4 15,37 1 12,87 8-

-18,058 -13.1 -8.2

2,744 7,982

321

1987 -28,655 -450

429 2,50 -816 -64 -27,635 -6 37,479 12,08 3,114 4,17

-12,571 -6.7 -7.9

-10,633 -688 -1,816 2,720 -2,154

1986

48 Alexander

H. Sarris and Rogier van den Brink

THE HIDDEN ECONOMY IN TANZANIA One of the major problems facing th e analyst of the Tanzanian economy i s the fact tha t th e number s tha t presumabl y describ e th e evolutio n o f aggregat e variables suffe r fro m inaccuracie s du e t o incomplet e coverage , a s wel l a s inaccurate estimates of the activities covered. This is particularly so in agriculture, mining, manufacturing, construction , an d trade. In Tanzania , becaus e o f th e post-196 7 emphasi s o n developmen t throug h public ownership and control of the major enterprises producing and distributing goods an d services an d the subsequen t nationalizations , a large segmen t o f th e formerly activ e private sector either stopped producing or went underground, in the sense that it kept producing but with higher costs due to the cost of evadin g the variou s governmen t controls . Give n th e officia l policy , however , thes e activities wer e no t officiall y recognized , an d henc e no effor t wa s mad e t o estimate them . Th e resultin g "hidden " o r "second " o r "parallel " o r "under ground" economy i n Tanzania (the terms will be used interchangeably below) i s thought t o hav e graduall y grow n a s publi c control s becam e mor e bindin g an d especially a s shortages of various goods became widespread with the post-1979 foreign exchang e crisis . A questio n o f majo r macroeconomi c relevanc e i s whether th e hidde n econom y ha s followe d th e fluctuation s o f th e officiall y observed economy, and if not, whether it has tended to compensate for accentuate the observed secular decline in economic activity . In the past decade, there has been an active literature concerned with estimates of th e undergroun d econom y i n develope d countrie s (se e Tanz i [1982 ] fo r a n early surve y an d Bhattacharyy a [1990 ] fo r a recen t analysis) . Thi s i s usuall y defined a s tha t econom y whic h i s no t measure d b y officia l GD P statistics. I t covers lega l bu t unreporte d o r unmeasure d activities , suc h a s thos e o f man y small-scale enterprises, usually known as the "informal sector," as well as illegal activities, suc h a s productio n an d smugglin g o f officia l expor t crops , illega l mining, hunting , an d s o forth . I n Tanzania , give n th e lega l monopol y ove r marketing an d distributio n o f agricultura l crop s an d th e lega l monopol y o f parastatal production of many consumer and intermediate items, parallel markets quickly arose. The problem has been analyzed extensively by Maliyamkono and Bagachwa (1990) , who attempted several techniques in an effort t o measure the size an d evolution o f th e secon d economy. Ther e are four major methods use d for estimating the second economy: those based on differences betwee n survey based incomes and expenditures; those based on labor participation; those based on monetary estimates ; and those based on analyzing tax returns (see Fre y and Pommerehne 1982) . Maliyamkono and Bagachwa (MB) used the first two in their

Macroeconomic Policy and Performance 4

9

analysis of Tanzania. In this report we shall review their results and present some new estimates. In their first approach, MB used one of the simplest available methods, based on th e assumptio n tha t th e currency-to-demand-deposi t rati o stay s unchange d over time (Guttman 1977) . On the basis of that assumption and using 197 7 a s a base year (a year in which the second economy wa s assumed to be negligible) , they foun d tha t a s a shar e o f officia l GD P the secon d econom y gre w fro m 9. 8 percent i n 197 8 t o betwee n 2 2 an d 2 9 percen t i n th e perio d 1980-198 6 an d reached 31.4 percent in 1986 . However, if one uses the same technique and base year to compute the second economy in years before 1977 , one finds that between 1970 and 197 2 the unofficial econom y wa s between 2 0 an d 30 percent o f GDP, while between 197 3 and 1979 the second economy wa s smaller than 1 0 percent of GDP , whic h seem s unlikel y i n vie w o f th e fac t tha t controls wer e quit e extensive durin g that period. Th e problems with this technique, a s well a s any other tha t assume s fixe d monetar y ratios , ar e wel l know n an d ar e readil y acknowledged b y MB . A preferable metho d i s on e based on the estimation o f a demand-for-currency equatio n (Tanzi 1983 ; Bhattacharyya 1990) . However, MB report that they were unable to estimate suc h an equation for Tanzania. MB also reported the results of a household surve y i n 1986 , which measure d expenditures an d income s fo r a sampl e o f urba n an d rura l households . B y blowing u p thei r expenditur e figures t o th e nationa l level , the y foun d tha t officially reported , privat e fina l consumptio n wa s underestimate d i n 198 6 b y about 3 0 percent . Give n tha t the aggregat e margina l propensit y t o consum e i s around 0.9 3 i n Tanzani a (Lipumb a e t al . 1988) , thi s implie s a 3 3 percen t underestimate o f GDP. One major aspect of the second economy is parallel exports. There is a large, unrecorded set of activities that generate domestic product and income and which result i n exports tha t are not recorded. Th e large overvaluation o f th e currency in th e 1970 s an d 1980s , combine d wit h th e wid e borde r wit h countrie s wher e Tanzanian good s coul d b e exchanged , mad e th e contro l o f export s almos t a n impossible task . MB , as well a s Bagachwa, Luvanga , and Mjema (1990) , report that there are several categories of illegal exports in Tanzania, including agricultural products (maize, wheat, beans, goats, sheep), cattle, traditional export crops (coffee, cardamon , cotton), hides and skins, products of hunting (ivory), mining products (gold , diamonds , othe r precious stones) , tourism (vi a parallel cashin g of foreig n exchange) , housin g service s provide d t o foreigners , an d over - an d under invoicin g o f officia l export s an d imports . Good s obtaine d i n exchang e include basi c consume r goods , whic h wer e i n shor t suppl y i n th e 1970 s an d 1980s, luxury consumer goods, and some intermediate goods.

50 Alexander

H. Sarris and Rogier van den Brink

A glimps e o f th e siz e o f paralle l export s ca n be obtaine d b y examinin g th e figures fo r "own-funde d imports, " whic h starte d i n 1984 . Impor t unde r thi s category are funded with the importers' own foreign exchange. It was thought in the early years of the scheme that such imports would exhaust much of the foreign exchange that residents had accumulated through the parallel market. However, own-funded import s not only di d not slo w down , but in fact increase d substan tially, afte r 1986 . Th e conclusio n follow s tha t th e foreig n exchang e fo r thes e imports must have come from a flow o f unrecorde d export earnings. Below w e utilize this idea and a methodology based on the income approach to balance-ofpayment adjustment to estimate the missing GDP that is consistent with the large inflow o f own-funded imports. The methodology, whic h w e shal l ter m the "missing income " methodology, starts by assumin g tha t second-econom y GD P i s no t supply-constrained a s total economy GD P or official GD P would be (Ndul u 1988 ; Lipumba et al. 1988) , but instead adjust s t o satisf y deman d fo r it , muc h a s i n th e Keynesia n economi c model. Ther e i s no governmen t interferenc e i n it sinc e by definitio n i t escape s public control. A simple ex-post equation that states the macroeconomic balance between unofficia l GD P (denote d b y Y u) an d th e expenditur e o n i t i s th e following: (3.1) where C u is unofficial, unrecorde d consumption of the product generated by the second economy ; E u i s unofficial , unrecorde d exports ; an d M u i s unofficial , unrecorded imports. We mak e the traditiona l assumption s tha t consumption an d imports depen d on generated income . (3.2) (3.3) Hence (3.1) can be written, (3.4) Taking the total differential o f (3.4) , we obtain: (3.5)

Macroeconomic Policy and Performance 5

1

where a i s the marginal propensity to consume and |i is the marginal propensity to import out of income . I f we ca n estimate value s fo r the parameters a an d n, then by utilizin g som e estimate s fo r the siz e o f parallel export s w e coul d infe r the size of unofficia l GD P (Yu) required to support this level of imports. To implemen t ou r method, w e utiliz e estimate s o f th e incom e elasticit y o f aggregate privat e consumptio n expenditure s an d the income elasticitie s o f im ports of consume r an d intermediate good s arrive d a t by Lipumb a e t al . (1988 ) using data for the officially reporte d aggregate series from 196 8 to 1984 . To use these estimates , w e mus t assum e tha t th e sam e parameter s ar e als o vali d fo r second-economy transactions , which we do. The reported values i n Lipumba et al. (1988) are the following :

where ec is the income elasticit y o f privat e consumption expenditures , an d e\fc and e\fi are the elasticities with respect to officia l GD P of import s of consume r and intermediate goods respectively . The next step is to combine the two import elasticities to form an elasticity of aggregate , second-econom y imports . W e assum e tha t import s tha t cam e through clandestin e route s wer e mainl y consume r an d intermediat e goods . We then utilize the breakdown of no-payment imports for the period July 198 4 to December 198 5 reported by Ndulu and Hyuha (1986) an d estimate that the share o f intermediat e import s (includin g spare s an d buildin g materials ) i n total import s o f intermediat e an d consumer good s i s 0.40 , leavin g a share o f 0.60 fo r consume r goods . These , alon g wit h th e abov e estimate s o f elastici ties, imply a n elasticity o f parallel import s with respect to parallel incom e o f 0.64. To transform the estimated elasticities to marginal propensities, we utilize the average propensity t o consume an d the average propensit y t o import compute d from officia l dat a o f th e perio d 1976-1983 . Thi s perio d include s bot h th e euphoria of 1976-197 8 an d the post-1980 crisi s years. The resulting estimate s are, fo r th e margina l propensit y t o consume , a = 0.6 8 an d fo r th e margina l propensity to import, P = 0.13, resulting in a value of 0.45 fo r the denominato r of (3.5) . The fina l ste p involve s estimatin g paralle l exports . Fo r lac k o f an y bette r numbers, we utilize the figures for the dollar value of own-funded import s from 1985 t o 198 8 reporte d b y Bagachwa , Luvanga , an d Mjem a (1990) . The y als o report figures for 1984 , but the program operated for only hal f of that year, and the agents were still learning; hence that year was dropped. The assumption that

52 Alexander

H. Sarris and Rogier van den Brink

the size of parallel, unrecorded exports is close to that of own-funded import s can be justified a s follows: o n the one hand, the figures woul d overestimat e parallel export s becaus e the y als o includ e drawdow n fro m accumulate d for eign exchange holdings abroad . O n the other hand, however, they coul d ver y well underestimat e th e size o f paralle l export s becaus e despit e th e indirect "legalization" of parallel exports , ther e mus t stil l be a substantial amoun t of products that are exported and bartered directly for unrecorded import goods . This mus t b e s o particularly fo r parallel agricultura l export s i n the border areas. In any case the estimates are only intende d to give ballpark indication s of magnitudes . In Table 1 5 we indicate the results of our calculations, whic h giv e a picture of a huge second economy (almost equal in size to the official GDP ) when parallel rates are used to translate own foreign exchang e to Tsh (row 8). The size of the second economy ranges between 1 6 and 45 percent of official GD P when officia l rates are used (ro w 7). It is obvious tha t the Tsh equivalent of parallel export s and own-funde d import s shoul d b e computed usin g paralle l foreig n exchang e rates. However, remember that the prices used to compute current market value of GDP are still mostly official prices , which for many products do not represent

TABLE 15 Tanzania: Estimate s of the Second Economy by the Missing Income Approach.

(1) Officia l GD P at current market prices (2) Own-funde d import s (million US$) a (3) Own-funde d import s in Tsh at official rate s (4) Own-funde d import s in Tsh at parallel rates (5) Implie d second-economy GD P from row (3) b (6) Implie d second-economy GDP from row (4) b (7) ro w (5) as a share of row (1) (%) (8) ro w (6) as a share of row (1) (%) (9) ro w (6) as a share of adjusted GDP C

1985

1986

1987

1988

120.6 505.0

159.6 476.0

219.0 514.0

311.5 638.0

8.8

15.6

33.0

63.3

50.9

78.5

92.5

134.0

19.6

34.7

73.3

140.7

113.1 16.3 93.8 60.7

174.4 21.7 109.3 67.6

205.6 33.5 93.9 58.9

297.8 45.2 95.6 59.6

Note: Figure s are in billion Tsh unless otherwise noted. a Source: Bagachwa, Luvanga, and Mjema (1990). b Usin g equation 3.1 (see text). c Se e text.

Macroeconomic Policy and Performance 5 3 actual market prices. The degree of underestimat e o f current-marke t GD P can be inferred by comparing the implied GDP deflator fro m official statistic s and the NCPI. Such a comparison reveals that the NCPI has grown much faster than the GD P deflator, startin g aroun d 1980 , whic h i s logica l give n tha t the crisi s started then , and official price s starte d losin g thei r significanc e a t that time . In fact, th e ratio o f th e NCP I to th e implie d GD P deflator, whic h wa s clos e t o one before 1980 , increased to 1.5 6 by 198 4 and has hovered between 1.5 5 an d 1.62 sinc e then . If we use the ratio of NCPI to the implied GDP deflator to augment the reported current value GDP in Table 1 5 and also use the Tsh estimates of unofficial export s translated at parallel rates, then we arrive at figures in the last row of the table. They indicate tha t th e secon d economy , a s reveale d b y unofficia l foreig n exchang e outflows, ranges between 59 and 68 percent of adjusted nominal GDP. This is clearly a very high level and much higher than the estimates reported by MB. Another estimate o f th e unobserved economy ha s been obtained recentl y b y Bagachwa and Naho (1990a) in a paper prepared for this project. They estimated by Ordinary Least Squares (OLS) a demand-for-currency equatio n as follows :

(3.6) where the period of estimation is 1967 to 1988, the figures under the coefficient s are t-ratios , an d th e variable s ar e a s follows : CUR is rea l currenc y holding s outside banks, namely, nominal currency holdings divided by the NCPI; YOFR i s real nominal incom e deflate d by the GDP deflator; RPFC i s ratio of private fina l consumption expenditure to total expenditure on GDP; NBCB i s index of number of National Ban k of Commerc e branches; INFL i s GDP deflator; GINT i s ratio of parastatal employee s ove r total employees , a measure o f governmen t interven tion; ATR i s average tax rate, equal to the ratio of the sum of income and corporate taxes to the sum of compensation of employees an d operating surplus , and PEXP is ratio of parallel exchange rate to the official one . The estimated equation appears quite robust, and all the coefficients hav e the correct signs and are significant. B y using the standard technique develope d by

54 Alexander

H. Sards and Rogier van den Brink

Tanzi (1983) , they estimate the size of the second economy, first, by estimatin g through the above regression total currency holdings as predicted by (3.6 ) wit h all variables. Then "legal" currency holdings are estimated from (3.6) by omit ting the "interference" variables GINT, ATR , and PEKP. Th e difference give s th e amount o f currenc y hel d fo r unofficial-econom y transactions . B y multiplyin g this by th e observed-transactions velocit y o f money (obtaine d by dividing offi cial nomina l GD P by Mi , whic h i s th e su m o f currenc y an d demand deposits) , they obtain an estimate of the second economy GDP. Bagachwa and Naho's results are shown in Table 16 . Their estimates show a very larg e siz e fo r th e secon d economy , whic h starte d fro m a modes t 20-2 5 percent o f nomina l GD P in th e lat e 1960s , expande d t o aroun d 40-50 percen t during th e 1970 s an d early 1980s , an d has grow n t o mor e tha n 6 0 percen t o f nominal GD P after 1984 . Their estimates fo r this latter period fal l betwee n ou r low an d high values indicated in Table 15 . The last two columns in Table 1 6 show the real value of the second economy, obtained by dividing the estimated second economy GDP by the NCPI, and its share of rea l GDP. The results indicat e tha t the second econom y gre w ver y fas t in the early 1970 s an d tha t fro m 197 7 t o now , i t ha s fluctuate d betwee n 3 5 an d 5 2 percent of real official GDP , with peaks in 1978 and 1985. In absolute real terms, the second economy doe s not seem to have grown since the onset of reforms i n 1986. On a per capita basis, in fact, the absolute size of the second economy has declined, from peaks of 683 Tsh in 197 2 and 677 Ts h in 197 8 (i n 197 6 prices), to 598 Ts h in 198 5 an d 502 Tsh in 1988 . Table 1 7 present s th e result s o f Tabl e 1 6 fro m anothe r viewpoint . Wha t i s reported i s th e total rea l GDP—tha t is, the sum o f th e officia l rea l GDP and the second-economy real GDP—the yearly changes in total and official rea l GDP, and the siz e o f th e secon d econom y a s a share o f th e total. I t i s interestin g t o not e that a very differen t pictur e o f tota l GD P is obtaine d fro m thi s table . Year s o f decline in total real GDP appear to be 1973 , 1979,1982 , 1983 , and 1986. Growth appears to have been quite strong in 1984, the first year of liberalization, and has continued to be strong since then. The first year of the ERP, 1986 , appears to have been marked by a fall i n total real GDP, albeit the official figures indicat e a rise. Since 1987 , however , growt h appear s t o hav e occurre d a t rate s faste r tha n officially reported . Of interes t als o i s th e las t column , whic h indicate s th e shar e o f th e secon d economy i n the total (officia l plu s unobserved) GDP. It appears that with minor fluctuations, tha t share has stayed remarkably constant at around 28-32 percen t throughout th e las t tw o decades . I t must , o f course , b e note d tha t al l thes e conclusions depen d strongl y o n the metho d o f estimatin g th e second-econom y GDP.

Source: Bagachw a and Naho (1990a). a Colum n (6) divide d by the NCPI (1977=100).

1967 1968 1969 1970 1971 1972 1973 1974 1975 1976 1977 1978 1979 1980 1981 1982 1983 1984 1985 1986 1987 1988

Year

Second Economy Real Size of Real Second Income as Percentage of Second Economy as a Actual Legal Total Illegal Velocity Second the Official Economy* Percentage of Currency Currency Currency Currency of Money Economy Nominal GDP (base 1976) Real Official GDP (9) (2) (6) (3) (5) (7) (4) (8) (1) 512.0 1,440.5 27.7 305.8 21.4 3,996 516.7 210.8 6.83 528.6 267.6 508.9 241.4 6.93 1,672.4 23.3 4,017 26.5 605.0 269.0 610.9 341.9 5.65 1,932.2 25.9 3,971 25.7 818.4 322.9 799.6 476.7 6.94 3,309.2 40.3 6,633 40.5 986.6 348.9 972.7 623.8 6.23 3,888.6 43.9 7,445 43.7 1,201.1 421.3 1,251.0 831.6 6.49 5,393.5 53.8 9,588 52.7 1,198.6 415.7 1,186.9 771.2 5.77 4,448.7 38.7 7,169 38.2 1,517.3 450.4 1,469.9 1,019.5 5.86 5,977.7 42.7 8,078 42.0 1,755.8 527.1 1,928.1 1,401.0 5.56 7,790.5 45.9 8,320 40.9 2,071.3 548.6 2,096.1 1,547.5 5.68 8,796.6 40.6 8,797 40.6 2,379.7 619.9 2,507.4 1,887.6 5.56 10,492.4 40.8 9,401 43.2 2,915.2 800.2 3,064.9 2,264.7 6.06 13,737.2 48.1 11,546 52.0 4,055.5 871.1 3,828.1 2,951.0 4.46 13,178.9 40.8 9,816 43.0 5,245.4 1,109.2 5,282.8 4,173.6 4.07 16,973.1 45.3 9,705 41.4 6,616.0 1,240.8 6,393.8 5,153.0 4.38 22,566.2 51.4 11,634 49.9 7,988.7 1,408.8 7,135.2 5,726.4 4.47 25,623.1 48.8 9,042 38.6 8,194.2 1,742.7 8,336.0 6,593.3 4.44 29,249.6 46.7 8,124 35.5 10,472.4 2,064.9 10,570.9 8,501.0 6.44 54,798.6 70.1 11,179 47.3 12,719.0 2,755.0 14,210.7 11,455.7 7.06 80,892.7 74.8 12,381 51.0 18,309.7 3,727.2 16,741.0 13,021.8 6.63 86,370.0 61.3 9,983 39.9 24,550.8 5,431.3 22,921.5 17,490.2 6.89 120,490.2 62.4 10,716 41.3 31,786.7 8,211.6 33,951.6 25,740.1 6.48 166,880.1 61.4 11,314 41.8

Estimated

TABLE 16 Tanzania: Estimate s of Secon d Economy GD P 1967-1988 (i n million Tshs).

56 Alexander

H. Sarris and Rogier van den Brink

TABLE 17 Tanzania: Siz e of Secon d Economy (i n million 197 6 Tsh).

Second Official Economy GDP Real GDP (2) (1) 1967 1968 1969 1970 1971 1972 1973 1974 1975 1976 1977 1978 1979 1980 1981 1982 1983 1984 1985 1986 1987 1988

14,438 15,186 15,465 16,362 7,046 18,192 18,748 19,217 20,352 21,652 21,739 22,202 22,849 23,419 23,301 23,439 22,882 23,656 24,278 25,008 25,972 27,039

3,996 4,017 3,971 6,633 7,445 9,588 7,169 8,078 8,320 8,797 9,401 11,546 9,816 9,705 11,634 9,042 8,124 11,179 12,381 9,983 10,716 11,314

Total Real GDP

(l)+(2) (3) 18,434 19,203 19,436 22,995 24,491 27,780 25,917 27,295 28,672 30,449 31,140 33,748 32,665 33,124 34,935 32,481 31,006 34,835 36,659 34,991 36,688 38,353

Percent Yearly Change of Total GDP (4) 4.17 1.21 18.31 6.51 13.43 -6.71 5.32 5.04 6.20 2.27 8.38 -3.21 1.41 5.47 -7.02 -4.54 12.35 5.24 -4.55 4.85 4.54

Percent Yearly Change of Official Real GDP (5)

Second Economy GDP as Percentage of Total GDP (6)

5.18 1.84 5.80 4.18 6.72 3.06 2.50 5.91 6.39 0.40 2.13 2.91 2.49 -0.50 0.59 -2.38 3.38 2.63 3.01 3.85 4.11

21.68 20.92 20.43 28.85 30.40 34.51 27.66 29.60 29.02 28.89 30.19 34.21 30.05 29.30 33.30 27.84 26.20 32.09 33.77 28.53 29.21 29.50

Source: Computed by authors.

CONCLUSIONS As must have become clear in the course of this chapter, economic policy during the las t decad e i n Tanzani a ha s bee n characterize d b y man y controversies . Those range d fro m th e diagnosi s o f cause s o f th e crisi s t o th e appropriat e policies t o follow . Fro m our examination o f recen t developments , however , i t became clea r that the Tanzanian governmen t underestimate d th e magnitude o f the crisis , a s wel l a s it s dependenc e o n externa l support . I t als o appear s tha t despite th e nominal adoptio n o f a massive adjustmen t program , actua l polic y changes have been implemented very slowly, and many distortions still prevail.

Macroeconomic Policy and Performance 5

7

Examples ar e th e influenc e o n th e mone y suppl y o f domesti c financin g o f inefficient parastatals , an d th e stil l larg e differenc e betwee n th e paralle l an d official exchang e rates. Our analysis als o raise d the issu e tha t official statistic s o n economic devel opments do no t represen t th e tru e underlyin g trend s i n the rea l econom y ver y well. Ou r estimates o f th e unobserved o r second econom y reveale d that a very significant par t of the economy goe s unrecorded. When this part is incorporated in the officia l figures, i t appear s that, with the exception o f a few crisi s years , the economy ha s fared much better than had been thought. If this conclusion i s corroborated b y other , especiall y micro , evidence, i t suggest s tha t the crisis i n the Tanzanian economy wa s for the most part a crisis of the formal part and not necessarily th e whole economy .

I 4 PROFILE OF INCOMES AND POVERTY IN TANZANIA

^The mos t recen t Worl d Ban k World Development Report (1990 ) rank s Tanzania as the fourth-poorest country of the world, with a 1988 per capita income o f onl y US $ 160 . Yet i n 1980 , a t the end of a decade of shock s and economic decline, the International Labour Office (IL O 1982) estimated that in the urban areas only abou t 1 5 percent o f household s migh t be considered as falling belo w a poverty line. In the rural areas, the estimate was about 25-30 percent, for a countrywide tota l of abou t 25 percent. Thi s is not a very high estimate, compared to other developing country poverty levels. In this chapter, we sho w tha t poverty i s substantiall y greate r than previously estimated . We present a more up-to-date profile of poverty in Tanzania mainland, emphasizing its structural aspects. Recent household surveys on which to base analysis of patterns of income and consumption ar e not available i n Tanzania. However , there is a very detaile d national household survey that was done in 1976/77, the results of which became available only recently. It is very helpful i n giving the structure of Tanzanian households. This is especially so because there was another such survey in 1969, with which the 1976/7 7 survey can be compared. Furthermore, there exist two detailed rural income survey s (Collier et al. 1986 ; Bevan et al. 1989 ) done in 1979/80 and 1983 , respectively, whic h can also be used for comparative pur58

Profile of Incomes and Poverty in Tanzania 5

9

poses, despite the fact that they were not representative of the entire country but only certain regions. These four sources provide the bulk of the information fo r our analysis. The nex t sectio n outline s som e basi c demographi c characteristic s o f th e Tanzanian household. In the following sectio n the structure and sources of rural and urban incomes are analyzed. We also analyze income differentiation an d we show that income in Tanzania is quite equitably distributed, compared with other countries. I n the following section , w e discus s th e observed patter n o f incom e differentiation. Subsequentl y w e analyze patterns of consumption among differ ent types of households, and we show that most of the food intake of the poor in both rural and urban areas comes out of subsistence production. We compute an absolute poverty line an d estimate tha t a very substantia l numbe r of Tanzania n households appear to have expenditures below it. (We also estimate poverty lines for every yea r until 198 9 an d show that they amounte d to less tha n even senio r civil servan t salarie s i n 1989. ) I n th e fina l sectio n w e summariz e ou r mai n conclusions. POPULATION AND SOME HOUSEHOLD CHARACTERISTICS According to the 198 8 population census, the population of mainland Tanzania was 22,533.8 thousan d people, and the average household siz e wa s 5.3, imply ing that there were about 4,252 thousand households. The 1978 census revealed a populatio n o f 17,036. 5 thousand , compose d o f 3,44 2 thousan d households , with an average of 4.9 members. The 196 7 census had revealed a population of 11,958.7 thousan d an d a n averag e househol d siz e o f 4.5 . Whil e th e averag e intercensal populatio n growth rate between 197 8 an d 198 8 has declined t o 2. 8 percent from 3.2 percent between 1967 and 1978, the average size of households seems t o hav e increased . I t i s no t clea r whethe r thi s tren d i s rea l o r due t o a different definitio n o f the household i n the different censu s years. Most o f th e Tanzanian population live s i n rural areas. Table 1 8 exhibits th e number of household s reporte d as rural and as urban in the 197 8 censu s an d in the 1976/7 7 Househol d Budge t Surve y (HBS) , b y region . Th e 198 8 censu s preliminary report does not distinguish between rural and urban households. The 1978 census reports 13 percent more households for the mainland, compared with the 1976/7 7 HBS, the overreporting being simila r for the rural and urban categories. Th e 197 8 census , however, reporte d a n average househol d siz e o f 4. 9 fo r the whole country, while in the HBS it was 5.65. The result is that the HBS implied a tota l populatio n i n 1976/7 7 o f abou t 17,15 4 thousand , whic h i s 0. 7 percen t above the census figure of one year later. While the error in total population does not appea r t o b e ver y large , i t seem s tha t th e HB S use d a slightl y differen t

2,917.91 157.00 111.45 18.39 186.43 184.03 105.41 152.93 106.39 106.10 196.72 173.61 156.46 206.81 86.50 99.68 213.53 119.49 137.29 181.90 217.80

517.54 18.72 8.07 186.50 16.98 18.25 12.23 16.59 11.69 10.73 22.09 28.52 21.82 34.15 11.36 8.95 12.84 12.88 24.48 34.24 8.46 3,435.45 175.72 119.52 204.89 203.42 202.27 117.64 169.51 118.07 116.83 218.81 202.13 178.28 240.96 97.87 108.62 226.37 132.37 161.76 216.14 226.26 2,585.05 117.34 99.45 0.00 149.10 157.28 95.11 148.78 112.97 109.81 176.02 151.05 166.37 209.66 64.98 101.85 189.25 95.17 133.92 146.21 160.74

453.70 22.46 0.00 165.21 32.34 10.88 8.05 16.96 6.28 9.74 10.81 12.93 11.22 31.54 8.92 7.96 14.76 9.93 15.19 33.85 22.66

3,038.75 139.80 99.45 165.21 181.45 168.17 103.16 165.74 119.24 119.55 186.83 163.98 177.59 241.19 73.89 109.81 204.00 105.10 149.11 180.06 183.40

4.90 5.30 4.30 4.10 4.70 4.50 5.50 5.30 4.40 6.20 5.00 4.70 4.30 6.00 5.10 5.20 5.80 4.60 5.00 4.70 4.50

5.30 5.40 4.90 4.30 5.00 4.80 5.80 5.40 4.60 6.70 4.90 5.30 4.40 6.40 5.30 5.30 6.30 5.30 5.70 5.10 4.90

5.65 5.33 5.65 5.02 5.13 5.32 6.06 5.91 4.90 6.67 5.87 5.53 4.97 6.75 6.05 4.89 6.57 5.80 5.52 5.65 5.00

Source: Tanzania , 7975 Population Census Preliminary Report, Tanzania Bureau of Statistics, Household Budget Survey, 1976/77 .

Tanzania Arusha Coast Dar es Salaam Dodoma Iringa Kigoma Kilimanjaro Lindi Mara Mbeya Morogoro Mtwara Mwanza Rukwa Ruvuma Shinyanga Singida Tabora Tanga West Lake

1978 Population Census Rural Total Urban (3) (2) (1)

Number of Households (in thousands)

Average Household Size Household 1976/77 Household Budget Survey Population Census Budget Survey 1976/77 1988 Urban 1978 Total Rural (4) (9) (8) (7) (6) (5)

TABLE 18 Tanzania: Rura l and Urban Households, 1976-1978 .

Profile of Incomes and Poverty in Tanzania 6 1 definition o f a household. I n 1978 , abou t 8 7 percent of th e population live d i n rural areas , compared wit h 9 3 percen t reporte d i n the 196 7 population census . Although recent data have not been analyzed in detail yet, it is thought that the rate o f urba n growt h i n th e las t decad e i s les s tha n th e 9 percen t annua l rat e experienced betwee n 196 7 an d 1978 . Abou t 9 1 percen t o f rura l households i n 1978 live d i n registere d village s (entitie s tha t hav e lega l an d corporat e statu s under Tanzanian law), 4.8 percent in unregistered, traditional villages, and only a small proportion outside villages. This is a result of the villagization campaign, as before 1970 , mos t o f th e rura l households live d i n unregistere d village s o r outside villages. Table 1 9 shows the frequency distribution of households by size in urban and rural areas circa 1977. It can be seen that, as expected, rural households have, on average, more members (mea n size 5.8) tha n urban households (mean size 5.0) . The population censu s o f 197 8 showe d that about 45 percen t of th e population is economically activ e (7,687.4 thousand), of which 88 percent is in agriculture. Clearly, agriculture is the main economic activit y fo r the bulk of Tanzanians. Table 2 0 exhibit s th e educationa l leve l o f head s o f household s b y secto r o f activity of head of household. Over half the heads of households (51. 6 percent) did not have an y education i n 1977 ; 36.9 percent had some primary education ; and 9.1 percen t had completed primary education. It is clear from the table that almost all of the household heads without school education are in agriculture (94 percent), whil e th e bul k o f househol d head s with post-primar y educatio n (8 6 percent) ar e engage d i n th e nonagricultura l sectors . Amon g thos e with som e primary education , agricultur e occupie s 7 7 percent . Education , an d especiall y

TABLE 19 Tanzania: Distributio n of Rural and Urban Households b y Siz e (percentages) . Household Size

Rural

Urban

1 2 3-4 5-6 7-8 9-

4.8 9.3 25.9 25.1 17.3 17.6

12.6 10.9 23.9 24.7 14.8 13.1

Total number of households (000 ) 2,585.

1 453.

Source: Househol d Budget Survey , 1976/1977 , Table 4C.

7

One to three years of secondary education

Completed primary education

One to six school classes completed

No school clas s completed

Educational Level

a b

a b

a b

a b

7,470 0.29 22.25

123,810 4.84 44.83

955,610 37.32 85.34

1,470,900 57.44 93.80

0 0.00 0.00

600 51.28 0.22

570 48.72 0.05

0 0.00 0.00

900 0.97 2.68

30,520 32.74 11.05

36,290 38.93 3.24

22,790 24.44 1.45

120 1.08 0.36

4,810 43.37 1.74

4,960 44.72 0.44

1,060 9.56 0.07

70 0.28 0.21

4,100 16.18 1.48

14,060 55.49 1.26

6,340 25.02 0.40

Mining Conand Agricul- Quarry- Manufac- Public strucing turing Utilities tion ture

4,000 6.05 11.91

13,170 19.92 4.77

29,740 44.99 2.66

16,060 24.30 1.02

Commerce

Industry of Head of Household

4,000 8.06 11.91

9,820 19.80 3.56

14,460 29.15 1.29

18,780 37.86 1.20

2,710 17.34 8.07

5,850 37.43 2.12

4,530 28.98 0.40

270 1.73 0.02

Total

14,310 6.62 42.61

83,470 38.64 30.23

33,580 1.11 100.00

276,160 9.09 100.00

59,580 1,119,78 0 27.58 36.85 5.32 100.00

31,870 1,568,06 0 14.75 51.60 100.00 2.03

Transport and Community Communication Finance Services

Tanzania: Househol d Distribution According to Educational Level and Industry of Head of Household, 1976/77 .

TABLE 20

a b

a b

a b

93,230 100.00 3.07

2,560,590 1,170 100.00 100.00 84.26 0.04

1,880 2.22 7.19

0 0.71 5.33

0.00 0.00

920 0.00 0.00

0.04 7.43

0.07 6.53

0 0.00 0.00

2,070 3.04 2.68

0 0.71 3.79

140 4.02 9.25

11,090 25,340 66,100 100.00 100.00 100.00 0.36 2.18 0.83

660 0.00 0.00

0 1.26 0.49

49,600 100.00 1.63

470 0.00 0.00

770 5.12 8.83

630 4.50 78.45

2,540 7.90 59.30

9,720 0.41 100.00

1,640 0.95 100.00

15,630 216,010 3,038,74 0 100.00 100.00 100.00 0.51 7.11 100.00

0 4.03 5.08

2,660 10.49 5.70

Source: Househol d Budget Survey, 1976/77 , Tabl e 27 . Note: Figure s in row "a" below number s correspon d to vertical percentages . Figures i n row "b " correspond t o horizontal percentages .

Total

Vocational course after primary school or form iv, vi aggregate

Completed form iv, v, or vi aggregate

64 Alexander

H. Sarris and Rogier van den Brink

post-primary education, are thus strongly associated with movement out of agricultural activities. A strikin g aspect of the table is the small proportion of househol d heads with post-primary education (only 2.5 percent). These figures describe a poor human capital situatio n in the late 1970s , despite efforts a t universal primary and adult education after the Arusha declaration and the villagization campaign. According to some observers, those policies bore fruit only toward the late seventies. For instance, according to the International Labour Office (IL O 1982), in 197 0 general illiteracy was 68 percent, and the primary school enrollment rate was 35 percent. By the late 1970s, illiteracy was down to 10 percent, and primary school enrollment was up to 70 percent (ILO 1982). STRUCTURE OF INCOME Tanzanians hav e diversified pattern s o f income . Tabl e 21 exhibit s th e source s of average , pe r househol d incom e i n rura l an d urba n Tanzania , includin g nonmonetary (subsistence ) income . Abou t 4 8 percen t o f tota l incom e i n rural areas is nonmonetary income, which is basically consumption of own-produced food. In the urban areas the proportion is much lower at 6.6 percent, as expected. If w e coun t al l nonmonetar y incom e a s incom e fro m agricultur e (th e actua l proportion i s 97. 5 percen t i n rural , an d 96. 7 percen t i n urban , areas) , the n own-account agricultura l activitie s (includin g fishing ) accounte d fo r 67.7 percent o f average , per-househol d rura l incom e i n 1976/7 7 an d 8. 9 percen t o f average urban household income . Wages and salaries are the main source of income in urban areas, accounting for 49.9 percent of tota l income, they account for only 7.7 percen t of total, and 16.2 of cash, income in the rural areas. In the rural areas the second major income source is the category of trade, enterprise, and professional activities, accounting for 1 7 percen t o f total , an d 35. 7 percen t o f cash , incom e fo r th e averag e household. Notic e tha t i n rura l areas , cash incom e fro m trade , enterprise , an d professions i s a s importan t a s cas h incom e fro m cro p sale s an d muc h mor e important than cash income from wages. This might be due to the fact that wage labor in Tanzania was illegal fo r many years and might be underestimated. The same holds for urban areas where 23.9 percent of total income (25. 6 percent o f cash income ) i s accounte d fo r by trade, enterprise, o r profession. Furthermore , this component of income is one of the most heavily underestimated, especiall y in urban areas (ILO 1982). Almost al l household s i n rural areas (86. 7 percent ) have som e cas h incom e from trade, enterprise, or profession, while the same holds for only 59.2 percent of urba n households. Also , a fairly larg e proportio n o f rura l household s (36. 8 percent) have some income from remittances or gifts.

Sale of assets

Remittances and gifts

Interests and dividends

Rents, sublets

Registered cooperative s

Trade, own enterprise, or profession

Wages and salaries

Fishing

Animal husbandry

Crop husbandry

Source of Cash Income ) )

) 6 ) 0 ) 2 ) 9 ) 8 ) 5 ) 1 )

4

0

(Table continues on the following page.)

3 2,14 4 ) (70.57 ) 9 1,51 2 ) (49.77 ) 5 11 4 ) (3.75 ) 7 1,06 3 ) (34.99 ) 8 2,51 0 ) (82.62 ) 9 61 3 ) (20.18 ) 3 8 7 ) (2.86 ) 8 9 7 ) (3.19 ) 1 1,08 1 ) (35.58 ) 7 25 5 ) (8.39 )

2061 8 (79.73) (18.32 1453 5 (56.21) (13.02 99 1 (3.83) (3.31 756 30 (29.25) (67.77 2,242 26 (86.73) (59.16 584 2 (22.59) (6.40 34 5 (1.32) (11.70 79 1 (3.06) (3.97 940 14 (36.36) (31.13 218 3 (8.43) (8.17

4

937 11 (17.31) (1.20 113 2 (2.09) (0.28 39 7 (0.72) (0.80 418 4,55 (7.72) (49.91 921 2,18 (17.01) (23.94 148 17 (2.73) (1.91 5 16 (0.09) (1.77 11 6 (0.20) (0.69 94 23 (1.74) (2.56 33 9 (0.61) (0.99

0 81 ) (13.64 6 10 ) (1.68 34 ) (0.74 7 1,03 ) (17.36 6 1,11 ) (18.60 4 15 ) (2.55 22 ) (0.49 31 ) (0.30 4 11 ) (1.93 04 ) (0.69

Number of Households Making Cash Income from Given Source (in thousands) Rural Urban Total

Average Household Income from Different Sources (in Tsh per household) Rural Urban Total

TABLE 21 Structure of Househol d Income i n Rural and Urban Tanzania, 1976/77 .

2,585.1 453.7

8,527 (93.39) 604 (6.61) 9,131 (100.00)

1 (0.08) 67 (0.73) 175 (1.92) 106 (1.16) 497 (5.44)

Urban

3,038.8

3,686 (61.75) 2,283 (38.25) 5,969 (100.00)

2 (0.03) 14 (0.23) 53 (0.89) 33 (0.55) 126 (2.11)

Total

Source: Computed from Household Budget Survey, 1976/77 , Tables 1 3 and 4A. Note: Number s in parentheses below figures in the first thre e columns denote vertical percentages.

Number of households ('000 )

Total income

Nonmonetary incom e

Total state d

Cashing of bank savings, securities, etc.

Loans and overdrafts fro m bank s

Loans from famil y o r friend s

2,836 (52.38) 2,578 (47.62) 5,414 (100.00)

1 (0.02) 5 (0.09) 31 (0.57) 20 (0.37) 60 (1.11)

Lottery, scholarship s

Pensions, insurance, provident fun d

Rural

Average Household Income Number from Different Sources Cash (in Tsh per household) (in

Source of Cash Income

(continued)

TABLE 21

2,585

453

Total

)

3,038

38 (1.25) 31 (1.02) 583 (19.19) 77 (2.53) 343 (11.29)

) (100.00

6 (1.32) 8 (1.77) 89 (19.65) 22 (4.86) 55 (12.14)

Urban

(100.00) (100.00

32 (1.24) 23 (0.89) 493 (19.07) 55 (2.13) 288 (11.14)

Rural

of Households Making Income from Given Source thousands)

Profile of Incomes and Poverty in Tanzania 6

7

The average reported cash income seems to be much higher for urban households, compared with rural ones, by a ratio of 3 to 1 . This, however, is counterbalanced by a more than 4-to-l advantag e in the rural areas as far as nonmonetary income i s concerned . Whe n added, the average , total per-household incom e i n the urban areas appears to be almost 70 percent higher than average, per-household rura l income . Whethe r thi s translate s t o a correspondin g rea l incom e differential, o f course , depend s o n quantitie s an d price s o f simila r type s o f consumed goods. The figures i n Table 21 hide the fact that within rural and urban areas, there are both farm an d nonfarm households wit h quite differen t pattern s of income . Table 2 2 exhibit s th e structur e o f cash incom e fo r thes e differen t type s o f households and compares it with the corresponding structure from the 1969 HBS. It can be noticed from th e table that in the rural areas, sources of incom e o f both far m an d nonfarm household s shifte d betwee n 196 9 an d 1976/7 7 towar d trade, enterprise, and profession, a s well a s "other" sources (which includes al l the other categories exhibited in Table 21). In the urban areas, the pattern is quite different. First , notic e th e shar p increas e i n th e numbe r o f so-calle d "farm " households livin g i n urban areas between 196 9 an d 1976/77 . Thes e ar e households whose heads ar e employed by agriculture-relate d enterprise s but who are not necessarily farmers . For these, cas h income shifte d towar d crop husbandry and "other," and away from trade or enterprise. For nonfarm urban households, the only noticeable shift is away from wages and trade and toward other sources of income . The change most noticeable in Table 22, however, is in the total value of cash incomes. In the rural areas, between 196 9 and 1976/77, the average cash income per household almost tripled for both farm and nonfarm households. In the urban areas, it appears to have declined in nominal terms for farm households, while it increased by less than 50 percent for nonfarm households. Given that the increase in th e nationa l consume r pric e inde x durin g th e perio d wa s 121. 7 percen t (obtained by averaging the last two quarters of 197 6 and the first two quarters of 1977 [Tanzani a Burea u o f Statistics , Economic Survey 1982]) , an d tha t th e definitions o f a household implie d a larger average househol d siz e i n 1976/77 , the data indicate a substantial drop in real per capita incomes in the urban areas, as shown in Table 23. If the degree of underestimation of incomes is the same in both the 196 9 and 1976/77 surveys, the data imply a small increase in rural, real per capita incomes an d an enormous declin e o f 6 8 percen t i n urban, per capita real incomes. Thi s was also the conclusion reache d by the IL O mission i n 198 2 (ILO 1982). The structur e o f rura l income , a s reporte d b y tw o othe r detaile d incom e surveys, don e i n 1979/8 0 an d 1982/83 , i s show n i n Tabl e 24 . I n th e 1979/8 0

68 Alexander

H. Sarris and Rogier van den Brink

TABLE 22 Tanzania: Structur e of Averag e Cas h Income per Household, Rura l and Urban Farm and Nonfarm Households , 196 9 and 1976/7 7 (percentages) . 1969 Farm Nonfarm Total Sources of rural cash household income Crop husbandry 36.90 11.94 Animal husbandry 16.10 2.63 Wages and salaries 10.58 56.74 Trade, enterprise, profession 26.91 21.36 Other 9.87 7.34 Total

38.94 4.68 5.68

13.04 1.65 45.48

33.04 3.98 14.74

24.95 9.06

34.67 16.03

25.14 14.69

32.48 15.76

1,826

982

2,501

5,207

2,836

354.3

2,637

2,264.7

320.4

2,585.1

0.76 0.40 60.39

8.44 2.88 22.34

0.30 0.58 57.76

1.29 0.86 53.44

29.22 9.23

40.01 26.32

23.63 17.73

25.64 18.78

Sources of urban cash household income Crop husbandry 1.89 0.56 1.13 0.37 Animal husbandry 19.71 63.04 Wages and salaries Trade, enterprise, 57.18 28.66 profession 20.08 7.38 Other Average cash income per household (Tsh) Number of households ('000)

31.36 12.53 22.61

100.00 100.00 100.00 100.00 100.00 100.00

Average cash income per household (Tsh ) 851 Number of households 2,282.7 ('000)

Total

1976177 Farm Nonfarm Total

100.00 100.00 100.00 100.00 100.00 100.00 3,800

6,299

6,036

3,434

10,742

8,527

16

136

152

137.5

316.2

453.7

Source: Compute d from Household Budge t Surveys , 196 9 and 1976/77 .

survey, th e share s o f subsistenc e a s wel l a s thos e o f wage s an d own-business , summarized unde r own business , appea r to be lowe r tha n i n the 1976/7 7 HBS, while the livestock share appears to be much higher. In the 1982/83 survey, crop income (including subsistence) appears to be lower, while own-business incom e is higher than in 1976/77 and 1979/80. The share of livestock income is also quite high. Althoug h th e tw o survey s sighte d abov e ar e no t representative—th e 1979/80 surve y coverin g eigh t region s an d th e 1982/8 3 surve y coverin g onl y four—the ver y larg e differenc e i n livestoc k incom e shar e merit s som e discussion.

Profile of Incomes and Poverty in Tanzania 6

9

TABLE 23

Tanzania: Nomina l an d Rea l Pe r Capit a Cas h Income s i n 196 9 an d 1976/7 7 (Tsh/capita). 1969 1976177 Rural Urban Nominal pe r capita cas h income (Tsh ) 21 Real pe r capita cas h incom e (1969 base ) 21 National CP I 10

Rural

3 2,41

4 489.

3 2,41 0 10

4 221. 0 221.

Urban 0 1,705.

0

0 769. 7 221.

0 7

Source: Computed from Household Budget Surveys, 1969 and 1976/77.

TABLE 24 Tanzania: Compositio n o f Rural Household Income, 198 0 and 198 3 (percentages).

Net income from crop sales Subsistence crop production Net livestock incom e Own-business Wages Remittances Per capita annual income (Tsh) Mean household size Per household annual income (Tsh)

1979/80

1982/83

14.1 41.4 21.0 19.5a

48.5 b

4.0 734.3 5.3 3,892.0

c

13.9 26.3 6.4 4.8 1,549.0 d d

Sources: For 1979/80, Collier et al. (1986), pp. 65-66; for 1982/83, Bevan et al. (1989), p. 54. a I n Collier et al., only total nonfarm earnings are reported. b I n Bevan et al. (1990), the crop income is reported as originating from food crops (41.5 percent) and cash crops (7 percent). c I n Bevan et al. (1990), 72 percent of all farm income (62 percent of crop income and more than 100 percent of livestock income) is derived from subsistence. This implies that 45 percent of total income is derived fromsubsistence. d No t indicated in the relevant source.

In bot h surveys , particularl y th e 1979/8 0 surve y (p . 64) , th e problemati c nature of livestock incom e is readily acknowledged. The reason is that although livestock yield s norma l output s (suc h a s milk , meat , offals , etc. ) tha t ca n b e valued in a standard way, it is, on the other hand, one of the main assets of rural households. The latter implies that sales of livestock shoul d better be thought of in a separate income category, that of sales of assets, in which sales of any other

70 Alexander

H. Sarris and Rogier van den Brink

asset woul d als o b e recorded . Similarly , purchase s o f livestoc k shoul d b e recorded separatel y an d mad e par t o f investment expenditures . I n th e sam e fashion, stock valuation adjustments, such as livestock births and deaths, should be include d a s a separat e componen t o f incom e fo r whic h ther e shoul d b e a corresponding, equa l investmen t outlay . Th e separat e recordin g o f livestoc k sales an d purchases a s asse t change s shoul d be done because conceptuall y th e net of these is equivalent to net dissaving i n livestock assets , and hence it is not a proper part of current or permanent income. The 1979/8 0 surve y include s livestoc k sale s a s part of curren t income, doe s not include livestock valuation, and is not clear as to whether it nets out livestock purchases. If only sales of livestock are included, this would tend to overestimate total income, as well a s income fro m livestock, an d indeed this seems to be the case i n the results. The 1982/8 3 surve y (p . 305 ) correctl y include d a s part of livestoc k incom e own-consumption o f produc e fro m livestoc k (suc h a s milk) . Ne t cas h sale s o f stock (sale s minus purchases) are also included, but they are again netted out in the stock valuation definition. Henc e i t appears that livestock incom e includes , apart from consumption or sales of products, net stock valuations—that is, births plus net gifts (received minus given), minus deaths, thefts, own-consumption (of stock), an d stoc k give n t o labor . I t i s no t clea r wh y al l stoc k valuation s ar e included as part of current income. While some parts of stock valuation, such as net birth s o r gift s o f livestoc k received , ar e certainl y par t o f curren t income , own-consumption o f stoc k o r gift s o f stoc k give n ou t ar e par t o f househol d expenditure, not income. Hence counting al l stock valuation s as part of incom e will probably bias the "normal" income considerably. For instance, in a bad year a liquidation of livestock herds would count as negative income, and in fact this is wha t seem s to account fo r the large negative livestoc k incom e o f th e poorer rural household s i n Beva n e t al . (1990) . Th e prope r wa y woul d hav e bee n t o record all such asset transactions separately as saving or dissaving and to include as part of current income only those asset changes that result from net births, net transfers of stock , an d stock given out as wages. In fact, this is what appears to have been done in the 1976/7 7 HBS, as sources of income fro m asset transactions, includin g livestock sales , are recorded separately, an d correspondingly investments , includin g purchase s o f livestock , ar e recorded separately as part of the total distribution of proceeds from all sources. As ca n be see n fro m Table 21 , however, i n the 1976/7 7 survey , proceeds fro m sales o f asset s ar e reporte d t o b e ver y low . I t i s no t clea r fro m th e survey' s methodological explanatio n exactl y ho w th e variou s incom e component s wer e measured, an d henc e i t i s no t clea r wha t exactl y i s include d unde r livestoc k income an d sales o f assets . The conclusion i s that some doub t must be cast o n

Profile of Incomes and Poverty in Tanzania 7

1

the rura l livestoc k incom e figures , especiall y thos e fro m th e 1979/8 0 an d 1982/83 surveys, particularly when one attempts to compare the composition o f rural househol d income s ove r time . I t appears , nevertheless , tha t rura l cro p income, including subsistence consumption , declined a s a share of total incom e from 1976/77 to 1982/83, while the share of income from trade and entrepreneurship seems to have increased. Table 2 5 exhibit s th e differen t source s o f cas h incom e fo r households wit h heads employed in the private, cooperative, or public sector. The greatest number of Tanzania n household s i n 1976/7 7 ha d heads wh o wer e privatel y employed , mainly i n agriculture . Onl y 11. 4 percen t wer e employe d b y cooperatives , parastatals, or in other public-sector jobs. The bulk of the cash income o f those households cam e fro m wage s an d salaries , an d their total cas h incom e wa s o n average more than twice that of households operating in the private sector. Given, however, tha t mos t privat e household s ar e rural , farmin g household s an d that subsistence incom e accounts fo r about half of their total income (cf . Tabl e 21), the difference i n total averag e incom e betwee n households whos e head s ar e in the private sector and those in the public sector does not appear to be very large. The education of the head of household appear s to make a significant differ ence in average household cash income in both rural and urban areas, as well as in the individual sectors , as shown in Table 26. Among rura l households, those whose head had completed primary school had cash incomes about twice as high as thos e wit h no schoo l education . Fo r thos e wit h secondar y o r vocationa l post-primary education , the ratio is close t o four to one . Those with high-leve l secondary education (completed form VI) had the highest overall cash incomes, but thei r number s wer e rathe r small , an d they wer e al l concentrate d i n publi c service. That education is positively associated with cash incomes was also found by Collier et al. (1986). Table 26 also exhibits the average household cash income by economi c secto r an d industr y o f th e hea d o f household . Household s wit h heads i n commerc e an d finance see m t o b e th e wealthies t overall . Ther e wer e 66,110 households whose head was in commerce in 1976/77 , only 2.2 percent of the total number; only 15,60 0 households had heads engaged in finance. The bulk of household heads (84.3 percent) wer e engaged in agriculture. Table 27 exhibits the annual average cash incomes of Tanzanian household s by secto r an d economic activit y o f hea d o f household . I t i s interestin g t o not e that households whos e hea d is engaged in agriculture do not exhibit much cash income differentiation irrespectiv e of whether the head is employed privately or by th e publi c secto r i n al l it s variou s forms . Whe n subsistenc e incom e i s considered, average income in agricultural households does not seem to be very different irrespectiv e o f whethe r the household hea d is engaged privatel y o r in an agricultural cooperative, agricultural parastatal, or agricultural public service.

908 (29.94) 114 (3.76) 51 (1.68) 237 (7.81) 1,190 (39.24) 165 (5.44) 24 (0.79) 20 (0.66) 105 (3.46) 40 (1.32)

Crop husbandry

Sale of assets

Remittances and gifts

Interests and dividends

Rents, sublets

Registered cooperative

Trade, own enterprise, or profession

Wages and salaries

Fishing

Animal husbandry

Private

Source of Cash Icnome 693 (9.61) 8 (0.11) 57 (0.79) 4,397 (60.96) 1,340 (18.58) 121 (1.68) 80 (1.11) 12 (0.17) 124 (1.72) 69 (0.96)

Registered Cooperative 139 (1.48) 7 (0.07) 3 (0.03) 7,417 (79.13) 702 (7.49) 55 (0.59) 68 (0.73) 12 (0.13) 161 (1.72) 81 (0.86)

415 (5.42) 31 (0.41) 6 (0.08) 5,896 (77.06) 404 (5.28) 112 (1.46) 20 (0.26) 6 (0.08) 186 (2.43) 44 (0.58)

Public Parastatal Service 43 (0.63) 15 (0.22) 0 (0.00) 5,206 (76.42) 839 (12.32) 27 (0.40) 78 (1.15) 14 (0.21) 213 (3.13) 9 (0.13)

Other 822 (22.07) 101 (2.71) 45 (1.21) 1,052 (28.25) 1,116 (29.97) 153 (4.11) 28 (0.75) 19 (0.51) 114 (3.06) 42 (1.13)

Total Stated

501 (22.71) 42 (1.90) 7 (0.32) 380 (17.23) 881 (39.94) 92 (4.17) 49 (2.22) 8 (0.36) 121 (5.49) 29 (1.31)

Sector not Stated

Tanzania: Annua l Cash Income in Private Households by Source of Income and Sector of Head of Household, 1976/77 (Tsh per household).

TABLE 25

0 (0.33) 7 (1.22) 8 (0.92) 4 (3.43)

(0.03)

9

3 (100.00)

Source: Househol d Budge t Survey , 1976/77 , Tabl e 23. Note: Numbe r in parenthesis indicate s percentage .

Number of households ('000) 2,55

Total stated 3,03

Cashing of bank savings, securities, etc. 10

Loans and overdrafts fro m banks, etc. 2

Loans from family o r friends 3

Pensions, insurance, provident fund 1

Lottery, scholarships 1

(1.12) (3.97)

(0.73) (1.08)

287

7,213 (100.00)

78

53

(0.44)

(0.72) (3.78)

289

55

(2.39)

163

30

150

(2.20)

(2.08)

159

(0.00)

0

(0.34)

23

(0.29)

22

(0.08)

(3.44)

128

(0.91)

34

(1.42)

53

14

(0.38)

(1.50)

33

(0.73)

16

(1.50)

33

13

(0.59)

1 (0.05)

2 (0.05)

144

164

68

2,963

76

6,812 3,724 2,206 9,373 7,651 (100.00) (100.00) (100.00) (100.00) (100.00)

372

105

166

(1.77)

(2.04)

147

(0.86)

81

6 (0.04)

(0.47)

34

04 (0.00)

0 0

10,961 22,919

0 29,267

2,870 2,591

0

2,636

9,786

8,544

0 0 0 0

10,493 14,106 3,313 11,189 24,051 12,066 0 0 0 17,101 35,165 0

10,443

5,145

3,584

9,154

4,777

5,150

0

0

5,156

2,829

8,035

3,030

0

5,494

2,330

2,337 9,561

8,334 8,212 13,697

9,574 4,714 12,119

7,092 5,806

8,220

0

34,871

12,804

0

0 27,218

0 56,665

0

8,857 5,714 16,617

0

0

0

8,139

0

0

0

8,186 13,020 0 0

8,721

8,099

7,196

8,225

7,198

7,286

6,330

3,623

2,654

10,380

8,735

45,649 25,565

17,889 13,717

0 11,659

3,686

29,267

18,485

10,403

13,184 10,825 12,574 17,702 13,739 21,014 0 0 0 20,599 24,337 28,140

6,854

7,793

4,622

Transport CommuCon- and nity struc- Com- Commution merce nication Finance Services Total

19,977 9,647 17,033 38,406 19,891 13,985 8,817 88,776 0 0 0 0 33,812 0 0 63,112

7,690

8,131

6,871

Mining and Manu- Public Quar- factor- Utilirying ing ties

Source: Househol d Budget Survey, 1976/77, Tables 27 and 33.

Total

No schoo l class completed One to six school classes completed Completed primary education One to three years of secondary education Completed form iv Completed form v Completed form vi Vocational cours e afte r primary school Vocational cours e afte r form iv Vocational course afte r form vi

Educational Level

AgriculRural Urban ture

TABLE 26 Tanzania: Annua l Cas h Income by Educational Leve l o f Hea d of Household , 1976/7 7 (Ts h per household).

Source: Househol d Budge t Survey, 1976/77 , Table 25.

Number of households ('000)

Total

Community service s

Finance

Transport and communication

Commerce

Construction

Public utilities

Manufacturing

3,033 84.202

2,580 80.969 6,182 0.003 8,646 0.667 25,298 0.026 3,104 0.212 18,164 1.844 16,128 0.113 18,465 0.037 9,906 0.330

Agriculture

Mining and quarrying

Private

Industry of Head of Household 4,847 0.239 6,527 0.014 8,512 1.21,4 9,888 0.094 9,640 0.090 7,477 0.085 9,093 0.525 10,005 0.242 10,375 2.242 9,373 4.745

5,176 0.069 0 0.000 8,956 0.227 6,913 0.013 3,941 0.066 8,001 0.121 3,914 0.070 7,840 0.105 7,339 0.275 7,213 0.945

Registered Cooperative Parastatal

3,724 97.504

6,812 2.222 7,651 5.390

3,038.880 100.000

2,603 81.776 5,150 0.039 8,220 3.068 8,857 0.365 5,714 0.834 16,617 2.175 8,139 1.632 10,308 0.508 8,735 7.108

Total Stated

4,315 0.196 3,949 0.009 7,028 0.491 8,791 0.059 8,013 0.108 8,502 0.122 6,688 0.528 8,987 0.023 6,771 0.685

Other

5,203 0.303 4,205 0.012 7,749 0.469 5,971 0.173 5,901 0.357 2,591 0.003 7,279 0.397 10,363 0.101 8,083 3.575

Public Service

2,206 2.486

2,206 2.486 0 0.000 0 0.000 0 0.000 0 0.000 0 0.000 0 0.000 0 0.000 0 0.000

Sector not Stated

Tanzania: Annua l Cash Income in Private Households by Economic Activity and Sector of Head of Household, 1976/7 7 (Tsh per household).

TABLE 27

76 Alexander

H. Sarris and Rogier van den Brink

This also seems to hold for those households whose heads are engaged in mining, manufacturing, an d communit y services . However , fo r th e household s whos e head is engaged in public utilities, commerce, transport, and finance, the private sector seem s t o offe r muc h better average income s tha n the public sector . Fo r construction, the reverse seems to be the case. For those households whose head is engage d i n construction vi a parastatals, average annua l cash income i s mor e than three times that of households whose head is engaged in private construction activities. DISTRIBUTION OF INCOME The 1976/7 7 Househol d Budget Surve y compile d informatio n abou t both cash and noncash incomes . Earlie r attempts t o loo k a t income distribution , suc h a s that o f th e IL O (1982) , wer e hampere d b y th e availabilit y o f cas h incom e statistics only a t the time. Table 2 8 exhibit s househol d incom e statistic s fo r Tanzani a mainlan d fo r 1976/77. I t is interestin g t o note tha t most incom e differentiatio n i s du e to the cash component. Th e ratio of averag e pe r capita nonmonetary incom e betwee n the highest income group and the lowest is 5.5, while the same ratio for per capita cash income is 43.9, and for total income i s 20.6. It clearly appears that despite government effort s durin g th e 1970s , there wa s substantia l incom e differentia tion an d inequality i n Tanzania in 1976/1977 . Nonmonetar y incom e wa s much more equitably distribute d than cash income . Tha t is to be expected, a s almos t all nonmonetary income consists of subsistence food production, which depends largely on labor. Nevertheless, it is interesting to note that higher-income households have higher per capita cash incomes as well as nonmonetary incomes. This could b e du e t o a produc t compositio n effect—namely , tha t higher-incom e households als o produce higher value foods (e.g. , livestock products) . In 1976/77 , averag e tota l incom e pe r household i n Tanzania stoo d a t 5,969 Tsh, of which 2,283 Ts h or 38.2 percent, was nonmonetary income . The figure s show the importance of subsistence agriculture in the economy. It is to be noted that fo r th e lowes t thre e incom e groups , whic h accoun t fo r 3 4 percen t o f th e population, average per capita subsistence income was equal to, or greater than, per capit a cas h income . Onl y i n th e highes t thre e househol d incom e groups , which account for only 1 6 percent of the population, was the share of subsistence in total per capita income les s tha n 30 percent. Given that traditionally, house hold budget survey s underestimat e incomes , particularly thos e o f high-incom e groups, thi s figur e als o most likel y i s a n underestimate. A t th e ver y bottom o f the distribution, 8 percent o f households , o r 5 percent o f the people, enjoyed 2 percent of total income. Forty-one percent of households, or 33 percent of people,

Total

106.6817,153.97 5.65 8.01

13.32 3,038.75

40,000and Over

88.52 84.16 66.65

82.22 76.46 57.32

69.07 61.17 42.17

Source: Compute d from Household Budge t Survey , 1976/77 , Tabl e 43. a Denot e Gin i coefficients .

405.94 6.30 7.70 9.34

440.33 13.15 15.29 15.15

382.67 28.44 28.43 23.40

98.82 98.31 91.33

553.93 10.30 14.14 24.67

99.56 99.38 95.21

100.00 100.00 100.00

969.22 1,330.79 0.74 0.44 1.07 0.62 3.88 4.79

876.65 1,290.88 2,858.78 6,808.15

607.01

487.95

0.36J 0.26J

404.42 100.00 100.00 100.00

652.96

870.45 1,047.34 1,282.59 1,844.94 3,828.00 8,138.94 1,057.38

2,159.00 2,890.00 2,801.00 4,292.00 7,949.00 10,658.0 0 2,283.00

2,753.00 3,984.00 6,049.00 10,002.0 0 23,446.00 54,525.0 0 3,686.00

4,911.00 6,874.00 8,850.00 14,295.0 0 31,395.0 0 65,183.0 0 5,969.00

184.06 8.20

191.37

4,876.47 2,623.20 1,320.45 2,426.11 5.64 6.56 6.90 7.75

399.68

25,00039,999 22.44

864.33

No. of households ('000 ) 14.91 234.18 985.41 No. of househol d members ('000 ) 28.97 809.02 4,779.0 1 Average household siz e 1.94 4.85 3.45 Average total incom e per household (Tsh ) 769.00 1,611.0 0 3,061.00 Cash income per household (Tsh ) 301.00 697.00 1,542.00 Nonmonetary incom e per household (Tsh ) 468.00 913.00 1,519.00 Per capita total incom e (Tsh) 395.83 466.33 631.16 Per capita cash incom e (Tsh) 154.94 201.76 317.95 Per capita nonmonetar y income 240.90 264.28 313.21 Percentage of household ?j 0.4 9 32.43 7.71 Percentage o f people 27.86 0.17 4.72 Percentage o f incom e 0.06 2.08 16.63 Cumulative % of households 0.49 8.20 40.63 32.74 Cumulative % of peopl e 0.17 4.89 Cumulative % of incom e 0.06 18.77 2.14

10,00024,999 313.12

4,0005,999

1,0001,900

2,0003,900

0-999

Income Group 6,0008,0007,999 9,999

TABLE 28 Tanzania: Distributio n o f Income i n Tanzani a Mainland , 1976/7 7 (Tsh) .

78 Alexander

H. Sarris and Rogier van den Brink

at the low en d enjoyed 1 9 percent o f tota l income . A t the very hig h end of th e distribution, 1 percent of households comprising 2 percent of the people, enjoyed 5 percen t o f tota l income , whic h i s probably a n underestimate. Eleve n percen t of households , comprisin g 1 6 percen t o f th e people , ha d 3 3 percen t o f tota l income at the top. These figures imply a rather equitable distribution of incom e compared with othe r countries . I n fact, th e Gin i coefficient , compute d o n th e basis of cumulative income s attribute d to households i n Table 28, is 0.363; the Gini coefficien t compute d o n th e basi s o f income s accruin g t o peopl e i s onl y 0.264. It must be realized a t this point that the income statistic s exhibite d i n Table 28 aggregat e rura l and urban households. W e could not obtain detailed, simila r statistics b y rura l an d urban classifications. I t might be argue d that the cost o f living i s highe r i n th e cities ; income s ough t t o b e highe r there , an d hence th e picture exhibited in Table 28 is distorted. On the other hand, however, while the cost of food is higher in the cities, the cost of nonfood items is higher in the rural areas. Sinc e foo d i s a very larg e componen t o f expenditure s i n Tanzania , th e relative cost of living i s probably higher in the cities, on balance. However, the distortion woul d ten d t o overestimat e inequalities . Henc e th e low Gin i coeffi cient found could be considered an overestimate of the true one. Earlier incom e distributio n studie s (e.g. , thos e reviewe d i n IL O 1982) use d data on cash incomes only an d utilized numbers of households, not people. For instance, th e Gin i coefficien t tha t was compute d by th e IL O mission (base d o n preliminary result s o n cas h incom e o f rura l households fro m th e 1976/7 7 HBS) was 0.4 9 (IL O 1982) . Tha t compare s wit h 0.36 4 compute d here , utilizin g ful l income figures an d aggregatin g ove r al l households . A s wil l b e see n below , however, tota l income s (cas h an d subsistence ) ar e muc h mor e equitabl y distributed. Table 29 exhibits regional income distributio n statistics computed from data similar to those of Table 28. With the exception of Kigoma and Mara, all regions exhibit Gin i coefficient s similar , t o o r smalle r than , th e on e fo r th e whol e o f Tanzania. Kigoma exhibits a high Gini coefficient, due to inequalities in reported nonmonetary income , whic h might impl y somethin g suspec t abou t the data. In per capita income terms, Dar es Salaam, Kigoma, Kilimanjaro, Mara, Morogoro, Rukwa, Tabora, and West Lake exhibit per capita incomes above the average for Tanzania as a whole. Table 30 exhibits the distribution of per capita incomes in Tanzania mainland in 1976/77 . Becaus e o f househol d compositio n effects , thi s distributio n i s no t expected to be the same as that derived from analyzing household incomes. For instance, Tabl e 28 indicate s tha t in all o f Tanzania only 28,970 people, o r only 0.2 percent of the population, lived in households whose total annual income was

139.8 99.4 165.2 181.4 168.2 103.2 165.7 119.2 119.6 186.8 164.0 177.6 243.2 73.9 109.8 204.0 105.1 149.1 180.1 183.4 3,038.8

2,881 2,481 10,988 1,829 1,961 2,814 5,492 1,960 5,843 2,874 4,607 2,143 3,205 3,634 2,599 2,706 1,881 4,319 3,638 4,834 3,686

Source: Compute d from Household Budge t Survey , 1976/77 .

Arusha Coast Dar es Salaam Dodoma Iringa Kigoma Kilimanjaro Lindi Mara Mbeya Morogoro Mtwara Mwanza Rukwa Ruvuma Shinyanga Singida Tabora Tanga West Lake Tanzania 2,228 1,342 319 2,196 2,326 7,317 2,665 1,837 1,784 2,192 1,938 2,037 1,869 5,226 2,477 2,237 2,657 2,816 1,599 2,058 2,283

5,109 3,824 11,306 4,025 4,287 10,131 8,157 3,797 7,627 5,066 6,545 4,180 5,074 8,860 5,076 4,943 4,538 7,135 5,237 6,892 5,969

958 676 2,251 784 806 1,672 1,381 775 1,144 863 1,183 841 758 1,463 1,039 753 782 1,294 927 1,377 1,057

Nonmonetatry Number of Cash Income Income Per Total Income Total Income Households Per Household Household Per household Per Capita

Tanzania: Regiona l Income Distribution Statistics, 1976/77 (Tsh).

TABLE 29

0.274 0.243 0.297 0.373 0.275 0.504 0.231 0.247 0.438 0.253 0.289 0.278 0.299 0.322 0.305 0.335 0.348 0.360 0.271 0.332 0.363

Gini Coefficient Based on Household Income 0.224 0.122 0.228 0.241 0.093 0.553 0.095 0.146 0.518 0.165 0.194 0.208 0.202 0.175 0.227 0.218 0.251 0.235 0.128 0.193 0.264

Gini Coefficient Based on Personal Income

0.122 15.670

61.537

501-1,000

921.046

459.306 3,038.674

1,001- 2,0002,000 and Above Total

238.040 1,167.615

401-500

174.635

301-400

Per Capita Income Group

0.000 0.001

0.137

0.904

3.614

12.416 350.000 2.711

5.747 8.191

0.016 0.025 0.983 4.226 25.000 75.00 0 150.000 250.000 0.000 0.001 0.136 0.767

0.543

2.025 3.243 8.315

0.027

0.023

0.516 0.958 2.568

0.004 0.009

0.023 0.016

8.001

22.724 450.000 4.386

16.149

7.834 10.308

15.115 8.348 100.000

30.311 26.220 84.885

38.291

75.484

100.000

91.652 100.000 65.432 750.000 1,500.000 3,106.00 0 30.290 37.193 24.520

54.574

38.425 42.708

100.000

0.140{

100.000 100.000

2.814 1.461 164.265 556.318 1,404.981 1,768.159 7,325.90 5 4,497.686 1,431.981 17,153.570 5.645 4.883 3.118 6.274 4.003 11.975 10.483 9.040 8.045 7.428

0.703

201-300

Source: Computed from Household Budget Survey, 1916/17, Table 43. a Denote s Gini coefficients.

Number of households ('000) Number of people ('000) Average household size Percentage of households Percentage of people Cumulative percentage of households Cumulative percentage of people Income per capita Percentage of income Cumulative percentage of income

0-50 51-100101-200

Tanzania: Distributio n of Per Capita Income, 1976/77.

TABLE 30

Profile of Incomes and Poverty in Tanzania 8

1

less than 1,00 0 Tsh . The per capita income o f this household grou p is 39 6 Tsh. In Tabl e 30 , however , w e observ e tha t i n 1976/7 7 ther e wer e 2. 1 millio n Tanzanians, or 12. 4 percent, living i n households whose per capita income wa s less tha n 400 Tsh . This implie s tha t within each household incom e class , ther e are many households whose per capita income is much below the mean per capita income for the household class as a whole. The 1976/7 7 HBS tables we obtaine d did no t specif y th e mea n pe r capit a incom e o f household s i n eac h pe r capit a income class . T o compute the Gini coefficient o f th e personal incom e distribution in Table 30, we assumed that the mean in each per capita income grou p is the unweighted average of the bounding per capita incomes. For the top income group, the mean was estimated so that the total income computed is equal to that reported. Th e Gin i coefficien t thu s compute d i s 0.14 , whic h i s quit e smal l b y international standards. Despite the possible underestimation of inequality inherent in household budget surveys, per capita income in Tanzania around 1976/7 7 appears to have been quite evenly distributed . Up to now, we have considered only income. We were not able to break down the distribution of income by rural or urban location. We have, however, obtained data on distribution of household cash expenditures by rural and urban divisions. Admittedly, it would have been better to have households classified by their per capita expenditures, rather than per-household expenditures, but those data were not available . Th e subsequen t analysi s i s therefor e carrie d ou t o n th e basi s o f groups classified accordin g to per-household expenditures . Table 3 1 present s a breakdow n o f rura l an d urba n household s int o cas h expenditure categories. The first thing to notice is that rural households with low cash expenditures (less than 4,000 Tsh) have, on average, higher total per capita consumption than urban households of simila r average cash expenditures. This, of course , is du e to the much higher amount of subsistenc e consumption i n the lower-income rura l households. By contrast , urban households wit h high monetary expenditure s hav e muc h higher total consumptio n expenditure s tha n rural households with simila r expenditures . Thi s las t resul t mus t be qualified , how ever, because rura l households i n the categories wit h cas h expenditure s abov e 40,000 Tsh are reported to have average monetary consumption smaller than the lower bound s o f thei r respectiv e ranges . I t i s no t clea r wh y thi s i s th e case . Another observatio n fro m th e tabl e i s tha t althoug h i n th e rura l area s th e nonmonetary income per household is fairly uniform across expenditure classes, it tends to decline for higher-expenditure household s i n urban areas. Inequality seem s t o be muc h higher i n urban areas than in rural ones. Thu s the ratio of mea n per capita total income s between the highest an d lowest cas h expenditure group s i n the rural areas i s 4.6, whil e i n the urban areas i t is 20.3. Slightly ove r 70 percent of urban households had total incomes below the mean

Rural households Number of households ('000) Percentage of households Number of household members per household Cash income per household Nonmonetary income per household Total income per household Per capita total income Monetary consumptio n expenditures per household Subsistence consumption per household Total consumption expenditure per householda Per capita total consumption 7 6,217 2,780 8,997 1,285 6,084 2,698 8,825 1,261

6 4,344 2,232 6,576 1,061 4,367 2,178 6,566 1,059

6 2,578 2,430 5,008 835 2,615 2,378 5,009 835

2,898 4,344 804 1,388 2,855 4,260 789

2,406 3,013 701

616

2,340

3,001 698

112 4

5 1,446

307 12

4 607

887 34

2,0003,999

723 28

1,0001,999

415 16

0-999

8 13,131 2,983 16,115 1,918 10,692 2,903 13,607 1,620

2,862 10,983 1,308 7,932 2,799 10,738 1,278

80 3

8 8,122

54 2

6 2,836 2,578 5,414 933 2,704 2,523 5,249 905

12 33,576 4,502 38,078 3,200 14,032 4,392 18,424 1,548

10 28,868 3,202 32,060 3,375

17,175 1,808

3,104

14,048

2,585 100

2 0 6 0

Expenditure Categor ies 4,000- 6,000- 8,000- 10,000- 25,000- 40,0007,999 9,999 24,999 39,999 and Over Total 5,999

Tanzania: Distributio n of Rural and Urban Households According to Cash Expenditures (Tsh).

TABLE 31

5 2,989 953 3,942 857 2,934 898 3,843 835

5 1,384 1,392 2,776 603 1,432 1,347 2,800 609

4 636

1,346 1,983 472

608

1,297

1,932 460

86 19

37 8

25 6

89 20

5,280 1,200

438

4,835

475 5,448 1,238

4 4,973

Source: Compute d from Household Budge t Survey, 1976/77 , Tables 2 and 3. a Include s some other minor items.

Urban households Number of households ('000) Percentage o f households Number of household members per household Cash income per household Nonmonetary income per household Total income per household Per capita total income Monetary consumption expenditures per household Subsistence consumption per household Total consumption expenditure per household8 Per capita total consumption 7,183 1,437

324

6,845

375 7,615 1,523

5 7,241

60 13

9,016 1,610

221

8,788

276 9,504 1,697

6 9,228

54 12

14,695 2,578

170

14,506

246 15,279 2,681

6 15,033

85 19

29,950 4,754

131

29,807

197 29,408 4,668

6 29,211

10 2

48,668 6,667

433

48,231

603 69,774 9,558

7 69,171

9 2

8,465 1,693

548

7,904

604 9,131 1,826

5 8,527

454 100

84 Alexander

H. Sarris and Rogier van den Brink

urban household income of 9,131 Tsh , while over 85 percent of rural households had total income below the rural mean of 5,414 Tsh. Note, however, that average household siz e i n rura l area s i s a much stronge r positiv e functio n o f averag e household income than in urban areas. This implies that the share of people living in households with cash expenditures below th e mean is about the same in rural and urban areas. The Gini coefficient compute d from Table 31 on the basis of households and cumulative total household incomes (both monetary and subsistence) is 0.208 i n rural areas and 0.418 i n urban areas. The Gini coefficient compute d on the basis of incom e accruin g to people (tha t is, after taking int o account household size ) is only 0.12 9 i n rura l area s an d 0.38 2 i n urba n areas . Whil e thes e statistic s certainly underestimate inequality, because households are not ranked according to tota l expenditur e o r incom e bu t only accordin g t o cas h expenditure, the y indicate tha t inequalit y i s muc h greate r i n urba n tha n i n rura l area s an d that , absolutely speaking , inequalit y i n rura l area s appear s t o hav e bee n lo w i n 1976/77 by international standards . The presence o f subsistenc e incom e moderate s inequalit y significantly . Fo r instance, had we computed the Gini coefficient i n rural areas on the basis of cash incomes only, a procedure that was tried by the ILO mission (ILO 1982), we would find a figure o f 0.413—twic e a s high as the Gini computed on the basis of total incomes. Table 3 2 exhibit s th e source s o f cas h incom e i n various cas h expenditur e categories. Give n that for most expenditure an d income groups (especiall y i n the lowe r range ) cas h expenditur e i s quit e clos e t o cas h incom e (se e Tabl e 31), i t is quite clea r that households wit h low cas h expenditure (an d income ) derive th e bul k o f thei r incom e fro m nonmonetar y sources , chiefl y subsis tence agriculture . Thi s wa s alread y apparen t fro m Tabl e 28 , wher e fo r th e three lowes t incom e groups , mor e tha n hal f o f tota l incom e cam e fro m subsistence agricultura l production . What is of interest in Table 32, however, is tha t fo r al l expenditur e groups , cas h incom e fro m agricultur e (includin g animal husbandr y an d fishing ) i s neve r mor e tha n 4 5 percen t o f tota l cas h income. Fo r th e lowes t thre e cas h expenditur e classes , i t i s 3 4 percent , 4 3 percent, an d 40 percent, respectively ; fo r the highest two , i t is only aroun d 2 percent. Clearly , then, the poor derive a proportionately highe r share of cash, as well a s total, incom e fro m agriculture. However , al l income group s have a well diversifie d patter n o f cas h earnings . I t i s interestin g t o not e i n thi s context tha t th e larges t nonagricultura l componen t o f cas h incom e i n low income household s i s fro m trade , enterprise , o r profession. Wag e an d salary income seem s t o be insignifican t a t lower income s an d is importan t onl y fo r the higher-expenditure households .

Sale of assets

Remittances and gifts

Interests and dividends

Rents, sublets

Registered cooperatives

Trade, own-enterprise, or profession

Wages and salaries

Fishing

Animal husbandry

(Table continues on the following page.)

256 551 853 1,152 1,618 2,057 3,269 13,750 27,992 1,110 (8.66) (12.91) (17.36) (18.22) (19.00) (20.09) (20.84) (45.30) (43.92) (18.60) 12 46 99 204 268 196 651 249 6,176 152 (0.41) (1.08) (2.01) (3.23) (3.15) (1.91) (4.15) (0.82) (9.69) (2.55) 1 4 12 23 71 62 78 390 2,253 29 (0.24) (0.36) (0.83) (0.61) (0.50) (1.28) (3.53) (0.49) (0.03) (0.09) 0 3 9 15 54 47 101 209 420 18 (0.00) (0.18) (0.24) (0.07) (0.63) (0.46) (0.64) (0.69) (0.66) (0.30) 53 62 102 143 219 226 344 215 94 115 (1.79) (1.45) (2.08) (2.26) (2.57) (2.21) (2.19) (0.71) (0.15) (1.93) 13 10 29 28 99 79 204 80 1,139 41 (0.44) (0.44) (0.23) (0.59) (0.69) (1.16) (0.77) (1.30) (0.26) (1.79)

1,355 1,328 1,630 510 814 562 928 1,039 179 285 (6.06) (13.16) (18.88) (16.43) (15.91) (12.97) (10.39) (0.94) (0.80) (13.64) 26 49 126 144 225 109 99 39 747 100 (0.88) (1.15) (2.56) (2.28) (2.64) (1.06) (0.63) (0.13) (1.17) (1.68) 4 13 40 78 92 185 80 196 0 44 (0.14) (0.30) (0.81) (1.23) (1.08) (1.81) (0.51) (0.65) (0.00) (0.74) 41 95 343 1,512 2,293 3,985 6,476 10,182 6,785 1,036 (1.39) (2.23) (6.98) (23.91) (26.92) (38.92) (41.29) (33.54) (10.64) (17.36)

Crop husbandry

Cash Expenditure Group 4,000- 6,000- 8,000- 10,000- 25,000- 40,0005,999 7,999 9,999 24,999 39,999 , and Over Total

0-999

2,0003,999

Source of Cash Income

1,0001,999

TABLE 32 Tanzania: Source s of Cash Income of Households According to Cash Expenditure Categories, 1976/7 7 (Tsh/household).

2 (0.04) 5 (0.10) 29 (0.59) 7 (0.14) 32 (0.65)

1 (0.02) 0 (0.00) 17 (0.40) 3 (0.07) 27 (0.63)

1 (0.03)

0 (0.00) 10 (0.34)

3 (0.10)

11 (0.37)

50 (0.79)

9 (0.14)

16 (0.25) 69 (1.09)

5 (0.08)

153 (1.80)

8 (0.09)

12 (0.14) 102 (1.20)

3 (0.04)

570 (3.63)

1,579 (5.20)

1,157 (3.81)

294 (1.87)

38 (0.37) 183 (1.79)

616 (2.03) 143 (0.47)

0 (0.00)

119 (0.76) 183 (1.17)

11 (0.07)

0 (0.00) 179 (1.75)

4 (0.04)

13,207 (20.72)

1,325 (2.08)

0 (0.00) 1,761 (2.76)

0 (0.00)

Cash Expenditure Group 4,000- 6,000- 8,000- 10,000- 25,000- 40,0007,999 9,999 24,999 39,999 and Over 5,999

126 (2.11)

33 (0.55)

14 (0.23) 53 (0.89)

2 (0.03)

Total

440.3

759 972.9

395.7

171.8

107.7

164.4

16.1

10.7

3,038.8

8,678 14,109 29,088 62,408 3,686 1,443 4,485 2,615 609 6,573 (20.61) (33.80) (53.22) (70.92) (77.18) (84.75) (89.95) (95.82) (97.91) (61.75) 1,562 1,576 1,332 1,839 1,944 1,268 2,283 2,346 2,826 2,299 (2.09) (38.25) (4.18) (79.39) (66.20) (46.78) (29.08) (22.82) (15.25) (10.05) 6,324 8,517 10,240 15,685 30,356 63,740 5,969 4,269 2,955 4,914 (100.00) (100.00 ) (100.00 ) (100.00 ) (100.00 ) (100.00 ) (100.00 ) (100.00 ) (100.00 ) (100.00 )

Source: Compute d from Household Budge t Survey , 1976/77 , Table 21.

Number of households C000)

Total income

Nonmonetary income

Total stated

Cashing of bank savings, securities, etc.

Loans and overdrafts fro m banks

Loans from family o r friends

Pensions, insurance, provident fund

Lottery, scholarships

2,0003,999

1,0001,999

0-999

Source of Cash Income

(continued)

TABLE 32

Profile of Incomes and Poverty in Tanzania 8

7

The figure s elaborate d abov e ar e corroborate d b y th e result s o f th e rura l household survey reported by Collier et al. (1986). In that analysis, poorer rural households were found to rely for about three-quarters of their total income (cash and subsistence) on crop income, of which 95 percent was subsistence consumption. Nonfarm income was found to be more important as a source of cash income than crop an d livestock produc t sale s fo r both poor an d rich rural households . The sam e resul t wa s foun d i n a later surve y (Beva n e t al . 1990 ) conducte d i n 1983. For the poorer households, nonfarm income from own-business and wages appeared to be greater than income from cash crop sales. However, the proportion of tota l incom e derive d fro m cash cro p sale s wa s foun d t o b e greate r amon g low-income group s than among higher-income ones . INCOME DIFFERENTIATION

The structur e o f incom e analyze d i n the previous section s exhibite d moderat e inequality i n rural areas but substantial inequalit y i n urban ones. What are the underlying components of this differentiation? A n answer cannot be given with the average figures s o far presented, although education and wage employmen t appear to be significantl y relate d to income differences . A thoroug h analysis of pattern s of rural differentiation wa s made by Collie r et al . (1986) . I n that study , i t wa s show n tha t mos t o f th e incom e difference s between rura l poor and nonpoor could be traced to differences i n ownership o f assets. However, size of landholding did not appear to be highly associated with income differences . Thi s can be expected i n a land-abundant economy, suc h as that o f Tanzania . Ownershi p o f livestoc k appeare d t o b e a fa r greate r factor , followed by education. It is not clear, however, whether ownership of livestoc k is the cause or the result of incom e differentiation. Sinc e livestock i s one of the main form s o f investment , i t seem s quit e reasonabl e tha t riche r household s should have more assets, namely livestock , a s a result of saving s o r investment decisions. Acces s t o wag e employmen t wa s see n i n tha t stud y a s a majo r explanatory facto r o f nonfar m incom e i n rura l areas . Unit return s fo r sale s o f different far m product s di d no t appea r t o b e ver y differen t betwee n poo r an d nonpoor; thus prices ar e not significantl y differen t betwee n poor an d nonpoor. The poor were far more likely to engage in their own business than the nonpoor, but the returns from suc h activitie s wer e muc h lower fo r th e poor than for th e nonpoor. It was thus concluded that the poor were somehow forced into a range of marginal, nonfarm activities. That the size of landholding was not strongly associate d with total per capita income amon g rura l household s wa s als o show n i n th e stud y o f Beva n e t al . (1990). Th e per capita total incom e (cas h an d subsistence) fro m foo d an d cash

88 Alexander

H. Sarris and Rogier van den Brink

crops i n households operatin g ver y smal l landholdin g wa s neve r less tha n half the pe r capit a incom e o f household s operatin g landholdin g 1 0 o r more time s larger. Pe r capita incom e derivin g fro m nonfar m sources , however , includin g own-business an d wages , wa s muc h large r i n household s operatin g smalle r landholding. CONSUMPTION PATTERNS The consumption patterns in Tanzania, a s revealed by th e 1976/7 7 Househol d Budget Survey , exhibit several interesting features . Table 33 exhibits expendi ture shares on 1 0 broad classes of expenditure, by expenditure category in rural areas. Tabl e 3 4 exhibit s th e sam e informatio n fo r urba n areas. The share s ar e exhibited fo r bot h monetar y an d subsistenc e consumptio n an d fo r combine d expenditures. Among bot h rural and urban households, food expenditure s dominate , comprising 75 percent of total spending in rural areas and 66 percent in urban areas, on average . Th e hig h foo d expenditur e o f th e riches t rura l clas s mus t b e a n aberration, since per capita total, and especially monetary , expenditures i n that class appea r much lower than those of th e next two wealthy classes ; that coul d easily bias the estimation of the shares. While food constitutes about 97 percent of subsistence expenditures, for most expenditure classes, whether rural or urban, it constitutes a much lower share of monetary expenditures—50 percent in rural areas an d 6 3 percen t i n urba n areas , o n average . I n rura l areas , th e shar e o f monetary expenditures on food does not vary much by expenditure class; it varies considerably in urban areas, from 65 percent in the lowest cash expenditure class, to 42 percent in the highest. Among other expenditure categories, the second-most-important expenditur e share, i n both rura l an d urban areas an d for al l expenditur e classe s excep t th e very highest, is clothing an d footwear, accountin g fo r an average of 3 0 percent of tota l cas h expenditure s i n rura l area s an d 1 3 percen t i n urba n areas . Othe r important item s of monetar y consumptio n expenditures amon g the expenditur e classes ar e fuel, ligh t an d water, furniture, an d utensils. Rents ar e important i n urban areas only fo r high-expenditure households , whil e househol d operation s also become importan t for high-expenditure classes . Another interestin g featur e o f th e table s i s reveale d b y th e figures fo r pe r capita consumption and for total expenditure in the various classes, which appear in th e botto m o f th e tables . Th e consumptio n expenditure s categor y refer s t o expenses in all 1 0 consumption categories listed in the tables. Total expenditure includes spending for education, other investments, and savings in various forms. The figures reported for total expenditures i n Table 31 correspon d (apar t for m

0-999

48.82 4.72 0.08 4.89 28.66 5.53 4.15 2.05 0.08 1.01 100.00

97.13 2.87 100.00

87.52 1.02 0.02

Expenditure

Monetary Food Drinks and tobacco Rents Fuel, light, and water Clothing and footwear Furniture and utensils Household operations Personal care and health Recreation and entertainment Transportation Total

Subsistence Food Other Total

Monetary and subsistence Food Drinks and tobacco Rents 82.08 1.54 0.02

97.25 2.75 100.00

47.84 3.77 0.08 4.17 31.22 5.67 3.05 2.39 0.26 1.55 100.00

1,0001,999

74.69 1.77 0.08

97.14 2.86 100.00

51.53 3.01 0.16 3.63 28.55 4.76 2.29 2.35 0.90 2.82 100.00

2,0003,999

69.49 2.12 0.13

96.85 3.15 100.00

53.67 2.95 0.20 3.04 28.38 5.06 1.74 1.80 0.79 2.37 100.00

4,0005,999

66.48 1.97 0.35

97.77 2.23 100.00

49.70 2.97 0.54 3.39 30.93 5.18 1.40 2.14 1.06 2.68 100.00

65.79 2.83 0.75

96.58 3.42 100.00

53.12 3.95 1.06 2.24 24.77 5.67 1.36 1.93 1.62 4.27 100.00

6,000- 8,0007,999 9,999

Income Group

57.61 2.20 0.52

96.48 3.52 100.00

43.35 2.84 0.71 2.53 35.24 5.03 1.87 2.10 0.98 5.35 100.00

73.74 1.62 0.00

90.50 9.50 100.00

63.94 1.52 0.00 10.91 10.60 1.98 2.03 4.30 0.91 3.81 100.00

40,000and Over

74.69 1.78 0.17

97.10 2.90 100.00

50.32 3.20 0.35 3.48 29.51 5.09 2.19 2.20 0.82 2.84 100.00

Total

(Table continues on the following page.)

55.97 1.60 3.66

86.03 13.97 100.00

47.74 1.93 4.66 6.52 22.08 3.41 2.10 4.94 1.06 5.55 100.00

10,000- 25,00024,999 39,999

TABLE 33 Tanzania: Expenditur e Shares on Various Consumption Categories by Different Incom e Groups, Rural (percentages).

3.12 5.70 1.10 0.83 0.43 0.02 0.25 100.00

679.5 135.2 544.3 687.5

Fuel, light, and water Clothing and footwear Furniture and utensils Household operations Personal care and health Recreation and entertainment Transportation Total

Per capita consumption expenditure (Tsh) Monetary Subsistence Per capita total expenditure (Tsh)

Source: Compute d from Household Budget Survey, 1976/77 .

762.9 234.3 528.6 785.8

2.79 9.59 1.74 0.94 0.74 0.08 0.48 100.00

0-999

Expenditure

1,0001,999

(continued)

TABLE 33

2.76 17.98 3.21 1.11 1.16 0.50 1.55 100.00

4,0005,999 2.91 20.13 3.37 0.91 1.42 0.69 1.75 100.00

6,0007,999 2.42 17.55 4.02 0.97 1.38 1.15 3.13 100.00 2.53 25.79 3.68 1.38 1.56 0.71 4.02 100.00

5.59 18.28 2.99 2.28 4.38 0.83 4.42 100.00

8,000- 10,000- 25,0009,999 24,999 39,999 1M 1.11 1.48 1.62 3.14 0.58 2.62 100.00

40,000and Over

780.5 958.4 1102.6 1,141.3 1,287.2 1,505.2 999.5 384.3 607.2 717.6 808.6 941.8 1,181.7 630.7 396.2 351.2 385.0 332.7 345.4 323.5 368.8 832.2 1,055.7 1,254.5 1,277.5 1,618.5 1,805.4 1,548.1

2.91 14.06 2.35 1.13 1.17 0.45 1.41 100.00

2,0003,999

Income Group

834.8 399.8 435.0 901.2

2.86 14.14 2.44 1.05 1.07 0.39 1.39 100.00

Total

0-999

64.58 2.81 0.26 6.61 17.42 4.02 2.55 0.88 0.06 0.81 100.00

97.60 2.40 100.00

84.65 1.10 0.10

Expenditure

Monetary Food Drinks and tobacco Rents Fuel, light, and water Clothing an d footwear Furniture and utensils Household operations Personal care and health Recreation and entertainment Transportation Total

Subsistence Food Other Total

Monetary an d subsistence Food Drinks and tobacco Rents 83.61 1.40 0.21

96.89 3.11 100.00

70.37 2.76 0.42 4.58 15.68 1.69 2.33 1.09 0.05 1.04 100.00

1,0001,999

74.89 1.62 1.25

96.48 3.52 100.00

61.74 2.14 1.66 5.14 16.20 2.49 1.61 1.31 0.58 1.12 100.00

2,0003,999

66.43 1.82 4.11

96.14 3.86 100.00

63.37 1.99 4.53 6.15 14.82 3.69 1.58 1.44 1.22 1.20 100.00

4,0005,999

70.44 1.33 3.15

97.31 2.69 100.00

68.92 1.40 3.33 6.08 12.42 2.63 1.37 1.30 0.55 2.00 100.00

6,0007,999

70.91 1.69 2.96

97.23 2.77 100.00

70.08 1.74 3.05 7.14 11.19 1.79 1.33 1.15 0.41 2.11 100.00

64.25 2.91 4.56

93.91 6.09 100.00

63.75 2.93 4.64 6.35 12.85 2.00 2.19 1.30 0.58 3.42 100.00

43.25 2.99 11.02

96.00 4.00 100.00

42.34 3.02 11.21 8.19 9.26 0.71 8.04 1.48 1.93 13.83 100.00

40,000and Over

66.00 2.27 4.17

96.59 3.41 100.00

63.14 2.46 4.56 6.40 12.90 2.29 2.60 1.40 0.79 3.46 100.00

Total

(Table continues on the following page.)

51.61 4.30 7.10

94.93 5.07 100.00

51.29 4.32 7.16 6.17 12.45 2.78 6.41 2.70 1.40 5.33 100.00

8,000- 10,000- 25,0009,999 24,999 39,999

Income Group

Tanzania: Expenditur e Shares on Various Consumption Categories by Different Incom e Groups, Urban (percentages).

TABLE 34

508.2 199.3 308.9 453.8

Per capita consumption expenditure (Tsh ) Monetary Subsistence Per capita total expenditure (Tsh) 586.6 293.8 292.8 604.1

3.75 7.85 0.85 1.16 0.58 0.03 0.55 100.00

1,0001,999

Source: Compute d from Household Budget Survey, 1976/77 .

4.01 6.85 1.57 1.01 0.36 0.02 0.32 100.00

0-999

Fuel, light, and water Clothing and footwear Furniture and utensils Household operations Personal care and health Recreation and entertainment Transportation Total

Expenditure

(continued)

TABLE 34

5.89 7.0 11.75 10.8 2.49 1.7 1.30 1.2 1.23 1.1 0.52 0.4 1.89 2.0 100.00 100.0

5.88 13.44 3.34 1.44 1.32 1.11 1.11 100.00 1 5 4 9 1 0 5 0

6.29 12.64 1.97 2.16 1.29 0.57 3.36 100.00

6.14 8.0 12.36 9.1 2.76 0.7 6.36 7.9 2.69 1.4 1.39 1.8 5.29 13.6 100.00 100.0

6.11 11.80 2.09 2.38 1.29 0.73 3.17 100.00

Total

4 1,283.2 0 1,173.6 4 109.6 8 1,690.3

7 0 0 1 7 9 0 0

40,00010,000- • 25,000- and 24,999 39,999 Over

6 1,821.0 : 2,784.9 3,518. 1 1,791.2 : 2,764.0 3,459. 5 29.8 20.9 59. 9 : 2,574.7 < 1,752.4 6,666.

6,000- 8,0007,999 9,999

4,0005,999

784.3 1,062.8 1,207.9 1,291. 589.0 963.3 1,143.1 1,252. 195.3 99.5 64.8 39. 833.2 1,198.4 1,433.8 1,608.

4.71 12.17 1.87 1.21 0.99 0.44 0.85 100.00

2,0003,999

Income Group

Profile of Incomes and Poverty in Tanzania 9

3

minor rounding) t o the bottom line s o f Table s 3 3 an d 34. It is quit e revealing , but also expected, that households i n the lower expenditure categorie s reporte d consumption expenditures quite close to total expenditures; in other words, very little saving. In fact, for the lowest urban class dissaving is reported, on average. In contrast , household s i n th e to p thre e o r fou r expenditur e classe s repor t substantial exces s o f expenditur e ove r consumption, hence substantia l savings . Nevertheless, almos t al l classe s repor t som e savings , albei t i n most case s the y are small. A fina l featur e o f th e table s i s th e breakdow n o f pe r capit a consumptio n between monetar y an d subsistence. Amon g rura l households, per capita subsistence consumption is highest among low-cash-expenditure groups , and declines for higher-expenditure groups, as expected. For the lowest two cash expenditure groups, subsistence consumption constitutes 8 0 percent and 70 percent, respectively, of total consumption. However, even for the highest expenditure groups, subsistence stil l constitutes a substantial shar e of total consumption (mor e than 20 percent). This is quite different tha n in urban areas. For the lowest-expendi ture households there , subsistenc e constitute s 61 percen t o f tota l consumptio n expenditures, but for the four highest expenditure groups it is less than 3 percent. Clearly, then , a substantia l portio n o f th e tota l consumptio n o f th e poores t households i n Tanzania is shielded from price fluctuations . Given tha t foo d constitute s a n overwhelmin g shar e o f consumptio n i n Tanzania, the next item of discussion concerns the composition of food consumption. Table 3 5 exhibit s th e consumption o f 1 2 categories o f food s i n kilograms per capita per year, by cash expenditures class in the rural areas. Table 36 exhibits the same data for the urban areas. Cereal and starchy root consumption per capita (the two major staple classes) seem to be quite even in the rural areas, irrespective of expenditure class. The composition of cereal and starch consumption changes, however, with rice becoming more important and sorghum flour less importan t as expenditures rise. Amon g starches , cassav a an d sweet potatoes becom e les s important, whil e cookin g banana s rise i n importanc e a s expenditur e rises. Al l other food items exhibit a positive expenditure elasticity, as expected, since they are largely nonstaple . In the urban areas, a rather similar pattern emerges. Tables 3 7 an d 38 exhibit th e total daily pe r capita calorie intake s of variou s kinds o f foo d b y monetar y expenditur e categor y i n rura l an d urba n areas , respectively. Thes e ar e computed by multiplying th e respective quantitie s con sumed by appropriat e calorie content coefficient s an d converting the m to daily equivalents. The first major observation is that daily per capita calorie intake in the rura l area s i s quit e evenl y an d narrowl y distribute d amon g expenditur e classes aroun d th e mea n o f 2,15 3 kilocalorie s pe r capit a pe r day . I n fact , i t appears that wealthier rural households consum e fewe r calorie s per capita than

0.2 93.1 23.3 17.8 20.0

0.0 90.9 32.1 14.0 7.2

1.4 43.5 3.3 11.4 4.0 4.0 1.6 8.1 0.2

Cereal products Starchy roots and starches Cassava Sweet potatoes Cooking bananas

Sugar and sweets Pulses, dry Nuts Vegetables Fruits Meat, meat products, poultry Fish and shellfish Milk and dairy products Oils and fats

Source: Compute d from Household Budget Survey , 1976/77 .

2.6 40.0 2.2 14.1 5.6 5.7 2.2 4.8 0.6

144.6 14.3 93.0 10.9

134.9 10.7 77.2 25.3

Cereals Maize, grain Maize, flour Sorghum, flour

0-999

1,0001,999

4.2 20.7 2.7 12.5 6.8 9.3 2.8 6.5 0.3

0.5 104.2 19.7 22.0 30.3

125.2 17.0 68.3 9.2

2,0003,999

6.6 22.1 2.6 13.4 8.2 13.5 3.4 3.9 0.6

0.8 76.0 10.8 15.5 31.8

142.3 20.2 74.7 12.9

4,0005,999

6.6 22.9 2.3 11.7 9.0 13.0 2.7 11.3 0.9

1.1 106.7 11.9 13.1 47.4

137.7 30.3 72.6 4.1

6,0007,999

7.9 15.2 3.2 11.3 11.9 14.0 2.7 12.0 1.5

1.8 102.1 15.2 12.0 62.3

146.3 42.7 64.0 2.0

11.2 19.4 3.7 11.0 9.9 16.3 2.6 9.6 1.3

1.4 120.6 13.5 21.5 69.0

105.4 15.8 54.0 2.0

14.9 15.7 3.4 22.1 9.6 16.2 3.3 21.9 2.9

1.4 104.9 11.2 30.0 60.2

91.8 3.3 56.5 0.1

8,000- 10,000- 25,0009,999 24,999 39,999

Expenditure Class (Tsh)

7.9 35.5 0.9 8.2 8.5 16.2 2.3 0.6 1.2

0.6 47.0 4.4 5.1 4.0

133.1 10.4 69.1 2.3

40,000and Over

4.3 28.3 2.6 12.8 6.7 9.0 2.6 6.6 0.5

0.5 96.2 19.8 18.3 28.4

133.1 17.1 75.5 11.2

Total

TABLE 35 Tanzania: Annua l Consumptio n in Private Households , 1976/77 , by Food Items by Househol d Expenditure Group, Rural (quantities i n kilogram/capita/year).

3.6 9.5 5.2 7.4 3.1 6.7 1.2 3.3 0.0

35.2 13.8 15.2 0.5

1.7

115.0 6.4 5.2 43.1 46.7

4.6 9.8 5.2 9.6 3.7 7.0 2.2 5.4 0.0

59.1 34.6 12.8 0.9

1.7

110.9 10.2 10.0 44.6 27.6

Source: Compute d from Household Budge t Survey , 1976/77 .

Sugar and sweets Pulses, dry Nuts Vegetables Fruits Meat, meat products, poultry Fish and shellfish Milk and dairy products Oils and fats

Starchy roots and starches Cassava Sweet potatoes Cooking bananas

Cereal products

Cereals Rice, husked Maize, grain Maize, flour Sorghum, flour

0-999

1,0001,999

7.0 12.2 6.5 14.8 5.0 10.7 3.3 3.9 0.4

67.8 21.1 12.6 13.5

4.1

94.8 12.4 18.5 48.3 10.4

2,0003,999

11.4 13.0 11.6 14.8 6.4 13.4 5.2 1.8 0.9

37.5 9.8 6.1 13.4

9.3

88.2 21.1 15.2 48.2 0.7

4,0005,999

13.6 15.0 18.4 17.2 9.6 14.6 4.4 1.2 1.0

31.0 6.6 6.8 9.4

32.2

91.8 27.8 7.4 52.0 0.6

6,0007,999

15.7 13.6 21.3 18.0 12.3 16.1 3.8 0.7 1.3

31.4 6.3 6.8 9.6

16.3

97.0 33.0 8.6 52.1 0.4

19.8 16.7 25.1 24.4 16.1 20.4 4.7 1.6 2.5

38.6 11.8 8.4 11.1

21.1

106.1 40.4 6.5 54.7 0.4

20.5 14.6 18.3 31.1 26.0 27.1 3.3 3.2 5.4

31.0 4.1 3.8 13.8

26.0

97.6 45.1 8.7 34.0 0.2

8,000- 10,000- 25,0009,999 24,999 39,999

Expenditure Class (Tsh)

24 A 19.9 4.2 24.9 22.9 22.7 1.8 6.4 7.5

52.3 3.3 5.3 24.1

16.3

99.2 41.4 2.9 30.0 0.0

40,000and Over

13.0 13.8 15.0 17.6 10.2 14.8 4.0 2.4 1.4

43.2 13.0 8.8 10.6

14.6

98.4 26.0 10.6 49.2 6.4

Total

TABLE 36 Tanzania: Annua l Consumptio n i n Private Households , 1916/17, b y Food Items by Househol d Expenditur e Group, Urban (quantities i n kilogram/capita/year).

Kilocaloriesl capital day

Cash Expenditure Class (Tsh) 4,000- 6,000- 8,000- 10,0005,999 7,999 9,999 24,999

Total

2,224.64 2,320.5 2 2,032.7 0 2,161.8 9 2,254.3 3 2,312.6 4 2,050.9 8 1,965.8 7 2,104.6 3 2,152.8 4

915.65 1,305.74 1,316.2 4 6.29 5.53 14.62 362.36 175.40 345.27 86.57 47.24 163.81 146.10 330.33 263.39 13.81 35.71 53.53 10.15 27.25 15.73 11.81 10.46 8.29 115.47 115.53 63.86 22.67 28.61 19.89 16.15 1.45 53.99 29.01 72.67 12.75

25,000- 40,00039,999 and Over Total

1,333.69 1,434.23 1,237.91 1,400.56 1,359.4 8 1,441.03 1,037.04 Cereals 19.08 8.62 15.26 5.34 12.21 1.98 0.00 Cereal products 334.08 338.76 375.56 263.50 377.92 365.30 426.94 Starchy roots and starches 86.11 122.64 72.47 45.66 72.02 28.41 15.29 Sugar and sweets Pulses, dry 405.10 372.60 192.51 205.83 212.92 141.94 180.76 38.68 32.61 28.88 47.46 49.32 34.40 36.30 Nuts 15.41 16.50 14.44 13.94 13.50 17.35 Vegetables 14.05 10.14 11.10 12.18 8.42 Fruits 4.87 6.85 14.68 96.51 28.16 Meat, meat products, poultry 92.60 100.07 116.18 40.89 66.48 19.48 24.84 29.70 23.80 24.01 22.96 Fish and shellfish 14.27 Milk and dairy products 9.54 11.87 20.07 16.03 27.83 29.65 23.78 8.22 15.91 21.14 38.16 13.70 5.73 Oils and fats 32.29

1,000- 2,0000-999 1,999 3,999

TABLE 37 Tanzania: Dail y Pe r Capita Calorie Intake of Rura l Households, 1976/77 , by Househol d Cas h Expenditure Class .

61.81 0.09 14.60 1.22 16.06 1.48 0.75 0.30 1.76 0.84 0.51 0.59 100.00

59.95 0.00 15.02 0.69 18.21 2.22 0.63 0.22 1.27 0.64 0.90 0.26

100.00

Source: Compute d from Household Budge t Survey , 1976/77 .

Total

Cereals Cereal products Starchy roots and starches Sugar and sweets Pulses, dry Nuts Vegetables Fruits Meat, meat products, poultry Fish and shellfish Milk and dairy products Oils and fats 100.00

60.90 0.26 18.48 2.25 9.47 1.79 0.76 0.41 3.27 1.22 0.79 0.40 100.00

64.78 0.40 12.19 3.35 9.52 1.51 0.76 0.47 4.46 1.37 0.44 0.74 100.00

60.31 0.54 16.76 3.19 9.44 1.28 0.64 0.49 4.11 1.06 1.23 0.94 100.00

62.31 0.83 15.80 3.72 6.14 1.67 0.60 0.63 4.33 1.04 1.28 1.65

Percent of Total

100.00

50.56 0.74 20.82 5.98 8.81 2.31 0.66 0.59 5.66 1.12 1.16 1.57 100.00

46.58 0.74 18.43 8.33 7.43 2.72 1.39 0.60 5.87 1.46 2.75 3.70

100.00

62.04 0.30 8.30 4.11 15.70 0.66 0.48 0.50 5.49 0.95 0.07 1.38

100.00

61.14 0.26 16.00 2.19 12.23 1.66 0.73 0.39 2.97 1.05 0.75 0.59

Total

933.77 44.13 251.34 76.24 113.40 73.08 18.23 6.16 75.88 28.59 9.65 10.72

875.82 99.56 134.57 124.53 120.67 126.21 18.21 7.85 95.52 45.83 4.48 22.42

912.25 344.05 109.53 149.04 139.73 199.73 21.21 11.84 104.00 38.58 2.96 24.66

967.34 1,061.97 173.63 224.95 108.95 134.90 172.21 217.26 126.42 155.25 234.49 270.80 22.24 30.06 19.90 15.19 114.48 144.97 32.88 41.53 1.76 3.89 30.82 60.56

Kilocaloriesl capital day 972.43 278.15 99.00 224.40 136.03 203.35 38.36 32.09 193.35 29.22 7.83 133.07

985.83 174.18 163.50 264.21 185.03 50.37 30.74 28.20 161.98 15.61 15.88 185.78

977.16 156.00 154.92 142.47 128.55 161.26 21.70 12.58 105.42 35.07 5.92 34.52

25,000- 40,00039,999 and Over Total

1,545.73 1,637.71 1,641.19 1,675.68 2,057.5 6 2,000.4 1 2,366.03 2,347.2 8 2,261.3 1 1,935.55

Cereals 1,131.40 1,092.21 Cereal products 18.58 17.81 Starchy roots and starche s 129.92 229.39 Sugar and sweet s 39.14 50.03 Pulses, dry 88.71 91.13 Nuts 59.69 58.01 Vegetables 9.10 11.79 Fruits 3.82 4.56 Meat, meat products, poultr y 49.55 47.49 Fish and shellfis h 10.44 19.06 Milk and dairy product s 8.22 13.40 Oils and fat s 0.00 0.00

1,000- 2,0000-999 1,999 3,999

Cash Expenditure Class (Tsh) 4,000- 6,000- 8,000- 10,0005,999 7,999 9,999 24,999

TABLE 38 Tanzania: Dail y Pe r Capit a Calori e Intak e o f Urba n Households , 1976/77 , by Househol d Cas h Expenditur e Class .

100.00

73.20 1.15 8.41 2.53 5.74 3.86 0.59 0.25 3.07 0.68 0.53 0.00 100.00

66.69 1.13 14.01 3.05 5.56 3.54 0.72 0.28 3.03 1.16 0.82 0.00

Source: Compute d from Household Budge t Survey , 1976/77 .

Total

Cereals Cereal products Starchy roots and starches Sugar and sweets Pulses, dry Nuts Vegetables Fruits Meat, meat products, poultry Fish and shellfish Milk and dairy products Oils and fats 100.00

56.90 2.69 15.31 4.65 6.91 4.45 1.11 0.38 4.62 1.74 0.59 0.65

44.34 16.72 5.32 7.24 6.79 9.71 1.03 0.58 5.05 1.87 0.14 1.20 100.00

52.27 5.94 8.03 7.43 7.20 7.53 1.09 0.47 5.70 2.73 0.27 1.34 100.00

100.00

48.36 8.68 5.45 8.61 6.32 11.72 1.11 0.76 5.72 1.64 0.09 1.54

Percent of Total

100.00

44.88 9.51 5.70 9.18 6.56 11.45 1.27 0.84 6.13 1.76 0.16 2.56 100.00

41.43 11.85 4.22 9.56 5.80 8.66 1.63 1.37 8.24 1.25 0.33 5.67

100.00

43.60 7.70 7.23 11.68 8.18 2.23 1.36 1.25 7.16 0.69 0.70 8.22

100.00

50.48 8.06 8.00 7.36 6.64 8.33 1.12 0.65 5.45 1.81 0.31 1.78

100 Alexander

H. Sards and Rogier van den Brink

poorer ones . Thi s coul d b e du e t o househol d compositio n effects , a s rura l households with larger families tend to have higher overall consumption, without necessarily implyin g a higher per capita consumption. Nevertheless , i t appear s that the rural poor do not consume fewer calories than the rural rich. This is not so i n urban areas, where th e average dail y pe r capita calori e intak e of th e poor is muc h lowe r tha n tha t o f th e urban , high-expenditur e categories . Th e mea n daily pe r capita calorie intak e i n urban areas is estimated a t 1,93 6 kilocalories , which is 1 0 percent lower than the average figure fo r the rural areas. Cereals (an d products) constitut e a rather uniform 6 1 percen t o f tota l dail y calorie intak e i n the rura l area s acros s expenditur e classes , whil e i n th e urba n areas, albeit on average cereals and products constitute 59 percent of daily calorie intake, thi s varie s considerably , fro m a hig h o f 7 4 percen t fo r th e poores t households t o a low o f 5 1 percen t fo r th e richest . Starche s an d pulses ar e th e second majo r sourc e o f calori e intak e i n rural areas , wit h th e othe r categorie s contributing only mino r shares. In urban areas, however, sugar , nuts, and meat also contribute a substantial shar e of daily calori e intake that rises significantl y with household expenditure . A most interesting revelation from the 1976/77 HBS concerns the shares of the various foods consume d that are from subsistenc e production or purchased. We used information o n monetary an d subsistence consumptio n (whic h is valued at average price s fo r simila r products ) fro m th e HB S (Table s 3 C an d 3D) , t o disaggregate each category of food expenditure into a portion purchased for cash and a portion obtained from own-production by expenditure class. Tables 3 9 an d 4 0 exhibi t th e share s o f tota l consumptio n an d o f variou s food groups consumed in rural and urban areas, respectively, that are obtained for cash. In the rural areas, for the lowest expenditure classes, only 21 percent of thei r total consumptio n i s obtaine d throug h monetar y purchases , an d only 12 percent o f foo d an d drink consumption . However , ther e i s a sharp differ entiation i n th e share s purchase d amon g variou s foo d categories . Fo r th e same, lowes t rura l expenditur e class , only 6 percent o f cereal s consumption , 3 percent of pulses, 5 percent of vegetables, 1 percen t of starch , and 5 percent of dair y consumptio n ar e purchased ; th e res t i s obtaine d fro m own-produc tion. B y contrast , 9 7 percen t o f cerea l products , 9 4 percen t o f sugar , 6 5 percent o f meat , 9 2 percen t o f fish, an d 8 2 percen t o f oil s an d fat s ar e purchased. Fo r higher-expenditur e rura l households , th e share s o f staple s purchased i s considerabl y higher—37-4 3 percen t fo r cereals, 14-8 0 percen t for starches , 24-5 3 percen t fo r pulses , 48-7 9 percen t fo r vegetables , an d 43-95 percen t fo r milk . Fo r nonstaples, th e share s purchase d ar e a s hig h o r higher tha n thos e o f th e lower-expenditur e classes . Surprisingly , th e sam e pattern emerges amon g urba n households, with poorer ones purchasing only a

0.33 0.19 0.09 0.78 0.03 0.96 0.06 0.27 0.10 0.30 0.71 0.88 0.15 0.89

0.21 0.12 0.06 0.97 0.01 0.94 0.03 0.09 0.05 0.28 0.65 0.92 0.05 0.82 0.52 0.35 0.21 0.98 0.10 0.98 0.18 0.39 0.20 0.37 0.74 0.90 0.27 0.91

2,0003,999 0.67 0.50 0.29 0.99 0.18 1.00 0.28 0.58 0.38 0.57 0.89 0.93 0.56 0.96

Source: Compute d fro m Household Budget Survey , 1976/77 , Tables 3C and 3D.

Total consumption Food and drink consumption Cereals Cereal products Starchy roots and starches Sugar and sweets Pulses, dry Nuts Vegetables Fruits Meat, meat products, poultry Fish and shellfish Milk and dairy products Oils and fats

1,0001,999

0-999 0.69 0.50 0.32 0.98 0.22 0.99 0.30 0.62 0.41 0.53 0.93 0.97 0.33 0.96

0.74 0.59 0.43 1.00 0.16 0.96 0.46 0.61 0.54 0.57 0.96 0.94 0.62 1.00 0.79 0.56 0.39 0.97 0.14 0.99 0.32 0.75 0.48 0.67 0.91 0.97 0.43 0.93

0.82 0.67 0.37 0.74 0.80 0.89 0.53 0.78 0.79 0.89 0.90 0.99 0.73 0.58

0.76 0.55 0.38 1.00 0.23 0.94 0.24 0.42 0.76 0.56 0.86 0.94 0.95 0.88

Cash Expenditure Class 4,000- 6,000- 8,000- 10,000- 25,000- 40,0005,999 7,999 9,999 24,999 39,999 and Over

TABLE 39 Tanzania: Fractiona l Shares of Household Total and Food Consumption Purchased for Money by Household Expenditure Class, Rural, 1976/77.

0.52 0.33 0.19 0.96 0.09 0.98 0.14 0.38 0.20 0.43 0.81 0.91 0.32 0.93

Total

0.32 0.31 0.16 1.00 0.05 0.98 0.26 0.56 0.16 0.40 0.48 0.88 0.28 0.99

0.77 0.68 0.54 0.97 0.35 1.00 0.55 0.88 0.57 0.77 0.94 0.90 0.51 1.00

0.52 0.43 0.28 0.95 0.14 0.99 0.39 0.72 0.26 0.48 0.78 0.67 0.06 0.99 0.92 0.87 0.71 1.00 0.68 1.00 0.85 0.97 0.83 0.88 0.99 0.98 0.99 1.00

0.95 0.93 0.86 0.99 0.79 1.00 0.86 0.98 0.93 0.92 0.97 0.96 0.98 1.00

6,0007,999 0.98 0.96 0.92 1.00 0.83 1.00 0.96 0.99 0.96 0.91 0.99 0.98 0.99 1.00

8,0009,999

Cash Expenditure Class

4,0005,999

Source: Compute d from Household Budget Survey, 1976/77 , Tables 3C and 3D.

Total consumption Food and drink consumption Cereals Cereal products Starchy roots and starches Sugar and sweets Pulses, dry Nuts Vegetables Fruits Meat, meat products, poultry Fish and shellfish Milk and dairy products Oils and fats

0-999

2,0003,999

1,0001,999 0.99 0.98 0.95 1.00 0.94 1.00 0.97 1.00 0.97 0.95 0.99 0.98 1.00 1.00

1.00 0.99 0.97 1.00 0.96 1.00 0.97 1.00 0.99 0.92 1.00 1.00 1.00 1.00

0.99 0.96 0.90 0.99 0.88 1.00 0.93 1.00 0.98 0.87 0.98 1.00 1.00 1.00

0.94 0.88 0.76 0.99 0.64 1.00 0.83 0.97 0.85 0.90 0.97 0.95 0.94 1.00

10,000- 25,000- 40,00024,999 39,999 and Over Total

TABLE 40 Tanzania: Fractiona l Shares of Household Total and Food Consumption Purchased for Money by Household Expenditure Class, Urban, 1976/77.

Profile of Incomes and Poverty in Tanzania 10

3

very small share of their basic foods, although they purchase 31 percent of their total food and drink consumption. Tables 41 and 42 present an even more vivid demonstration of the importance of subsistence consumption of foods for poor household in both rural and urban areas. The tables exhibit the daily calories per capita that are obtained from cash purchases o r subsistence production . I n Table 39 , i t wa s see n tha t the poorest rural household s obtaine d onl y 1 2 percen t o f thei r tota l foo d consumption through cash purchases. From Table 41, it appears that the corresponding shar e of tota l calorie s consume d i s onl y 6. 9 percent . I n other words, for th e poores t rural groups , onl y a minuscul e portio n o f thei r tota l dail y calori e intak e i s obtained for cash. Surprisingly, the same appears to be the case with poor urban groups. For the two poorest urban classes, over 70 percent of their daily calori e intake is from subsistence production. Whether this comes from own-production or foo d remittance s fro m rura l relative s i s no t specified . I t seem s doubtful , however, that poor urban households could rely for as much as 80 percent of their calorie consumption on remittances. The share of calories obtained for cash in rural areas can be seen from Table 41 to be almost always lower than the share of total food and drink obtained with cash, as reported in Table 39. The same seems to hold for the urban areas, when one compare s Tabl e 4 2 wit h Tabl e 40 . Ther e i s significan t differentiation , however, betwee n rura l an d urban areas i n the proportion o f calorie s obtaine d from the market. For any given household cash expenditure category, the share of calorie s obtaine d fo r cas h i s alway s substantiall y highe r i n the urban areas. On average, rural households obtain only 20.9 percent of their calories from the market, while for the average urban household, the share is 80. 8 percent . POVERTY LINE AND EXTENT OF POVERTY Given the exposition o f incom e an d expenditure distributio n statistic s an d the analysis of consumption patterns of the previous sections, this section attempts to estimate a poverty line circa 1976/7 7 and to estimate the incidence of poverty in rural an d urban mainland Tanzania . Althoug h th e yea r 1976/7 7 i s quit e fa r removed from current conditions, it is the only year for which a detailed national household budget surve y i s available . A n analysis o f povert y i n that year wil l reveal structura l pattern s tha t ca n b e use d t o updat e th e povert y lin e i n late r years, given information o n prices. An earl y analysi s o f povert y i n Tanzania wa s don e by th e ILO mission (ILO 1982), which calculated a poverty line by costing three different subsistence diets for 198 0 an d supplementin g thes e wit h som e nonsurve y informatio n abou t nonfood costs. It was thus calculated that the basic needs income (BNl) circa 1980

Share of total calorie intake (%)

Total

Monetary Cereals Cereal products Starchy roots and starches Sugar and sweets Pulses, dry Nuts Vegetables Fruits Meat, meat products, poultry Fish and shellfish Milk and dairy products Oils and fats

6.90 10.6

0 22.7

0 32.2

0 35.4

0 43.2

105.50 22.22 10.25 29.87

328.32 14.79 55.11 121.64 57.52 34.71 6.45 8.11

51.42 20.68 5.18 11.82 99.20 18.67 1.38 25.42 103.98 28.38 39.54 41.98

0 38.7

0 56.0

0 35.0

0 20.9

0

4

218.94 5.31 31.60 46.27 36.96 15.84 3.08 3.54 386.47 6.29 21.89 81.31 78.53 3.51 7.71 5.86 291.71 10.84 285.19 146.32 77.81 43.50 21.51 10.55

Total

6 794.4 8 1,101.2 9 736.2 5 450.6

96.43 22.55 18.25 38.07

548.31 19.07 65.83 82.84 64.59 26.15 7.54 8.34

4 997.9

86.33 23.11 9.12 20.24

85.80 27.72 5.39 15.29 9 797.2

396.74 11.97 82.70 71.54 63.79 19.91 5.90 5.89

342.64 8.52 47.89 72.40 57.17 20.77 6.26 5.75

4 695.5

49.33 22.31 4.32 7.51

231.91 5.23 36.52 44.86 34.94 17.38 3.14 3.09

2,0003,999

3 460.5

29.18 17.14 1.83 12.13

18.22 13.12 1.09 4.73

153.17 245.1

109.40 1.55 10.24 27.16 21.33 11.37 1.72 2.09

1,0001,999

76.08 0.00 5.25 14.36 12.44 5.86 0.68 1.35

0-999

Cash Expenditure Class (Tsh) 4,000- 6,000- 8,000- 10,000- 25,000- 40,0007,999 9,999 24,999 39,999 and Over 5,999

TABLE 41 Tanzania: Dail y Pe r Capita Calorie Intake of Variou s Foods i n Rural Households fro m Subsistence an d Purchases, 1976/77, by Expenditure Class (kilocalories/capita/year) .

Source: Compute d by authors.

Share of total calorie intake (%) 93.1

Total

2 1 7 7 5 3 0 4 4 6 0 9

962.74 892.7 0.24 0.0 295.22 299.4 0.47 3.2 149.12 77.3 8.98 12.5 8.54 6.4 5.20 6.3 6.27 3.6 0.68 1.4 18.71 11.4 0.89 0.0

10.68 11.4 0.74 0.2 13.53 14.4 2.42 30.7

708.72 623.9 0.48 3.7 371.84 77.1 1.00 17.4 123.24 68.2 12.74 10.0 7.06 5.7 4.07 1.2

0 89.4

0 77.3

0 67.8

0

64.60 56.8

0 61.3

0

16.33 12.4 1.22 1.9 0.07 10.9 3.58 0.9

919.27 1,097.3 0.00 0.2 153.51 313.6 5.26 0.9 251.80 226.4 10.30 19.8 2.45 12.6 4.61 4.7

4 9 8 3

0 2 7 7 3 7 4 5

0 79.1

0

8 1,368.38 1,702.2 0

9 3 5 0

4 8 7 9 9 3 4 6

44.00 65.0

2,071.46 2,075.3 9 1,572.16 1,466.30 1,457.09 1,314.6 8 1,256.51 864.5

Subsistence Cereals 1,257.62 1,324.83 1,006.00 1,057.92 Cereal products 0.00 0.43 0.11 0.10 Starchy roots and starches; 328.8 3 328.52 339.04 215.61 Sugar and sweets 0.94 1.26 0.80 0.07 Pulses, dry 392.66 351.28 157.57 148.67 Nuts 43.45 23.02 18.92 11.84 12.27 10.24 Vegetables 13.37 15.63 Fruits 3.53 4.76 5.34 4.40 Meat, meat products, poultry 9.94 11.71 17.15 10.70 Fish and shellfish 1.15 2.34 2.53 1.97 10.04 Milk and dairy products 18.98 11.71 4.15 Oils and fats 1.00 1.57 0.71 0.62

Share of total calorie intake (%)

Total

20.30 31.5

0 55.6

113.65 32.24 1.75 30.80

100.96 37.16 2.90 24.60

94.18 44.86 4.42 22.40

144.03 40.64 3.89 60.30

880.85 1,000.66 173.09 224.76 90.29 123.88 172.21 216.88 121.19 150.05 232.56 270.12 21.24 29.31 13.79 18.96

767.10 340.89 82.84 148.37 120.33 198.57 19.69 10.92

610.73 99.37 86.60 124.50 102.76 124.65 15.08 6.90

778.25 171.79 135.20 263.77 172.26 50.25 30.13 24.66 158.73 15.61 15.88 185.58

938.85 278.02 94.23 224.36 132.14 203.31 37.84 29.37 192.69 29.22 7.83 133.00

102.34 33.31 5.57 34.44

705.05 155.11 91.16 142.21 106.33 158.57 18.41 11.31

Total

0 79.8

0 89.7

0 94.2

0 96.5

0 98.0

0 88.5

0

80.80

5 1,336.4 6 1,854.3 3 1,883.6 7 2,283.5 0 2,300.8 6 2,002.1 1 1,563.8 0

71.55 25.68 4.90 10.70

38.61 12.74 0.75 0.00 0 912.6

454.70 42.87 81.26 76.23 61.94 67.77 10.32 4.73

2,0003,999

277.43 17.60 29.27 49.72 35.88 49.36 3.05 2.18

1,0001,999

313.93 516.6

Monetary Cereals 148.07 Cereal products 17.76 Starchy roots and starches 5.09 Sugar and sweets 38.21 Pulses, dry 22.99 Nuts 44.40 Vegetables 1.50 Fruits 1.53 Meat, meat products, poultry 22.86 Fish and shellfish 9.20 Milk and dairy products 2.32 Oils and fats 0.00

0-999

Cash Expenditure Class (Tsh) 4,000- 6,000- 8,000- 10,000- 25,000- 40,0007,999 9,999 24,999 39,999 and Over 5,999

TABLE 42 Tanzania: Dail y Pe r Capita Calorie Intak e of Variou s Foods i n Urban Households fro m Subsistenc e an d Purchases, 1976/77, by Expenditure Class (kilocalories/capita/day) .

Source: Compute d by authors.

Share of total calorie intake (%) 79.7

Total

4.33 2.90 4.74 0.03

10.94 6.32 12.65 0.00

0 68.5

0 44.4

10.30 5.8

203.23 116.7

339.22 0

3 4 1 2

3.04 0.8 1.42 0.6 0.06 0.0 0.06 0.0

1.34 0.97 0.06 0.02

0

4

9 4 6 0 3 3 9 0

145.16 86.4 3.17 0.5 26.68 18.6 0.67 0.0 19.39 5.2 1.16 1.9 1.52 0.9 0.92 1.4

265.09 0.19 47.96 0.03 17.91 1.56 3.13 0.95

0 20.2

728.54

479.07 1.26 170.08 0.01 51.47 5.31 7.90 1.44

814.78 0.99 200.12 0.31 55.24 8.65 8.74 2.37

1,231.80 1,121.11

Subsistence Cereals 983.33 Cereal products 0.05 Starchy roots and starches 124.83 Sugar and sweets 0.93 Pulses, dry 65.72 Nuts 15.29 Vegetables 7.60 Fruits 2.29 Meat, meat products, poultry 24.62 Fish and shellfish 1.23 Milk and dairy products 5.90 Oils and fats 0.00

3.50 2.0

82.53 46.4

0.94 0.6 0.88 0.0 0.00 0.0 0.26 0.0

61.31 33.5 0.18 0.1 11.02 4.7 0.37 0.0 5.20 3.8 0.68 0.0 0.75 0.5 0.94 2.7

0

2

5 0 0 7

8 3 7 3 9 4 2 3

11.50 19.2

259.20 371.7

3.25 3.0 0.00 1.7 0.00 0.3 0.20 0.0

207.58 272.1 2.39 0.8 28.30 63.7 0.44 0.2 12.77 22.2 0.11 2.6 0.61 3.2 3.54 1.2

0

5

8 6 5 8

1 9 6 5 2 9 9 7

108 Alexander

H. Sarris and Rogier van den Brink

was 600 Tsh per month for a family o f five, o r 1,440 Ts h per capita per annum. On the basis of that figure an d aggregate income statistics, it was estimated that about 1 5 percent of urban households and 25-30 percent of rural households fel l below BN I around 1980 , a rather mild degree of poverty. I t will be see n below , however, that these figures wer e most likely large underestimates of the degree of poverty . Two methodologie s ar e followe d here . Th e firs t on e wil l conside r calori e intake and relate it to food expenditure an d total expenditures. The second wil l consider the relation of foo d expenditure s t o total incom e o r total consumption expenditures. Th e relatio n o f calori e intak e t o tota l foo d expenditure s (bot h monetary an d subsistence ) wa s propose d an d use d b y Gree r an d Thorbeck e (1986) t o derive a food povert y lin e usin g cross-section , household-leve l data . In the case a t hand, we did not have information a t the level o f households, but only b y cash expenditur e classe s o f households , a s alread y exhibite d i n th e previous sections. We therefore used the interval means from the aggregated data for the analysis. The first ste p in the analysis involve s relatin g per capita total calorie intak e to per capita total expenditure and per capita total food expenditure. The following relationships wer e estimated. (4.1) (4.2) where TCALPC i s tota l calorie s consume d pe r capita ; PCFE is tota l (monetar y and subsistence ) per capit a foo d consumptio n expenditures ; an d PCTE i s pe r capita total consumption expenditures (monetar y an d subsistence) excludin g savings. The dat a use d hav e bee n exhibite d i n th e previou s tw o sections . Th e OLS estimates o f th e abov e equation s ar e exhibited i n Table 43. I t can be see n that there doe s no t appea r to b e an y significan t relatio n between pe r capita calori e intake and food expenditure s o r total per capita expenditures in rural areas, but that a significan t an d positiv e relatio n appear s t o exis t fo r urba n households . Several other functional forms were estimated, but the results were quite similar. The elasticit y o f pe r capita calori e intak e wit h respect t o foo d expenditure s i n the urba n areas , fro m th e estimate d equations , i s 0.34 , whil e th e elasticit y o f calorie intak e wit h respec t t o tota l expenditure s i s 0.17 . B y solvin g equation s (4.1) an d (4.2) fo r per capita food, o r total, expenditure consisten t with a given per capita calorie level , w e ca n obtain a n estimate o f a possible food , o r total, poverty line, that is, a level of per capita expenditure consistent with a given per

Profile of Incomes and Poverty in Tanzania 1 0

9

capita calorie intake . The results of thes e calculations fo r the urban areas are exhibited in Table 44 for five different levels of per capita calorie intake. Based on FAO/WHO energy recommendations and the age and sex composition of th e Tanzanian population, it has been estimated that the average daily per TABLE 43

Tanzania: Econometri c Estimates of Calorie Expenditure Cross-Section Relations, 1976/77. Independent Variables Dependent Variable Rural

TCALPC TCALPC

Urban

TCALPC

TCALPC

Constant

In PCFE

3886.8 (1.623)a 3423.7 (3.950)

-263.4 (-0.722)

-2523.0 (-4.410) -486.3 (-1.098)

663.6 (7.833)

7 ? 2b

N.O.c

-0.064

9

0.124

9

0.883

9

In PCTE

-179.6 (-1.461)

331.6 (5.533)

0.787

Source: Compute d by authors. a Figure s in parenthesis denot e t-statistics . b Adjuste d R-squared. c Numbe r of observations .

TABLE 44

Tanzania: Pe r Capita Food and Total Expenditures in Urban Areas Consistent with a Given Per Capita Calorie Intake in 1976/77. Per Capita Calorie Intake (kilocaloriesl capital day) 1,800 2,100 1,900 2,000

(1) Per capita food expenditur e (Tsh/annum) (2) Per capita total expenditure (Tsh/annum) Ratio (l)/(2 ) Source: Compute d by authors.

2,200

675

785

912

1,061

1,233

987

1,335

1,804

2,439

3,298

0.68

0.59

0.51

0.44

0.37

110 Alexander

H. Sarris and Rogier van den Brink

capita calori e requiremen t circ a 198 0 was 2,22 9 kilocalorie s (Tanzani a 1987b , Agriculture an d Livestock) . O n th e basi s o f th e dat a i n Table s 3 7 an d 38 , i t appears tha t almos t ever y rura l household , irrespectiv e o f expenditur e class , achieves o r come s clos e t o thi s level , whil e onl y th e thre e highes t househol d expenditure classe s i n urba n area s achiev e th e standard . I t migh t wel l b e tha t calorie requirement s i n rural area s are , on average , highe r than in urban areas. But eve n i f w e arbitraril y adop t a n urba n minimu m calori e standar d o f 2,00 0 kilocalories, i t is only abou t 48 percen t of urba n households (cf . Table s 38 and 31), thos e with cas h househol d expenditure s abov e 6,00 0 Tsh , tha t achiev e it . These ar e th e household s tha t exhibi t averag e pe r capit a dail y calori e intak e above 2,000 kilocalories i n Table 38. Notice, however , tha t th e secon d lin e i n Tabl e 4 4 indicate s tha t dail y pe r capita calorie consumption of 2,000 kilocalories is consistent with an annual per capita expenditure level of 1,80 4 Tsh. From Table 31, it appears that this average per capita expenditur e leve l i s achieve d i n the urban areas only b y household s with total cash expenditures of 8,000 Tsh and above, or only about 35 percent of urban households . Base d o n thes e consideration s on e mus t conclud e tha t i n 1976/77 abou t 50-60 percen t o f urba n households coul d be classifie d a s poor. This figure i s substantially highe r than what was estimated by ILO for 1980 , but given tha t 1976/7 7 wa s a muc h bette r yea r i n Tanzani a tha n 1980 , th e IL O underestimate i s bound to be even greater. To chec k th e consistenc y o f th e abov e estimate s an d t o obtai n a povert y estimate for rural areas also, we use the second methodology, whic h relates per capita foo d expenditure s t o pe r capita tota l expenditure s an d tota l income , a s reported fo r th e respectiv e interval s i n th e 1976/7 7 HBS . For thi s purpos e w e estimated by OLS the following tw o sets of equations: (4.3) (4.4) where PCFE and PCTE are as defined earlier , and PCTY is per capita total incom e (including savings) . Table 45 exhibits the estimates of parameters a an d p. The results indicate significant and positive relations between per capita food expenditures an d total expenditures , a s wel l a s tota l income , i n both rura l and urban areas, with the best fits between per capita food expenditures and per capita total consumption expenditures. The good fit in the rural areas, when contrasted with the poor, earlier fit between per capita calorie intake and food expenditures, indicates that as total per capita expenditures increase in rural areas, the composition o f foo d intak e change s i n favor of higher-qualit y foods , eve n though the

Profile of Incomes and Poverty in Tanzania 11

1

TABLE 45 Tanzania: Econometri c Estimate s o f Relations Between Tota l and Food Expenditures. Dependent Variable Rural

In PCT E In PCT E

Urban

In PCT E

In PCT E

Constant

Independent Variables In PCTE In

4.094 (12.502)a 5.205 (16.520)

0.350 (7.534)

2.929 (13.221) 3.216 (10.574)

0.519 (17.300)

PCTY

0.188 (4.316)

0.475 (11.668)

K2b

N.O.c

0.875

9

0.688

9

0.974

9

0.944

9

Source: Compute d by authors. a Figure s in parenthesis denote t-statistics. b Adjuste d R-squared. c Numbe r of observations .

total calorie intake stays roughly constant/This is noticeable i n Table 35, where it can be see n that as household cash expenditures increas e in rural areas, more sugar, fruits, and meats are consumed. Within the starchy root component, it can be seen that as expenditures rise, a substitution away from cassava and into sweet potatoes takes place, while within cereals, rice consumption rises at the expense of maize. The elasticit y o f pe r capita foo d expenditure s wit h respec t t o tota l pe r capita consumption expenditure s i s 0.3 5 i n rura l an d 0.52 i n urba n areas , wherea s th e income elasticities of food expenditures are lower, at 0.19 i n rural and 0.48 i n the urban areas. These figures are consistent with growing savings as income rises. The estimated equations can be used to derive a level of poverty by using the ratio o f foo d expenditur e t o tota l expenditur e o r tota l incom e a s th e variabl e determining poverty. Notice in Table 44 that, using the calorie method, the ratio of foo d expenditur e t o tota l expenditur e fo r a household wit h mean per capita daily calori e intak e o f 2,00 0 kilocalories , i n urba n areas , i s 0.51 . Tabl e 34 , however, indicate s tha t i t i s onl y th e tw o highes t cas h expenditur e classe s i n urban areas that achieve a ratio as low or lower, while all the others have a much higher ratio. Among rural households, no class achieves a ratio as low as this, as can be seen from Table 33. We can use different values of this ratio to derive from the estimated equations corresponding poverty line s in rural and urban areas. Notice that equation (4.3 ) can be written as follows:

112 Alexander

H. Sarris and Rogier van den Brink (4.5)

where (4.6) and exp (• ) denotes the exponential function . Usin g alternativ e value s o f p , w e can estimate the corresponding values for per capita total expenditures. Table 46 exhibits the results of suc h calculations. The result s ar e interestin g i n man y ways . First , notic e tha t th e estimate d poverty lines are not very different betwee n rural and urban households. This is probably s o because suc h a large shar e of consumption a t low levels of incom e in bot h rural an d urban area s i s ou t o f subsistenc e production . I n fact, fo r th e three highest ratios of food to total expenditures, the rural poverty line is abov e the urban poverty line. As shown in Table 44, in urban areas, a per capita daily calorie intake of 2,00 0 kilocalories implie d a ratio of food t o total expenditure of 0.51 an d a per capita poverty level of 1,80 4 Tsh. In fact equation (4.5) implies a poverty level fo r p = 0.51 o f 1,78 9 Tsh , whic h i s quit e clos e t o wha t wa s estimate d with th e firs t method. Thi s implie s tha t th e tw o method s giv e compatibl e results , an d th e definition of a poverty level depends on what one assumes about either minimum daily per capita calorie intake or the ratio of food expenditures to total expenditures. We shall use the latter method, as it indicates poverty levels for both rural and urban households. Since it appears that the estimated poverty lines are quite similar in rural and urban areas, we can use Table 30, showing the distribution of per capit a incomes in Tanzania. Although th e information se t out in Table 30 aggregates rural and urban households, it is the only table we have obtained that specifically consider s TABLE 46 Tanzania: Povert y Levels Estimate d According t o the Ratio of Food to Total Expenditure (Tsh/capita/annum) . Ratio of Food to Total Per Capita Expenditure 0.50 0.55 0.60 0.65 0.70 0.75 0.80 Total expenditure Rural households Urban households Source: Compute d by authors .

1,579 1,364 1,193 1,055 1,864 1,529 1,276 1,080

941 926

846 802

766 702

Profile of Incomes and Poverty in Tanzania 11 3 per capita, and not per household, incomes or expenditures. We use simple linear interpolation to estimate the proportion of households or people in poverty when the poverty lin e falls a t a given income interval. Notice that Table 30 describe s the distribution of per capita incomes and not expenditures. We could have used the secon d se t of regressions , exhibite d i n Table 45, t o relate the ratio of foo d expenditures to total income via a procedure similar to what is indicated in (4.5). However, w e believe that the expenditure dat a are less biased, especially a t the high levels and hence the resulting regressions are more reliable. Second, as was exhibited i n Tabl e 31 , wit h th e exceptio n o f ver y hig h level s o f incom e an d expenditure, household total expenditures are quite close to household incomes. Hence th e bias arisin g fro m ou r procedure i s smal l and , if anything , i t under estimates the degree of poverty. Table 47 exhibits the estimates of the percentage of households and people in Tanzania mainlan d circ a 1976/7 7 tha t wer e livin g i n poverty. W e hav e use d a weighted averag e poverty leve l fro m Table 46. The weights ar e the proportions of population living in rural and urban areas, which, using figures from the earlier section ("Populatio n an d Some Househol d Characteristics" ) ar e 0.87 an d 0.13, respectively. The results indicate very high levels of poverty even for extreme assumptions. For instance, eve n if w e assum e a definition o f povert y accordin g to which the ratio o f foo d t o tota l expenditure s i s 0.8 , a n admittedl y extrem e situation , 3 6 percent of households or 55 percent of people would have been classified as poor in 1976/77 . Althoug h i n rural areas this woul d no t imply sever e calori e under nutrition, it would mean very severe undernutrition in urban areas. The IL O mission, i n it s estimate s o f th e BNI , use d a rati o o f foo d t o tota l expenditure o f 0.66 . A s show n i n Tabl e 47 , i t appear s tha t fo r thi s ratio , 56. 3 TABLE 47 Tanzania: Percentag e o f People an d Households i n Poverty i n 1976/7 7 Usin g Poverty Line s According t o the Ratio of Food to Total Expenditure. Ratio of Food to Total Expenditure 0.50 0.55 0.60 0.65 0.70 0.75 0.80 National poverty line (Tsh/capita/year) 1,61 Percentage of households below poverty line 73. Percentage of people below poverty line 81. Source: Compute d by authors.

6 1,38 5 1,20 4 1,05 8 93

9 84

0 75

8

2 66.

2 60.

8 56.

3 49. 9 42.

3 36.

0

6 75.

5 70.

8 67.

0 60. 2 51.

8 44.

8

114 Alexander

H. Sarris and Rogier van den Brink

percent of households an d 67 percent of people in Tanzania would be classifie d as poor. These poverty level s ar e clearly much higher than what was estimate d by th e IL O mission, whic h di d not have the results of the full 1976/7 7 HB S and used aggregate data . Taking ou r estimated national per capita total expenditur e of 1,05 8 Tsh , whic h correspond s t o a food-to-total-expenditur e rati o o f 0.6 5 (similar t o th e IL O assumption), an d using th e Nationa l Consume r Pric e Inde x (NCPl) average d ove r 1980/81 , w e obtai n tha t th e correspondin g pe r capit a poverty lin e i n 198 0 (th e yea r o f th e IL O estimate) i s abou t 1,87 1 Tsh , o r 3 0 percent higher, than the 1,44 0 Tsh that the ILO had estimated. If their estimate is deflated t o 1976/77 , i t yield s a pe r capit a povert y lin e o f 81 4 Tsh , which , according to Table 47, would imply that about 40 percent of households an d 49 percent of peopl e wer e living i n poverty. Clearl y then, poverty i s substantia l i n Tanzania an d ca n b e grossl y underestimate d usin g aggregat e nationa l incom e data. To obtai n a n ide a o f wha t constitute d povert y i n 1989 , w e hav e use d th e component pric e serie s o f th e NCPl , whic h give s a n inde x fo r eac h o f th e 1 0 categories o f consume r good s indicate d i n Table s 3 3 an d 34 , an d th e group specific weight s indicate d i n those table s fo r th e fiv e lowes t cas h expenditur e classes. Th e average per capita total annua l expenditure i n any of thes e classe s in 1976/77 , rural or urban, does not exceed 1,20 8 Tsh , which implies (base d on the figures i n Table 47) tha t we are examining th e consumption pattern s of th e five poores t classes of rural and urban households. Using this methodology, w e computed th e correspondin g pe r capit a tota l annua l expenditur e i n 198 9 tha t would b e equivalen t i n purchasin g powe r t o th e averag e pe r capit a annua l expenditure in each of the five lowest 1976/7 7 classes. The results are indicated in Table 48. Assuming that the highest interval exhibited in Table 48 illustrate s the poverty threshold, and using an average household size of 5.3 (from the 1988 census), it follows tha t a household would be classified a s poor in 198 9 if it had monthly tota l (monetar y an d subsistence) incom e o f les s tha n 8,25 8 Ts h in the rural areas and less than 8,885 Tsh in the urban areas. For reference, in 1987/88 , 89 percent o f al l civil servant s had monthly salarie s o f 3,61 0 Ts h (Level M S 3) or less, and the highest grad e civil servant s (M S 17-19) made , on average, onl y 7,980 Tsh. It is quite clear that a 1989 household headed by a civil servant needed other sources of income it were not to be poor. CONCLUSIONS

The upsho t o f th e analysi s i n thi s chapte r is , first , tha t povert y i n 1976/77 , a relatively goo d year from a macroeconomic an d agricultural production standpoint, was extensive, an d much more so than has been previously estimated . It

Profile of Incomes and Poverty in Tanzania 11 5 TABLE 48 Tanzania: Povert y Line s i n 1989 , Household Cas h Expenditure Categor y i n 1976/77 (Tsh/capita/annum) . 0-999

1,0001,999

2,0003,999

4,0005,999

6,0007,999

Average per capita total expenditure, 1976/7 7 Rural Urban

680 508

763 587

781 784

958 1,063

1,103 1,208

Average per capita total expenditure, 198 9 Rural Urban

11,720 8,756

13,130 10,056

13,352 13,267

16,359 17,587

18,698 20,118

Source: Compute d by authors. appears that in that year poverty wa s quite widespread, an d hence that incom e distribution was quite even. A major factor in this evenness was the prevalence of subsistenc e consumptio n i n the rural areas but also among the urban poor. It als o appeare d fro m ou r analysi s tha t inequalit y wa s greate r i n th e urba n areas tha n i n th e rura l ones . Give n tha t a large shar e o f th e income s o f urba n middle- and upper-class, and rural nonfarm, household s come s from wage s and salaries, much of that from the formal sector , it seems that changes in the wider public sector performance can have a significant impac t on these households, as will be demonstrated later.

I 5 PERFORMANCE OF AGRICULTURE

In order to evaluate the performance of the Tanzanian economy before and after adjustment, a correct assessmen t o f agricultura l growt h i s crucia l give n th e large share of agriculture in GDP. Ever since the mid-1970s, Tanzania's agricultural secto r has—accordin g t o officia l estimates—grow n a t rates abov e thos e of the nonagricultural sector and has increased its share in the GDP at the expense of the nonagricultural sector s (se e Tables 2 and 9). Such a pattern runs counter to wha t i s usuall y considere d a "normal " patter n o f economi c growth . Thi s atypical growt h o f th e agricultura l secto r i s attribute d t o increase s i n th e production o f foo d crops , sinc e stagnation ha s characterize d th e performanc e of export crops. In this chapter we start with a brief overview of the changes in the institutional setting of agriculture over the last three decades. We then present an analysis of the trend s i n foo d an d expor t cro p productio n an d sho w tha t th e clai m o f consistent expansio n in food cro p production i s not consistent wit h other types of information. We also show that agrarian structure and technology seem to have stayed remarkably constant during the period of institutional upheaval and crisis. Finally, although we will argue that agricultural producers were clearly affecte d negatively b y th e stagnatio n an d deterioratio n o f Tanzania' s econom y befor e

116

Performance of Agriculture11 7 adjustment, some of the main potential benefits of structural adjustment were not effectively transmitte d to the majority of agricultural producers. THE INSTITUTIONAL SETTING

Tanzania's agricultural sector performed remarkably well in the 1950s and the early 1960s . However , th e adven t o f a state-controlle d econom y afte r th e Arusha declaration in 196 7 created an environment that was generally repressive for agriculture. The socialist policies adopted turned against private production and marketing of export crops and substituted for them state farms and monopsonistic marketing boards. The inefficient operatio n of the various agricultural parastatals virtually decimate d Tanzania's expor t crop sector. Given that agricultural export s accounted for approximately 8 0 percent of total exports, the collapse of the export crop sector can be seen as a major endogenous cause of the macroeconomic crisis of the early 1980s. Tanzania's peasant far m secto r (a s oppose d t o estat e plantation an d other large-scale operations ) provide s aroun d 8 5 percen t o r more o f th e followin g major export crops: coffee, cotton, cashews, tobacco, and pyrethrum. Additionally, peasant farms are responsible for approximately 25 percent of tea production, 50 percent of the officially markete d rice production, and virtually all of the legally marketed maize production. During the 1950s and 1960s, this peasant sector expanded its share in agricultural exports considerably and dominated the country's export performance (World Bank 1983). However, the existence of a considerable numbe r o f plantation s an d large-scal e far m enterprise s give s Tanzania a diverse mix of agricultural producers. This typical mix of peasant and estate farming was basically already in place by the end of German colonial rule. At independence, the marketing structure was characterized by a combination of private traders and a strongly emerging cooperative structure, much as had been th e cas e earlie r i n th e century . Afte r independence , th e "improvemen t approach," as defined in the First Five-Year Plan of the newly independent state, strongly emphasize d the cooperative movement , probably partially becaus e it was see n a s a n appropriate countervailin g powe r agains t "non-African " elements, namely, Tanzanians of Indian and Arab origin. Partly as a result of high prices during the drought of the 1960/61 season, the government decreed in 1962 the Agricultural Products Control and Marketing Act, which defined wha t became known as the three-tier single-channel marketing system. At the apex level, crop-specific marketing boards were established, which were responsible for the final sal e of agricultural produce. The marketing boards bought from regional cooperative unions, who i n tur n wer e supplie d b y primary cooperatives or

118 Alexander

H. Sarris and Rogier van den Brink

directly by private producers. Crops falling under this regime had their prices set by th e governmen t an d wer e calle d "scheduled " crops . The y include d maize , paddy, wheat , oilseeds , cashe w nuts , and cotton. All bu t cotton were markete d by the National Agricultural Products Board (NAPB). Final marketing of all other crops (i.e., cotton, tobacco, coffee, pyrethrum, sisal, and tea) took place through crop-specific marketin g boards , whic h derive d produce r price s fro m actua l export sales. Under th e three-tie r single-channe l marketin g system , ther e existe d signifi cant geographical price differences du e to variable marketing costs. Pricing was only fixed with respect to the price that the marketing boards paid to the regional cooperative unions for the final delivery of produce (the "into-store" price of the marketing board). Unions an d co-ops deducte d their (officially approved ) marketing costs (e.g., transportation) and farmers received the residual. The vigo r with whic h th e governmen t promote d th e cooperativ e for m o f organization resulted, as earlier in the century, in a cooperative structur e which became increasingl y governmen t dominated . Especiall y with respec t t o th e domestic marketin g o f foo d crops , cooperativ e marketin g ha d no comparativ e marketing advantages , an d as a result the structur e had to be virtually impose d from above. Additionally, when registration standards for cooperatives relaxed , the quality, performance o f the cooperative organizations plummeted, and there were increased fraud and nonrepayment of credit . By 1973 , the three-tie r single-channe l marketin g syste m consiste d o f abou t 2,300 primar y co-op s (whic h usuall y combine d severa l village s an d markete d several crops), 20 regional cooperative unions, and agricultural marketing boards for coffee , oilseeds , cashe w nuts , cotton, pyrethrum , sisal , sugar , tea, tobacco , and cereals. In 1973, the government replaced the marketing boards with parastatal crop authorities , possibl y i n a n attemp t t o reduc e cooperativ e contro l ove r marketing. Th e NAP B wa s als o abolishe d i n 1973 , an d th e Nationa l Millin g Corporation (NMC), which until that time had been only involved in milling, took over the NAPB marketing functions . Whereas th e union s ha d specialize d geographicall y b y region , th e cro p authorities wer e specialize d b y crop , nationwide. I n the 1974/7 5 cro p season, the "into-store price," which had resulted in differential pricing at the regional level, was replaced by a "producer price," which became fixed for the whole nation (a "pan-territorial price") . Additionally, th e list o f schedule d crop s wa s extende d to includ e sorghum , bulrus h millet , finge r millet , an d cassava—th e so-calle d "drought crops. " Various pulse s wer e adde d to th e lis t late r in the 1970s . Th e pan-territorial pricin g policy , whic h affecte d producer s a s wel l a s consumers , was i n effec t fro m 197 4 t o 198 1 fo r al l schedule d crop s with th e exception o f coffee an d sisal.

Performance of Agriculture 11

9

The Village Development Act was passed in 1975 , after "Operation Sogeza" (planned villagization), an d the village became a legal person able to enter into contracts with other legal entities. All marketin g function s o f th e cooperativ e syste m wer e transferre d t o cro p authorities, whic h receive d considerabl e fundin g fro m wester n donors . A t th e same time, most private retail shops in rural areas were closed under "Operation Maduka," and Regional Tradin g Corporations become responsible fo r the retail distribution of food and consumer goods. Due to the general inefficiency o f the new system , consumer goods rationing soo n followed. Man y observers point to the induce d disincentive s fo r productio n o f suc h rationin g (see , fo r example , Bevan el al.1989). This combination of restrictions with respect to the marketing of agricultura l produc e an d supply o f consume r goods mus t have amplified th e attractiveness o f paralle l market s i n genera l an d illega l export/impor t i n particular. By 1983 , the parallel, open markets had become so active that the government, in a last-ditch effort to reassert its control over the economy, declared a "War on Economic Saboteurs, " that is, private traders . The effects o f increased controls and roadblocks temporarily reduced the quantities traded on the open market and caused open market prices to increase. In 1984 , a first se t o f liberalizatio n measure s too k effect . Fo r instance , th e amount of food grains allowed to be privately traded was raised to 500 kilograms, and the roadblocks were removed. Although the reinstated cooperatives formally held a marketing monopoly , th e government implicitl y allowe d trader s to dea l with villagers. In 1985 , th e governmen t relaxe d it s contro l ove r internationa l trad e b y implementing th e own-funde d impor t scheme . A s a result, privat e import s o f trucks jumped by 300 percent in 1986. Open markets for maize grew sharply, as transportation problem s decrease d whe n fue l an d spar e part s becam e mor e readily available . According t o estimates fo r Tandale market in Dar es Salaam , open-market sales doubled to over 50 percent of total quantity traded from 198 4 to 198 7 (Gordo n 1989) . After severa l increases , al l remainin g quantit y restriction s o n interregiona l private grai n trade were lifted whe n the permit syste m was abolished in March 1987. However , th e officia l marketin g channel s wer e stil l unde r obligatio n t o purchase al l quantitie s offere d fo r sale . Moreover , i n th e mor e remot e mai n maize-producing regions , th e officia l produce r price s wer e abov e th e open market produce r price , a s a result o f th e increase s i n quantity availabl e i n th e open marke t an d favorabl e weathe r condition s ove r th e 1985-198 7 period . During the 1984-198 7 period, NMC stocks of maize reached record levels (u p to about 200,000 tons ) with averag e increase s i n official purchase s o f 4 0 percen t

120 Alexander

H. Sards and Rogier van den Brink

per year , NMC' S domesti c maiz e sale s wer e depressed , becaus e th e officia l consumer price was higher than the open-market consumer price. In an effort t o decrease it s accumulate d stocks , th e NM C opened ove r 10 0 retai l shop s i n December 1988 , selling at below-open-market prices. In June 1988 , the NMC officially los t its monopoly o f th e grain trade. Offic ially, private traders were still not allowed to buy directly from primary cooperatives. However , th e private secto r di d not see m ver y intereste d i n purchasin g from the unions, given the price situation in the open markets. Thus, the unions continued to rely on the NMC to sell their produce. Both the unions and the NMC realized considerable losse s o n their marketing transactions i n the 1988/8 9 an d 1989/90 seasons . Moreover , i n Decembe r 1988 , th e governmen t wa s force d t o assume responsibility fo r the substantial overdraft s o f the NMC. In Septembe r 1989 , th e governmen t free d maiz e marketin g a t th e primar y cooperative level : privat e trader s wer e allowe d t o bu y directl y fro m primar y co-ops. B y 1990 , considerable foo d grai n stocks (approximatel y 250,00 0 tons ) were stranded, predominantly i n the regions of Arusha, Rukwa, and Shinyanga. NMC operations virtually came to a halt in 199 0 after the government prohibited it fro m undertakin g unprofitabl e operations . Th e financia l losse s o f th e NMC's and other parastatals had risen to such levels that they were posing a threat to the Tanzanian banking system . OFFICIAL SOURCES OF AGRICULTURAL DATA

An assessmen t o f agricultura l performanc e i n Tanzania ca n be base d o n thre e sources of information. First, several Tanzanian government agencies regularly publish regionally and nationally aggregated agricultural data. Second, one can use a number of studie s that contain agricultural production estimates obtaine d through nationally representativ e surveys . Third, one can obtain corroboratin g evidence fro m a numbe r o f indirec t sources ; fo r example , agricultura l pric e information, internationa l trad e statistics, health statistics, and the like. In this chapter, w e wil l explor e a number of thes e source s an d ascertain whether i t i s possible to arrive at a consistent picture of the performance o f agriculture ove r the last 25 years. Various officia l agricultura l outpu t serie s exist , produce d b y a numbe r o f government agencies. Sometimes the differences betwee n these series are minimal; sometimes the discrepancies ar e quite substantial. One of the problems fo r users o f suc h statistic s i s tha t i t i s no t eas y t o discer n fro m th e publication s exactly how the different output series were generated. The publications are often rather cursor y o n methodologica l information . Onl y throug h direc t interview s

Performance of Agriculture 12

1

with the relevant officers i n the government can one ascertain how certain series are estimated. At present , th e situatio n doe s no t see m t o hav e improved . Th e Burea u o f Statistics produces its ow n agricultural statistic s o n food crops , derived from a nationally representativ e sampl e surve y undertake n b y th e bureau i n 1986/87 . However, for the computation of the national accounts, the bureau does not use its own information, but uses instead production estimates from the Early Warning an d Crop Monitoring Burea u (EWCMB ) of th e Ministry o f Agriculture . Th e EWCMB estimates food production on the basis of a yield estimate obtained from a physiological cro p model that uses rainfall dat a and certain agronomic parameters. The yield estimat e i s combined with area estimates fro m th e agricultura l extension agents to obtain total production. The main pitfall of the model is that it i s no t regularl y calibrate d t o objectiv e productio n estimate s i n th e field . Nonetheless, th e EWCM B estimates ar e widel y accepte d a s th e bes t estimate s available and are therefore used to construct the national accounts. The Marketing Development Bureau of the Ministry of Agriculture also accepts and reports the EWCMB estimates . All th e different agencies , includin g th e EWCMB, rely o n the estimates fro m the Villag e Extensio n Officer s an d th e Distric t an d Regiona l Agricultur e an d Livestock Developmen t Officer s (RALDOs) . Thes e agent s d o no t emplo y objectifiable method s o f area , yield , an d productio n measurement . Give n th e nature of some of the major farming systems in Tanzania (viz., the savannah and semiarid systems), it is likely that the agents have a reasonable grasp of year-toyear qualitativ e change s i n overal l productio n level s fo r a give n villag e ("higher", "lower," "about the same") but a less-reliable estimate of quantitative changes in area and yields per hectare. Additionally, the various agencies employ different method s to compile or "correct" the RALDO estimates. A secon d source of error might be that the initial evaluation of the extensio n officer i s usuall y presente d t o th e villag e authoritie s befor e i t i s officiall y reported. Villag e authoritie s ma y hav e a n interest i n influencin g th e estimate s upward, sinc e th e party award s prizes fo r record agricultura l production . Simi larly, the government has issued a number of decrees that might have influence d the reliabilit y o f th e are a an d productio n estimates , particularl y durin g th e Ujamaa period. Thus, the Rural Lands (Planning Utilization) Act of 197 3 established certain rules and bylaws stating minimum and compulsory areas for certain crops. At times, stringent fines or imprisonment have been imposed on defaulters (Mtetewaunga 1986) . Th e compulsory minimu m acreag e ma y hav e produced a phantom growt h o f thos e crop s i n a n understandabl e effor t t o appeas e th e authorities. Th e growth in per capita food cro p production recorde d during the

1 2 2 Alexander

H. Sarris and Rogier van den Brink

height o f Ujama a policies , (1973-1976) , ma y b e partiall y th e resul t o f suc h politicized production estimates. A FIRST ASSESSMENT BASED ON OFFICIA L STATISTICS

Using th e officia l agricultura l statistic s a s a reference, th e followin g broa d picture of agricultural performance ove r the last 25 years emerges. Per capita production o f foo d an d expor t crop s i s graphe d i n Figur e 4 fo r th e perio d 1966-1989 usin g Divisia (valu e shar e weighted) indices . The overall conclusion, based on the official series , is that a dramatic decline i n the per capita marketing of the major export crops was compensated by an equally dramatic rise in per capita food crop production. FIGURE 4

Per Capita Production of Food Crops and Purchases of Export Crops.

0.5 l— | 1

1965 196

1

9 197

1

3 197 Year Food crops -

1

1

H

7 198

1 198

5 198

9

- - Expor t crops

Source: Authors' calculations. Notes: The Tornquist-Theil inde x is a discrete approximation of th e Divisia index. The index was chained and based to 1977 . The index is used in this and all following figures. Divergence with the corresponding Laspeyres index were found to be negligible in all cases. Food crops composing the index: maize, paddy, wheat, sorghum, millet, cassava, and beans. Export crops: coffee, cotton , cashew, pyrethrum, tea, tobacco, and cocoa. Sisal was excluded. Sisa l production has decreased dramatically ove r the period. For export crops, official purchase s are recorded as production.

Performance of Agriculture12 The declin e i n officiall y markete d expor t cro p productio n i n Tanzani a ha s been well-documented, and a broad consensus exists on the basic facts. However , the data refer to official purchase s only; no direct estimates for actual productio n or area figures exist . Consequently, ther e i s no direct o r indirect informatio n o n the siz e of unofficia l market s for expor t crops over time. Thus we have no dat a about the exten t t o which parallel export s have been abl e to compensate fo r th e decline i n the official expor t crop marketings. Per capita export crop purchases showed a gradual, and at times steep, decline between 196 6 an d 1975—th e perio d o f nationalizatio n o f privat e estates . Th e period 1976-198 0 sa w relatively hig h worl d market prices for expor t crops an d was characterized b y stabilizatio n o f per capita export cro p marketings. A stee p decline fro m 198 0 to 198 4 was followed b y a temporary recover y fro m 198 5 to 1987, afte r whic h th e declin e seeme d simpl y t o continu e o n it s earlie r path . I n spite of certain temporary ups and downs, then, over the entire 1965-1989 period, per capita expor t cro p production wa s more than halved . Turning to per capita food cro p production, we can distinguish the followin g subperiods. Food production kept up with relatively high population growth fro m 1966 unti l 1971 , a perio d durin g whic h th e governmen t outline d it s basi c development strategy , highlighte d b y th e Arush a declaratio n o f 1967 . Rapi d growth i n foo d productio n i s als o reporte d i n th e officia l serie s betwee n 197 2 and 1976 , whe n man y o f th e socialis t policie s wer e actually , an d a t time s forcefully, implemente d (e.g. , monopsonistic cooperativ e marketing , villagiza tion, relocatio n o f urba n unemploye d t o rura l areas , compulsor y minimu m acreage, pan-territorial pricing) . The drought o f 1973/7 4 als o fell i n this perio d but seems to have had little effect o n official productio n figures . The period between 197 7 and 1983 is characterized by stagnation in per capita food production . Thes e ar e the years o f "boo m an d bust," the coffe e boo m an d the events that led to the economic crisi s of the early 1980s : the abolition o f th e cooperatives, the war with Idi Amin, the second oil crisis, and the general failur e of stat e socialism t o generate economic growth . The unsustainability o f the official marketin g syste m was accentuated durin g the 1981/8 2 an d 1982/8 3 droughts . Whe n th e governmen t i n effec t liberalize d private grain marketing in 1984 , a dramatic surge in production was recorded fo r 1983/84. After 1984 , however, a decline i n per capita food productio n seem s t o have se t in, although 1988/8 9 was recorded by the EWCMB as an extraordinaril y good year . Summarizing, over the entire 1966-198 9 period , we observe that the officia l statistics report a doubling of national per capita food crop production, primaril y caused by a rapid increas e in the mid-1970s .

3

124 Alexander

H. Sarris and Rogier van den Brink

AN ALTERNATIVE SCENARIO Some observers broadly agree with the trends implied by the official data . They contend that food production in Tanzania has consistently outpace d population growth sinc e th e 1950 s an d ha s i n fac t accelerate d sinc e th e mid-1970 s (Odegaard 1985 , Lundah l an d Ndul u 1987) . I n th e contex t o f a food-versus cash-crop argument , the y posi t tha t a majo r increas e i n foo d productio n ha s occurred, mainl y cause d b y governmen t pric e policie s tha t favore d th e foo d crops, whic h resulte d i n a shif t o f resource s awa y fro m expor t crop s t o foo d crops. Other observers hav e entirel y dismisse d th e assertio n o f positiv e growt h i n per capit a foo d productio n an d posit a negative trend . Thus , Collier , Radwan , and Wangw e (1986 ) mak e th e argumen t tha t foo d productio n pe r capit a mus t have declined over the 1967-197 8 period, partly because of the limited potential for substitution of labor out of export crops. Could we formulate an alternative scenario with respect to agricultural growth in Tanzania? In particular, it seems that acknowledging the existence of positive incentives fo r an increase i n food production need not preclude questioning th e magnitude o f th e increase . Fo r instance, rapi d per capita growt h wa s recorde d during the 1971-1976 period, concurrent with a major drought, villagization, and the abolition of the cooperative structure . It is the 1971-197 6 period , in particular, that accounts for a significant portion of the total increase in per capita food production from 196 6 to 1989 , during which time the index shows a doubling of aggregate pe r capita foo d production . Note, moreover , tha t this acceleratio n i n food crop production in the 1970s is supposed to have come on top of an already impressive growt h rate during the 1960s , when Tanzania had the highest rate of increase o f domesti c foo d productio n o f th e entir e Africa n continen t (se e als o Lofchie 1988) . In contrast to the thesis o f rapi d per capita growt h of foo d cro p production, and to tha t of declinin g pe r capita foo d production , on e coul d entertain a third hypothesis. Fo r the perio d u p to th e mid-1980s , on e coul d posi t a scenari o o f stagnation an d modes t increase s i n pe r capit a foo d productio n base d o n tw o related arguments: first, th e profitability o f off-farm , income-generatin g activi ties declined, increasing the relative profitability o f farm activities and inducing an increase in the availability of farm labor. Additionally, the relative profitabil ity of export crop production declined, causing a shift o f resources from expor t to food crops . So far, the thesis is a basic cash-versus-food-crop argument . However, second , farmers di d not continuously expan d per capita food cro p production significantl y abov e subsistenc e level s du e t o th e increasin g uncer tainty of the economic environment in the period 197 1 to 1984 . That period saw

Performance of Agriculture 12

5

decreasing availability o f incentiv e goods, an inability o f the official marketin g system to assure producers a consistently profitable market, and often-high risks associated with usin g th e paralle l markets . I n other words, th e increas e o f th e relative profitabilit y o f foo d cro p productio n di d no t resul t i n a continue d expansion of food crop production over and above subsistenc e levels . A small but significant increas e in food crop production above former subsistence level s may , nevertheless , hav e occurre d i f farmer s perceive d th e overal l riskiness o f th e marke t environmen t t o have increased . T o hedge agains t suc h risks, they may have increased food production with the objective of increasin g on-farm food stocks. Moreover, the intermittent availability and increasing price of incentive goods may have induced farmers to hold more liquid assets, including food stocks . Producers' reaction s t o marke t failure s ma y hav e include d a continuou s increase in on-farm food stocks. In the absence of large increases in real incomes, such increased storage constitutes the only hypothesis that simultaneously allows for a modest increase in per capita food production and a protracted increase in the real price of food o n the open markets. We wil l furthe r investigat e thes e hypothese s below , confrontin g the m wit h data from various sources. PRICES In discussions of Tanzanian agriculture, one often encounters debates about the importance of "non-price" factors, such as the weather, the effects o f viilagiza tion, labor shortages, lack of processing capacity , transportation problems, and scarcity o f consumer , o r incentive, goods . Car e shoul d b e taken , however , t o define exactly what is meant by "price." For instance, transportation bottlenecks affect th e transformation o f marke t or consumer prices int o producer, o r farm gate, prices. In a situation wher e transportation i s very poor or lacking, i t may be tempting t o say that non-price factor s ar e important, using the term to draw attention t o th e larg e margi n betwee n officia l marke t price s an d far m gate , producer prices . Ye t th e fac t tha t the manipulatio n o f officia l price s ma y no t substantially affec t th e actua l profitabilit y o f rura l farmin g i s hardl y proo f o f the importance o f non-price factors . Evidence fo r low estimate s of elasticities o f suppl y i n sub-Saharan Africa i s usually derived from simulation models, which assume direct resource trade-offs between foo d an d cash cro p production and/o r rigidities i n produce an d facto r markets. Thus, for Tanzania, Renkow, Leonard, and Franklin (1983) simulat e a household model and come up with own-price marketed food supply elasticitie s

126 Alexander

H. Sards and Rogier van den Brink

well below 1 (from .11 to .74). The empirical validity of the trade-off assumptio n is highly dependent , however, on the particular crops under consideration. In Tanzania , th e empirica l definitio n o f pric e i s o f eve n greate r than usua l importance, given the particular marketing environment, which includes substantial paralle l markets . I n this respect , reference s t o "suppl y elasticities " shoul d often b e interprete d a s meanin g th e elasticit y o f quantitie s markete d throug h parastatals, with respect to officially announce d producer prices. The empirical econometric evidence for Tanzania on the magnitude of this type of elasticity o f supply run s contrar y t o th e estimation s derive d fro m th e simulatio n exercise s undertaken b y Renkow, Leonard , an d Franklin (1983). I n general, econometri cally estimate d suppl y elasticitie s ar e positiv e an d quit e high . Ndul u (1980 ) estimates short-ru n suppl y function s fo r th e main maize surplu s regions whic h vary fro m 3. 2 t o 7.2 . Gerrar d an d Ro e (1983 ) giv e 2.2 9 fo r th e own-pric e elasticities o f maiz e an d rice production. Lundahl an d Ndulu (1987 ) giv e pric e elasticities of officially markete d quantities of maize and paddy of 1.6 7 an d 3.1, respectively. However , al l o f th e abov e empirica l evidenc e i s base d o n th e officially recorde d data. The apparently substantial responsiveness of Tanzanian food crop marketing to official pric e may thus exaggerate the responsiveness o f total production . Inclusion o f open market information i n the regression mode l is likely t o reduce the estimates. For instance, using the relative price of maiz e on th e ope n marke t with respec t t o th e officia l pric e o f annua l expor t crops , Odegaard (1985 ) come s t o a considerabl y lowe r pric e elasticit y o f officiall y marketed supply o f 1.04 . Estimates of price elasticities for export crops should be more reliable than those fo r foo d crops , give n th e lesse r significanc e o f paralle l market s an d more reliabl e recordin g o f markete d quantities . Gwye r (1971 ) an d Malim a (1971) arriv e at positive elasticitie s fo r sisal and cotton (e.g., 2.5 fo r cotton). Moreover th e period-by-perio d consistenc y betwee n th e pric e an d quantit y indices o f expor t crops , a s depicted i n Figure 5 , seem s t o make it inappropri ate to downplay th e impact o f th e producer price facto r o n the declin e o f th e export cro p sector . Fro m Figur e 5 i t i s apparen t tha t declin e an d stagnatio n have characterize d th e tren d i n rea l produce r price s o f expor t crop s fo r th e entire period, with the exception of a temporary boom in 1975/1976. A modest and fragil e increas e wa s recorde d fro m 198 4 t o 1987 . Th e modest y o f thi s upward trend is associated with the low transmission of real export crop price increases t o th e produce r level . T o se e this , notic e tha t durin g 198 4 t o 198 7 producer price s o f bot h Arabic a an d Robust a coffe e shoul d hav e rise n con siderably, du e t o a combinatio n o f favorabl e worl d marke t price s an d th e effects o f devaluation. However, as can be gleaned from Table 49, these gain s were no t passe d o n t o producers . I n th e 1986/8 7 an d 1987/8 8 marketin g

Performance of Agriculture 1 2

7

FIGURE 5 Export Crops, Price, and Quantity Index, 1965-1989 . 1.8

1.6 J \ 1.4

•2 1 2 > X

T2 1. 0

0.8 0.6 0.4 T 1 1965 196

1 9 197

r 3 197

7 Year Per capita quantity

I•

1981 198

5

—T

1989

Real price a

a Nominal price index deflated b y the NCPL Source: Authors' calculations.

seasons, the share of world prices received by producers declined substantially compared with the fiveprevious seasons. A similarprocessoccurredfortheothe r major export crops. The benefits of structural adjustment policies have mainly been gobbled up by the parastatal sector. Not surprisingly , then , pe r capit a expor t cro p productio n continue d it s de cline, interrupte d onl y temporaril y b y a positive suppl y respons e i n th e earl y stages of the adjustment period. The early, but temporary, supply response seems to confirm th e argument s of Collie r and Gunning (1989 ) tha t during the initia l years of recovery , increase d availabilit y o f consume r good s provide d th e eco nomic incentive s tha t explai n th e positiv e suppl y reaction . I n th e lon g run , however, real prices need to improve in order to sustain the supply reaction. Figure 6 exhibit s th e aggregat e rea l pric e o f foo d crop s i n th e paralle l an d official market s from 1967 to 1989 . The deflator used is the nonfood componen t of th e National Consume r Pric e Index (NCPl) . When foo d import s increased i n the mid-1970s , th e governmen t activel y use d pric e incentive s t o stimulat e th e production of foo d crops . Startin g in the 1974/7 5 cro p season, a pan-territorial

22.22 11.11 21.10 95

11.14 68

18.96 73

17.38 83

16.38 8.19

26.06 20.85

20.97 16.78

32.70 81

40.38 20.19

28.59 73

39.16 31.33

1983/84

30.14 65

46.67 23.33

36.60 69

52.78 26.39

57.25 79

37.10 67 Robusta

72.36 57.89

1985/86

55.61 44.49

Arabica

1984/85

65.00 51

75.40 58

131.04 65.52

82.50 48 75.94 51

126.90 63.45

171.99 137.59

1987/88

150.00 120.00

1986/87

Source: Unite d Republic of Tanzania, Marketin g Development Bureau , Annual Review of Coffee (Kahawa) 1988, Table s 1 3 and 15 . Notes: Al l price s are in TSh/kilogram unless otherwis e indicated .

Average export price (official exchang e rates) Parchment equivalent price (50%) Producer price (advance + interim + final, clean ) (Tsh/Kg) Producer price/export price (percent)

Average export price (official exchang e rates) Parchment equivalent price (80%) Producer price (advance + interim + final, clean ) (Tsh/kg) Producer price/export price (percent)

1981/82 1982/83

TABLE 49 Tanzania: Pric e Analysis, Arabica and Robusta Coffee .

Performance of Agriculture 12

9

FIGURE 6 Food Prices, 1969-1989 , Officia l an d Parallel Real Prices.

Year Source: Authors' calculations. Note: Nomina l prices deflated b y the nonfood componen t of th e NCPI. Official pric e index is composed of maize, paddy, wheat, sorghum, millet, cassava, and beans. Parallel price index is composed of maize, paddy, and beans only.

price system came into effect, an d compulsory minimum acreage were declared. From 1979/8 0 t o 1983/84 , however , officia l price s fell , i n rea l terms , an d the "dual signal " policy syste m o f officia l price s an d compulsory acreag e sen t ou t conflicting signal s t o producer s (Lundah l an d Ndulu 1987 ; Ban k o f Tanzani a 1984). If we compare the real price level in the parallel market at the end of the 1960 s with that of the early 1980s , we find a considerable appreciation. Parallel market prices ros e steepl y i n th e mid-1970 s i n respons e t o th e inefficiencie s o f th e official marketin g syste m an d remained a t approximately th e sam e hig h level s until the government relaxed enforcement o f the official marketin g channels i n 1984. The observe d pric e trend s o f foo d an d expor t crop s see m onl y partiall y compatible with a cash-versus-food-crop argument. On the one hand, the relative price of foo d o n the parallel marke t with respect to the official pric e of expor t

130 Alexander

H. Sarris and Rogier van den Brink

crops (no t shown ) ros e considerabl y durin g th e perio d 1973/74-1975/7 6 an d between 1981/8 2 an d 1984/85 , an d export cro p productio n fel l steepl y durin g those periods. This part of the food-versus-export-crop thesis seems uncontested. However, an y thesi s tha t posits a complementary, rapi d increase o f pe r capita food productio n i n th e 1970 s need s t o addres s th e consistent , simultaneou s increase o f th e real price o f foo d o n the open markets. The prolonged increas e in real food price s contradict s hig h officia l foo d productio n growt h rates: food production pe r capit a canno t hav e increase d dramatically , becaus e tha t woul d have been reflected in a general decline in real food prices on the parallel market, at unchanged or declining per capita income levels. Only a significant, and highly unlikely, pe r capita real incom e growt h coul d be compatibl e wit h a higher per capita food outpu t with real foo d pric e increase s o n the parallel market , wher e most of the surplus is marketed. The impac t o f foo d marke t liberalization o n food price s afte r 198 4 support s our reasoning. A t that time th e increase d suppl y immediatel y cause d rea l foo d prices to fall. SURVEY DATA

Comparisons betwee n th e Househol d Budge t Survey s i n 196 9 an d 1976/7 7 imply a n annua l growt h o f pe r capit a consumptio n o f th e mai n foo d crop s (excluding bananas ) o f 3. 5 percen t i n th e rura l areas , an d 3. 1 percen t i n th e urban areas . Give n tha t populatio n growt h average d 3. 3 percent , thi s implie s that production of these crops must have grown by well over 6 percent per year over the period 196 9 to 1976/77 . Thi s is a respectable growt h rate. Accordin g to th e inde x base d o n th e serie s o f th e Ministr y o f Agricultur e an d Livestoc k Development, however , pe r capit a foo d productio n wa s suppose d t o hav e increased even more over this period, by over 1 1 percent per year. We have compared the most commonly cited official series—namely , the ones originating from the Early Warning and Crop Monitoring Bureau (EWCMB) of the Ministry o f Agricultur e an d Livestoc k Development—wit h th e result s o f th e recently publishe d Agricultura l Sampl e Surve y 1986/8 7 (AGSASU ) (Tanzani a Bureau of Statistics 1989) . The AGSASU methodology used a nationally representative sampl e an d employe d objectiv e are a an d yiel d measuremen t methods . Note, however, tha t the AGSASU estimates includ e neither production o n largescale farms nor production by "urban farmers," that is, persons cultivating plots in peri-urban and rural areas, but not resident in those areas. If we compare the official EWCM B estimates for some selected crops with the results of the AGSASU study, we se e that for 1986/8 7 the official serie s reported a national production of 2,359,000 metric tons of maize, whereas AGSASU reports

Performance of Agriculture 13

1

2,017,000 metri c tons. The estimates see m reasonably close . However, lookin g at the estimates for paddy, we se e that the Ministry of Agriculture arrives at an estimate o f 644,00 0 tons , and AGSASU reports onl y 445,00 0 tons— a differenc e of 30 percent. Wheat was not included in AGSASU. For the nontraded food crops, the tren d o f cassav a productio n i s th e mos t remarkable : th e officia l figure i s 1,875,000 metri c tons , bu t AGSAS U report s onl y 305,00 0 metri c tons . Bean s production i s officiall y estimate d t o be 347,00 0 tons ; AGSASU reports 282,00 0 tons. Mille t an d sorghu m togethe r accoun t fo r 570,00 0 ton s accordin g t o th e official series , bu t AGSAS U estimates thei r production t o b e close r t o 431,00 0 tons. Acros s th e board , then , th e serie s reporte d b y EWCM B are significantl y above the numbers reported by AGSASU. Another way to compare the estimates is to convert the various crop quantities into calorie s b y usin g standar d calori e content s an d assumin g certai n fixe d conversion rates , such as that for the conversion o f maiz e grai n to maize flour . The results of suc h an exercise ar e depicted in Figure 7. It should be noted that the figures do not include a number of crops important in the calorie budget, such as swee t potatoes , cookin g bananas , nuts , vegetables , fruit , meat , fish, dair y products, oils an d fats, and a number of mino r cereals an d pulses. The graph is solely meant as a crude check on consistency, not as an indication of total calorie production in the country. In Figure 7, total calories available from gross production are plotted over the 1966-1989 perio d an d this tren d is compare d wit h three independen t observa tions: the two consumption estimates from the Household Budget Surveys (HBS) in 196 9 and 1976/77, and the production estimate from the Agricultural Sampl e Survey conducte d i n 1986/87 . Th e comparison s sugges t tha t wherea s i n 1969 , official productio n statistic s seeme d t o impl y pe r capita calorie intake s simila r to the 196 9 HBS, in later years official productio n statistics seem to consistentl y and substantiall y overestimat e th e availabilit y o f stapl e foo d crops . Th e thre e independent observations , take n fro m HB S and AGSAS U respectively, see m t o indicate tha t foo d productio n pe r capit a increase d significantl y betwee n 196 9 and 1976 . Betwee n 197 6 an d 1986 , however , i t showe d a gradual decline , o r stayed approximately constant if one assumes that production by urban residents (around 1 4 percen t o f th e nationa l population ) woul d ad d 1 5 percen t t o th e AGSASU figure .

IMPORTS Another chec k o n consistenc y i s provide d b y dat a on cerea l import s ove r th e period unde r consideration . Th e agroclimati c condition s o f Tanzani a shoul d normally no t warrant any structura l food aid . Emergency foo d ai d should onl y

1 3 2 Alexander

H. Sarris and Rogier van den Brink

FIGURE 7 Caloric Compariso n between Official Serie s an d Survey Estimates , Major Cereals, Cassava, an d Beans. 2,000 I

1

1,800 -j A

1,600 J /

\J

\

1,400 J / 1,200 A /

1,000 J f-J 800 A V^v

600

L,

1966 197

/

,

0 197

,

4 197

,

8 198 Year

,

2 198

J

6 198 9

Sources: Marketing Development Bureau and other Ministry of Agriculture an d Livestock Development publications; Household Budget Surveys: International Labour Office, 1982 ; Bureau of Statistics : Agricultural Sampl e Survey , 1986/87 . Notes: HBS = Household Budget Survey; and AGSASU = Agricultural Sampl e Survey. Crops and calorie content s used: maize 3,630; paddy, 3,540; wheat, 3,440; sorghum, 3,350; millet, 3,360; cassava, 1,530 ; beans, 1,040 .

be require d i n exceptionally ba d years. Nonetheless , considerabl e amount s o f food grains have been imported during a number of years, over and above what would be expected i n light of th e official productio n data. Figure 8 shows tw o periods of considerable grai n imports, namely, 197 4 and 1980-1986 . Not e that Tanzania imported 291,000 tons of maize in 1974/75 . This is nearly 40 percent of tota l productio n th e previou s season , o r 20 percen t o f tha t o f th e 1974/7 5 season. Odegaard (1985,151) attributes the increase in imports wholly to a shift in foo d marketin g fro m officia l marke t channel s t o th e paralle l foo d market s coupled with a decline in wheat production in the large-scale farming sector. In other words , th e increas e i n import s i n th e mid-1970s , accordin g t o on e ob server, was mainly caused by a failure of the official marketin g system to supply Tanzania's urba n consumers with food .

Performance of Agriculture 1 3

3

FIGURE 8 Net Cereal Imports, 1966-199 0 300

-100 h -

1990 Maize Ric

e—

— — Whea t

Source: Internal files, Economic Research Service, United States Department of Agriculture.

The next , mor e prolonge d perio d o f foo d import s too k plac e i n th e perio d 1980/81 unti l 1986/87 . Durin g thi s period , maiz e import s average d aroun d 165,000 metri c ton s pe r year , whic h i s abou t 8- 9 percen t o f averag e annua l production over the same years. Note that around 16 percent of the total population lived i n urban areas during the period, and Dar es Salaam itself constitute d around 5-6 percent of the total population. If one assumes that food imports were mainly destined for urban maize consumption, one is left wit h an unexplainable growth i n national production—an d henc e nonurba n consumption—sinc e tota l maize production , accordin g t o officia l estimates , ros e fro m 1,839,00 0 tons i n 1980 to 2,359,000 tons in 1986, a total of 28.3 percent, or 10.3 percent more than population growth. PARALLEL EXPORTS Some o f th e majo r maize-producin g area s ar e o n Tanzania' s borders . Durin g the perio d o f consume r good s shortage s (mid-1970 s unti l mid-1980s), consid -

134 Alexander

H. Sarris and Rogier van den Brink

erable quantitie s o f maiz e ar e likel y t o hav e bee n trade d fo r incentiv e good s with Kenya , Zambia , Zaire , Malawi , an d Rwand a (Lofchi e 1988 ; 1989) . Th e Marketing Development Bureau (MDB) of the Ministry of Agriculture undertook a stud y o f th e parallel markets for food grain s in 198 3 an d identified norther n Zambia an d southern Kenya a s the main target areas of paralle l grai n exports. Imports wer e consume r goods , suc h a s soap , batteries , an d cloth . Th e MD B estimated tha t 40,00 0 t o 100,00 0 metri c ton s o f maize , rice , an d beans migh t have ende d u p i n paralle l expor t market s (Tanzani a 1983) . Give n tha t th e official productio n estimat e fo r thes e thre e crop s combine d wa s aroun d 2,300,000 metri c ton s i n 1982/83 , th e MD B estimate implie s tha t onl y 1.7 5 t o 4.38 percen t of total production wa s exported illegally . Thus although there is evidenc e tha t parallel exports of food crop s to neighboring countrie s wer e significant , the y canno t b y themselve s explai n th e dra matic rise in officially recorde d food production, given the high transaction costs associated wit h illegal, parallel export markets. For some export crops, substantial parallel markets did exist, however. Coffe e is Tanzania's most important export crop: during the 1980s , the value of coffe e exports constitute d betwee n 2 5 an d 5 0 percen t o f th e valu e o f al l exports . Moreover, an estimated 1.78 million people depend on coffee for their livelihood (Agland Investmen t Service s 1989) . Althoug h producer s hav e reportedl y dou bled Tanzania' s coffe e are a i n th e las t 1 5 years , t o aroun d 234,00 0 hectare s (Kristjanson e t al . 1990) , recorde d yield s hav e decline d dramatically . Whil e reduced yields may be partly the result of a failure of farmers to regenerate their coffee farms , it is likely that some of the yield reduction is in fact imaginary and reflects a n increas e o f paralle l exports , particularl y o f Arabic a coffe e fro m Arusha and Kilimanjaro to Kenya. Supporting this hypothesis is the fact that the quality of Tanzanian coffee ha s fallen rapidly . I n 1968/69 , 8 7 percen t o f th e mil d Arabic a productio n wa s o f medium quality or higher. In 1986/87, the share was only 50 percent. Moreover, high-quality coffe e (clas s 1-5 ) ha s dropped from 1 6 percent to 1.6 2 percen t o f the tota l (Aglan d Investmen t Service s 1989 , 30) . Par t o f th e fal l i n qualit y i s attributed to the lack of a premium for quality i n the official marketin g system . Part, however, may also be attributable to an increase i n parallel export s to the Kenyan, and to a much lesser extent, Ugandan and South African, market s (th e latter would be reached via Zambia). There i s substantia l anecdota l evidenc e o f a lively coffe e smugglin g trad e from Kilimanjar o an d Arusha regions t o Kenya. Interviews wit h coffee trader s in Moshi reveale d that coffee smugglin g wa s virtually nonexisten t i n 197 5 and 1976 but picked u p afterward, du e t o rapidly increasin g price s i n Kenya. Suc h price difference s di d no t remai n great , bu t th e shortag e o f consume r good s i n

Performance of Agriculture 1 3

5

Tanzania i n th e lat e 1970 s an d earl y 1980 s provide d powerfu l incentive s fo r parallel barte r exports. I n 198 3 an d 1984 , parallel export s dropped , due to th e official crackdow n o n paralle l trad e an d th e re-registratio n o f th e origina l cooperative unions . Afte r 1985 , however , paralle l export s agai n steadil y in creased. These anecdote s see m entirel y consisten t wit h th e pric e trend s depicte d i n Figure 9, which show that price incentives did indeed rise steeply from 1975 until 1977. Durin g th e year s o f th e crisis , pric e incentive s wer e less , bu t i t seem s logical to assume that the general consumer goods shortage in Tanzania compensated for this converging trend . The fact that after devaluation of the Tanzanian shilling relativ e produce r price s i n Tanzani a worsened i s als o brough t ou t b y Figure 9.

FIGURE 9

Real Coffe e Prices , 1970-198 8 (NCPI) ; Parallel Market and Official Produce r Price. 400 350 J § 30 0 -\ ii

S> 25 0

150 A 100 A 50 A T

1970

1973 Parallel price

T

1976

1979 Year

1988 Official produce r price

Source: Authors' computation. Note: Paralle l prices were computed by changing Kenyan producer prices to Tanzanian currency at parallel rates. Both official an d parallel nominal prices were deflated b y the NCPI.

136 Alexander

H. Sarris and Rogier van den Brink

Whereas suc h paralle l market s ar e quit e substantia l fo r som e high-valu e export crops, they tend to have a highly regional character. Producers in regions with access to parallel export markets, such as Arusha and Kilimanjaro, certainly took advantag e o f the m to compensate fo r the crisis of th e official econom y i n Tanzania. Thu s compensatin g effect s fro m paralle l expor t income s hav e ha d a considerable regiona l bias. This points agai n to the overriding importanc e o f a realistic assessment of the performance of the food crop sector for the economic analysis of agricultura l producer welfare i n Tanzania. MALNUTRITION A recen t Worl d Bank/FA O mission reporte d a puzzling "calori e overhang " i n Tanzania. Relying on official productio n estimates and increasing the FAO/WHOrecommended foo d requiremen t fro m 60 0 gram s t o 70 0 gram s pe r person pe r day "i n order to allo w fo r post-harvest losses, " the mission conclude d tha t i n 1988/89, nationa l foo d productio n wa s abov e foo d requirement s b y abou t 3 9 percent. Energ y requirement s wer e exceede d b y abou t 2 2 percent , eve n usin g the latest 198 5 FAO/WHO/UNU recommendation of 2,780 kilocalories per person per day (a n increase o f 2 1 percent ove r the old level o f 2,30 0 kilocalories) . I n other words, energy requirement s wer e exceede d b y 2 2 percen t (reporte d sur plus), plu s 1 7 percen t (postharves t losses) , plu s 21 percen t (increas e i n th e FAO/WHO/UNU recommendation)—a gran d total of 60 percent above the recommended leve l o f 2,30 0 kilocalorie s pe r person pe r da y (Yambi , Kavishe , an d Lorri 1990) . Paradoxically, the same mission also reported that 50 percent of the children below the age of 5 years are underweight. Accordin g t o the same mission, "the rates o f malnutritio n i n differen t area s o f th e countr y ar e no t correlate d with agricultural production, " an d i t i s conclude d tha t "adequat e aggregat e foo d availability a t national level doe s not translate into household foo d securit y fo r all" (Yambi, Kavishe, and Lorri 1990 , 1) . Is it possible that another explanation exists for the high prevalence of malnutrition in Tanzania, that is, that aggregate per capita food production is below wha t is reported? Our analysis i n chapter 4 (see Tables 37 and 38) suggested that in the good year of 1976/77 , daily calorie intakes were, on average, below the 2,300 kilocalorie FAO/WHO/UNU recommendation. It seems difficult to believe that after a decade of almost continuous crisis, calorie intak e coul d hav e increase d a s muc h a s th e missio n suggests . O r i s intrahousehold inequality , an d i n particula r th e us e o f grai n t o mak e bee r consumed b y mal e househol d head s i n Tanzania , s o larg e tha t i t ca n b y itsel f explain the phenomenon? That is difficult t o accept on a national scale .

Performance of Agriculture 13

7

AGRICULTURAL STRUCTURE AND TECHNOLOGY

In a relativel y land-abundan t setting , suc h a s ca n b e foun d i n Tanzania , a dramatic increas e i n per capita food production shoul d normally be associate d with an increase i n farm area, holding technolog y constant . To investigate thi s hypothesis, one can compare changes in farm size and size distribution between 1971/72 an d 1986/87 , a s reveale d b y th e Burea u o f Statistic s Agricultura l Census and the Agricultural Sample Survey (AGSASU) respectively (se e Figures 10 and 11). Between 1971/7 2 an d 1986/87 , ther e appea r t o b e fe w significan t change s with respect t o far m siz e distribution , apar t from a n increase i n the number o f farms smalle r tha n 0.5 0 hectare s whic h seem s t o hav e com e mainl y fro m th e 0.50-2.00 hectare s category . Similarly , a look a t th e change s i n are a actuall y cultivated reveals that there has not been a net increase. In fact, except for farms larger than 1 0 hectares, cro p land per holding actuall y seem s t o have decline d (Figure 12) . Overal l increase s i n far m area , o r distributiona l shift s capturin g economies of scale seem, therefore, unlikely candidate s to explain high agricultural growth rates in Tanzania. What is perhaps most strikin g i s that according to thes e data , the highly interventionis t agricultura l policie s durin g th e perio d under consideration seem to have had little actual impact in terms of the size and distribution of farm area per household. On-farm labor resources have increased somewhat during the period 1971/721986/87 (see Figure 13). This would seem to be in line with the slowing of urban population growth from 10. 7 percent during the years 1967-197 8 t o 5.4 percent for 1978-1988 . Th e increas e i n on-far m labo r resource s seem s t o hav e bee n absorbed mainly by the farms larger than 2 hectares. Labor resources per hectare have also increased (see Figure 14), but only marginally, and again mostly in the larger size classes. In order to ascertain whether a shift of land resources from cash to food crops has taken place, one can first examine changes in the area under food crops. The comparison o f the two survey s in terms of ne t cultivated area under food crop s is technicall y difficult , give n difference s i n definition s betwee n th e tw o cen suses. However, sinc e the two different estimate s that we constructed produce d approximately th e sam e overal l pattern , w e onl y repor t the result s o f on e (se e Figure 15) . I f w e compar e th e are a recorde d fo r th e mai n season , tha t is , th e masika season , in 1986/87 with the 1971/7 2 data, there is evidence that the area devoted to food crops actually decreased over the period. The decrease becomes less prominent , o f course , if fo r 1986/8 7 th e combine d are a of masika an d vuli seasons i s used , bu t th e decreasin g tren d stil l remains . However , give n th e different way s in which the results were reported in the two surveys, one should

1 3 8 Alexander

H. Sarris and Rogier van den Brink

FIGURE 10 Changes in Farm Size Distribution , 1971/72-1986/87 . 40 35 U

• 1971/7 • 1986/8

23 0k

I 25 I i* 2 0 kh

2 7

(D

21 5 U

I

10

PL)

0

?

4

J

Size class (hectare)

0 0 oo • '-< °'-

Sources: Compute d from data in the Tanzania Bureau of Statistics, Agricultural Censu s 1971/7 2 and Agricultural Sampl e Survey of Tanzania Mainland 1986/87 .

FIGURE 11 Changes in Farm Size, 1971/72-1986/87 . 15

Size clas s (hectare ) Sources: Compute d from data in the Tanzania Bureau of Statistics, Agricultural Censu s 1971/7 2 and Agricultural Sampl e Surve y of Tanzania Mainland 1986/87 .

Performance of Agriculture 1 3

9

FIGURE 12

Crop Land per Holding, 1971/72-1986/8 7 8 7 hL •3 6

M

• 1971/7 2 • 1986/8 7

k