Tanzania - Basic economic report (Vol. 5 of 8) : Income distribution and growth : A simulation model

Report No. 1616-TA FIE COP Tanzania Basic Economic Report Annex IV-Incomne Distibuton and Growth: A Simulaton Model December 1977 Eastern Afnca Country Programs I FOR OFFICIAL USE ONLY Document of the Worfd fank This document has a restncted distnburon and mav be used by recipients only in the perfomTance of their offcia dudes-. ltscontents may iot otherwise be discfosed without Wodd Bank auton,w-nn.  CLRICY EQUIVALENTS 1/ SDR 1.00 = Tanzania Shilinig (T.Sh.) 9.66 US$ 1.00 = Tanzania Shilling (T.Sh.) 8.30 T.Sh. 1.00 US$0.12 TANZANIA FISCAL YEAR July 1st- June 30th 1/ In October 19'75 the Tlanzaniau Shilling was pegged to the value of the Special. Drawing Rights (SDR) of the International Monetary Fund. The US Dollar/Tanzania Shilling exchange-rate is therefore subject.to change. .~~~~~~~~~~~~~~~~~~~~~~~~~~  FOR OFFICIAL USE ONLY TANZANIA - BASIC ECONOMIfC REPORT ANNNEX IV INCOME DISTRIBUTION AND GROWTH: A SIMJLATION )IBDEL Prepared by Mr. J. W. Gunning Assisted by Mr. R. G. Pad ula The full report consists of the following separately bound volumes: Main Report Annex I - Domestic Finance and Resource Use Annex II - Fiscal Implications of Universal Primary Education and Universal Rural Water Supply Annex III - Labor Market Allocation and Income Distribution Annex IV - Income Distribution and Growth: A Simulation Ymdel Annex V - Industry: Perspective and Strategic Choices Annex VI - Key Issues in Agriculture and Rural Development Annex VII - Appropriate Technology in Tanzanian Agriculture: Some Fnpirical and Policy Considerations 1This docume11t has a restricted distribution and may be used by recipients only in the performalnc: of their officiai cdutues. Its contents rnay not otherwise be disciosed without Worid Bank authonzation.  - ii - TABLE OF CONTENTS Page No. INTRODUCTION ........... ........................ v-vii PART A: DESCRIPTION OF THE MODEL I. PRODUCTION. SAVING A.ND INVESTMET . ............... . . .... 1 1. Introduction.. . 1 2. Definition of Sectors. .......................... 2 3. Production Block .. 3 4. Savings and Investments .. 4 5. The Input/Output Table . II. THE URBAN INCOME DISTRIBUTION. 9 1. Introduction.... 9 2. Regular Wage Earners.... 10 3. Self-Employed ... . . 12 4. Casuals ....12 5. Unemployment and the Free Entry Sector Equilibrium . 13 6. The Derivation of the Household Income Distribution. 14 III. RURAL INCOMES AND MIGRATION . ....19 1. Introduction . . .19 2. Rural Cash Income ...... ..19 3. Rural Subsistence Income . . .22 4. Migration ...... ..... 22 IV. CONSUMPTION .25 1. Introduction ..25 2. The Consumption Function . .25 3. Aggregation ...27 V. PRICES .29 1. Introduction ......... . . .29 2. Gross Output Prices ....29 3. Consumer Price Indices. . ..... .... 30 4. Consumer Prices after 1975 ....32 5. GDP Deflators ....32  Paae No. VI. INDUSTRY ................................................ 33 1. Introduction ....................................... 33 2. Subsectors ...................................... 33 3. Updating the 10 Table .............................. 36 VII. PUBLIC FINANC . .......................................... 38 1. Introduction ........................... ..38 2. The Recurrent Budget . . .......................... 38 3. The Composition of Domestic Savings .............. .. 39 PART B: SIMULATION RESULTS VTII. SIMULATION RESULTS: An Overview ......................... 40 IX. TEE BASE RUN ............................................ 43 L. Introduction ....................................... 43 2. Production, Savings and Investment ............ ..... 43 3. The Urban Income Distribution ...................... 46 4. Rural Income and Migration ..... ...................... 50 5. Consumption ........................................ 53 6. Prices ... ;3 7. Industry ........................................... 55 8. Public Finance ..................................... 58 X. SOME SENSITIVITY RUNS ................................ .... 59 1. Introduction . . . ........................ . i 9 2. Lower Migration (Run 2) . .......................... 59 3. Alternative Industrial Strategy (Run 3) . . .......... 61 4. Lower Wage Increases (Run 4) ..... .................. 62 5. Lower Public Consumption (Run 5) . .................. 65 6. Other Results . . .............. 67 V. AN ALTERNATIVE RUN ...................................... 69 1. Introduction ............... * . ...................... 69 2. Production, Savings and Investment ............. .... 70 3. The Urban Income Distribution ..... ................. 71 4. Rural Income and Migration ................. ........ 74 5. Other Results ....... ............................... 76  - iv - PART C: APPENDICES A. List of symbols B. The 1969 lnDut-Outzut table: the matrices D, B and MIG C. Wage Employment D. Results of the 1975 HBS Pilot Survey E. The derivation of the urban income distribution 1.. The matrix FW 2. The distribution FLSE 3. The matrices NEK and G 4. The urban income distribution in 1969 F. Data on consumer prices C. The matrices WtULC, T4RLC, WRSC and rtRSS 1. The EBS tape 2. Rural and urban consumption baskets H. Aggregation of consumption: the matrix CAM LIST OF TABLES REFERENCES  PART A: DESCRIPTION OF THE MODEL  INTRODUCTION i. This oaper presents a model for Tanzania and describes some of its simulation results. ii. Developing an economy-wide model which fits reality enough to be of some practical use is a hazardous undertaking for any country. In the case of Tanzania, a country for which no macro-economic model exists I/ and the data base of which is generally (not quite fairly) considered to be very poor, this is especially true. This model has not been validated econometri- cally and much uncertainty surrounds some of the crucial specifications and coefficients. Nevertheless, we consider the model useful, if not for pre- diction, certainly for the 'kind of policy simulation experiments which are described in Part B. iii. As far as the data base is concerned, it is ironic to find that it is considerably weaker for the "traditional" Darts of the model (ICOR's, sectoral production in constant prices, trade indices, sectoral allocation of investment, etc.) than for the model's less conventional parts concerning income distribution. Combining data from the 1969 Household Budget Survey, the 1971 Urban Survey, the 1975 Manzese Survey, the .1976 Pilot Sousehold Budget Survey and various issues of the Survey of Employment and Earnings leads to a data base in which we have considerable confidence. This Annex relies heavily on the work reported in Annex III in this area. iv. Even an economy-wide model cannot address all issues. Many issues are not considered at all in this model because the required data do not exist, because they are not very important in the Tanzanian context or simply because we have to draw the line somewhere. One will e.g. find nothing on the supply response to changes in agricultural producer prices, choice of technique as a function of factor prices, the factoral distribution of in- come, exchange rate policy, substitution effects in private consumption demand or the welfare effect of public expenditures. This still leaves a very complicated model in which we hope questions of growth and income dis- tribution can be explored usefully. v. The complexity of the model is as much a concern to the author as it probably is to the reader. Even if one accepts the chosen specifications as correct, the question remains whether the same results could not be ob- tained after some simplification: deleting some equations, introducing less disaggregation, etc. Such an attempt to "collapse" the model has not yet been made. We intend to do so, however, but, as yet, do not expect too much from it: many of the model's more interesting (and often to some extent counter-intuitive) results can be traced directly to e.g. the very detailed treatment of income distribution. I/ Faaland and Dahl (1967)'s attempt cannot qualify as such: in this "model" all variables are a function of GDP, which itself is exogenous.  - vi - vi. The paper is divided into three parts. The model is described in Part A (chapters I-VII, pp. 1-39). There is a short chapter on each of the model's main groups of equations. The non-technical reader can get an overall idea of the model's structure by reading the first section of each chapter. vii. Simulation results are discussed in Part 3 (pp. 40-76). Two runs (the "baserun" in chapter IX and the "alternative run" in chapter XI) are treated at some length. Four other runs are discussed in much less detail in chapter X. viii. The appendices form Part C. Except for the first one (which gives the list of symbols) these are devoted to data sources and manipulations. With the exception of appendices D and E (on income distribution data) these are probably mainly of interest to the specialist. ix. The model considers eleven sectors. Output is endogenous (except for agriculture and mining), being determined by past investment (manufactur- ing) or by final and intermediate demand (demand-oriented sectors). The supply-demand balance holds for each sector. The level of investment is endogenously determined as the sum of domestic and foreign savings. The requirement to satisfy the material balances and to have zero net-imvorts for sectors producing non-tradeables makes the allocation of investment between sectors endogenous. One of the instruments in the model is the choice of an industrial strategy: within the manufacturing sector the allo- cation and timing of investment for 30 subsectors is exogenous. Total im- ports have to satisfy foreign exchange availability. The implied rate of import substitution provides a feasibility test of the model's results. x.' Agricultural production is exogenous and this is an important limitation of the model, but the data required to make it endogenous in a meaningful way do not exist. This means that little can be said about the intra-rural income distribution. The model does pay attention however to interactions between the rural and urban sector, including rural-urban migration and terms of trade. In the urban sector the income distribution is treated in more detail. Sectoral output levels determine wage employ- ment; wage rates are exogenous (and form an important policy instrument). The size of, and income levels in, the informal sector (self-employed and casuals) is determined by supply and demand and depends on the growth rate of formal employment, wage rates, urban income levels and the size of the labor force (as affected by migration). Hence, the relations between the formal and the informal sector are included and the choice between (open) unemployment and informal employment is coasidered in a general equilibrium framework. The distribution between the formal sector (on which distribu- tional policies tend to focus) and the informal sector (over which the government has virtually no control) enables us to consider possibly unin- tended side effects of incomes policies. Rural-urban migration is endo- genous and depends on rural and urban income levels and the probability of findiag urban formal employment. The income distribution over house- holds is derived from the distributions over earners, allowing a differen- tiation between the direct effect of policies on the latter and their ulti- mate impact on the former.  - viid - xi. For both rural and urban households income-elastic consumption functions are estimated for each consumption good: household saving and the level and composition of private consumption are endogenous. xii. Prices are determined on a cost-plus basis as function of wage rates, labor productivity, import prices, subsidies and indirect taxes. Relative price changes have no substitution effect but do affect consump- tion indirectly through changes in real incomes. xiii. The model has been designed to help answer questions such as: (a) the implications for the growth rate,- the atructure of output and the required import-substitution of a projected decline in foreign savings; (b) the effect on output, imports and employment of different industrial strategies; (c) the effect on prices, income distribution and public finance of changes in direct and indirect taxes; (d) the effect of wage and producer price policies on income distribution (taking induced changes in migration, informal sector developments and the unemployment rate into account) and, through personal incomes, on the levels of consumption and savings; (e) the effect of public consumption on employment, public finance, prices, the sectoral composition of output, income distribution and the growth rate; (f) the required investment allocation given the need to avoid supply-demand imbalances.  1. PRODUCTION, SAVINGS AND INVESTMENT 1. Introduction 1. The model distinguishes eleven sectors: agriculture, mining, manu- facturing, small-scale industry, public utilities, construction, trade, trans- portation, real estate, public administration and services. Output levels are determined as follows: (a) for agriculture and mining, growth is exo- genous; (b) for manufacturing, output is determined by total investment in the sector as a whole and the allocation thereof between thirty subsectcrs (this disaggregated approach is further discussed in chapter VI); (c) all other sectors are considered demand oriented, i.e. output is determined by the level of final and intermediate demand. Since the output of the demand- oriented sectors consists largely of non-tradeables, an excess of domestic demand over production cannot be met from imports. Hence, if these sectors are not to become a bottleneck in the development of the rest of the economy, their output must be equal to domestic demand. In order to make the necessary adjustments feasible, the demand-oriented sectors have a first claim on in- vestible resources. 2. The 1969 Input/Output Table is used to derive intermediate deamnd and the relation between value added and gross output. Technical change is assumed to occur mainly in the manufacturing sector. Hence, technical coef- ficients are constant for all sectors except manufacturing and for this one they change continually and depend on the industrial strategy followed. The I/O Table is corrected for all sectors to incorporate the changes in indirect taxation which occurred between 1969 and 1973 (the model's base year). 3. Material balances are satisfied for each sector. This equation is used for the demand-oriented sectors to derive output levels and for other sectors to derive imports residually. This excess-demanded specifi- cation of imports implies that the issue of the reconciliation of the savings and trade gaps does not arise in the model: foreign savings are exogenous, domestic savings are endogenous and since ex post and ex ante consumption are identical in the model, any adjustment needed for the equal- ization of the two gaps is implicitly made ou the imports side. 4. Aggregate growth is determined by both endogenous and exogenous variables. First, the savings rate depends on: (a) the persoual income distribution, (b) taxation, (c) consumer prices, (d) government consumption and (e) foreign savings. Secondly, the aggregate ICOR depends on: (a) the exogenous growth rates for mining and agriculture, (b) the level and compo- sition of final demand, in particular of private consumptiou, (c) the indus- trial strategy and (d) foreign exchange availability. Of the exogenous factors, foreign exchange availability is the most important one; with for- eign savings and export levels specified exogenously, the level of imports is predetermined and this, as it turns out, presents a severe constraint on growth.  -2- 2. Definitions of Sectors i. The definitions of sectors used in the model differ somewhat from those used in the Tanzanian National Accounts. Table I.1: DE7INITb0N OF SECTORS IN THE MODEL AND CORRESPONDING SECTOR NUMBERS IN TEE NATIONAL ACCOUNTS (NA) AND THE IiNP!T OUTPUT TABLE (IO) NA IO 1. Agriculture, hunting, forestry and fishing L 1-13 2. Mining and quarrying 2 14,15 3. Manufacturing 3 16-28 4. Small scale industry 3 29-31 5. Public utilities (electricity and water supply) 4 32 6. Construction 5 33 7. Trade 6 34,35 8. Transport, storage and coamunication 7 36,37 9. Real estate 8 39 10. Public administration 9 41 11. Services, namely: (a) finance, insurance and business services 8 38,40,42-45 (b) imputed bank service charges 10 (c) other services 9 The reasons for the differences with the National Accounts definitions are simple: (a) a small scale industry sector is distinguished because, sup- posedly, this is where the self-employed (other than those in services) are concentrated. Treating this sector separately, therefore, enables one to construct a better proxy for the demand facing the self-employed; (b) public administration is treated separately because the growth of this sector can be controlled directly; (c) the output of the real estate sector consists largely of the value of owner-occupied housing. Since estimates of this value (especially in rural areas) are highly unreliable and since  -3- increases in the final demand for this sector are largely unrelated to changes in personal incomes inclusion of real estate in an aggregate services sector would be misleading. 3. Production Block 6. The basic equations of the model's production block are: (1) XNi = 2>i (i=1,2) (2) 2i - ZjAij=j + CPi + CGi + IS1 +Ei M (i4, ,11) (3) VACi =i - ZjAji - 3Zi - B3i - 34i + B5i + B6i (i=l,..11) (4) YXi - VACt 2Ni i1. ,) (5) GfP'C - ZiYNi (6) GDPMP - Zi (1 -ZjAji) XNi - GDPFC + i(B2i + B3i)xMi (7) Hi-Mi (i=4,.. ,11) (8) = =ZjAij2Nj + CPi + CGI + ISi + - xNi (i-1,2,3) where KS stands for gross output, YN for value added at factor cost, CP for private consumption, CG for government consumption, IS for investment (by source), E for exports, M for importsi VAC for value added coefficient and GDPFC and GDPMP for GCP at factor cost and market prices. 1/ All values are in 1969 prices. 7. The matrix A gives the requirements for intermediate goods, both domestically produced and imported. The rows of B give: 2/ 1/ Exogenous variables are denoted by a bar over the variable name. 21 The first row gives intermediate goods imports and is relevant only when used in conjunction with D where A - D + MIG and MIGi; are the imported intermediate goods of sector i used per unit of output of sector J. Since Z1MIG13 - B3j both ZAij + Zi#, 3j 1 and +iD: - 1.j  -4- Row Contents 2 customs duty 3 other indirect taxes (net of subsidies) 4 wage and salaries 5 depreciation 6 operating surplus 8. For agriculture and mining, output is exogenous; for manufacturing, output is determined in another part of the model and for all other sectors, output is determined by equation (2) as the sum of 'inal and intermediate demand. Non-zero exogenous imports are specified only for sector 11 (imports of non-factcr services); for all other dpemnd-oriented sectors, the model generates zero import levels. Fior the first three sectors, imports are defined by equation (8) as the excess of demand over supply. This specifi- cation implies that the model does not recognize non-competitive imports. The model's results must, therefore, be checked for the feasibility of the implied import substitution (as measured by the change since 1969 in the ratio of domestic production and demand). 4. Savings and Investments 9. Government consumption (CGT = ZiCGi) is exogenous but private con- sumption (CPT = riCPi) is endogenous (and strongly dependent upon the model's specizication of income distribution). Given these tvo levels domestic saving (GDS) follows from the identity: (9) GDS - GDPME - CPT - CGT 10. Foreign savings consist of net capital inflows, net factor service income, net transfers and gains from terms of trade: (10) FS = (CAP + FSY + NETTR)/IPTM + (Zi IPIEiEi/IP2f -Zi Ei) where I?fl and I?IEj are import and export price indices (1969=1) and CAP, FSY and NETTR are net capital inflow, factor service income and transfers in current prices. 11. Total savings and investment are now determined: (11) TS = GDS + FS (12) IT - TS 12. Investment requirements are derived from exogenous ICOR's. The amount available for investment in manufacturing then follows residually:  _ 5_ (13) IDi t-1 = ICORi (YNi - YNi,t-1) (i=1, 2, 4,..,1!) (14) ID3 IT -Z if3 IDi (15) ISi = ISIDi IT, Z iISIDi = 1i (i=1,..,L) where ID is investment by destination. 13. Three points should be noted about these equations. First, there is only a one year gestation lag in equation (13). Although this assumption can be relaxed for sectors one and two, for the demand-oriented sectors, it is the price one has to pay if one wants to avoid intertemporal solution problems. 1/ The implicit assumption that output levels can be adjusted very quickly in these sectors is most obviously unrealistic for the public utilities and the transportation sectors. The former case is taken care of by making investment in electricity generation exogenous (the investment pro- gram is in fact already pre-determined) but maintaining equation (2). In that version of the model, ICORs become variable, reflecting a difference between output and capacity; capacity is increased in discrete steps and output remains continuously below it.. 14. Secondly, equation (14) is certainly extreme in that any reduction in savings will affect only the manufacturing sector. It means that there are in fact two investment decisions in the model: one is the allocation of investment within the manufacturing sector. The other policy instrument is the allocation of investment between agriculture, mining and manufacturing. This latter decision enters in the model through the exogenous growth rates of sectors 1 and 2: if one wants to increase investment in manufacturing, this can only be achieved by a reduction in the growth rates for agriculture and mining (with feedback effects on the demand-oriented sectors). 11 Equations (13), (14), (15) are solved as follows: given a trial value of ID3 t-l the model can solve for all output levels and then use equa- tions (13) to solve for non-manufacturing investment. Since ITt.i is, of course, predetermined, equation (14) will ncw give a new value for ID3,t-I and one has to iterate between years t-1 and t until convergence is reached, but one does not have to solve the whole model again for year t-1: only equation (14) matters. This is possible because (a) there is only a one period lag in (13) (b) equation (15) is a special case of ISi - =ZISID..ID. namely the case when in any row of ISID all elements are identical. if either condition is not met one would have to solve the whole model again for at least one preceding year.  -6- 15. Thirdly, it should be noted that (15) is a degenerated version of a full (Ilx1l) matrix mapping investment by destination into investment by source. Such a matrix is not available for Tanzania and the simplistic assumption is therefore made that the composition of total investment (by destination) is irrelevant for its disaggregation to investment by sector of origin. 16. Obviously, the preceding equations guarantee the ecuality of the savings and trade gaps: riXNi - XijAiXN- C2T + CGT + IT + ET - MT, or GDPMP - CPT - CGT = IT + ET - MT, or MT - ET - IT - GDS = FS 17. It should be noted that the three components of domestic savings: corporate savings (CS), government savings (GS) and household savings (HS) cannot be specified independently. Personal income determines both CPT and HS, government revenues and current expenditure (CGT, which is exogenous) determine GS; GDS then follows from (9) and corporate savings can be derived only residually (cf. section VII.3). 5. The Input/OutDut Table 18. The input/output table plays a central role in the production block. A 45-sector I/O table is available for Tanzania both for 1969 and 1970. We have opted for the earlier one since 1969 is also the year of the Household Budget Survey and, hence, it'is desirable to work in 1969 prices. 1/ The I/O Table gives three matrices: (a) D (for use of domestically produced inter- med'iate inputs), (b) B (for imports of intermediate goods, indirect taxes and value added), and (c) MIG (for imports of intermediate goods by sector of origin and sector of destination. 2/ Except for the column corresponding to the manufacturing sector (which is changed every year during the simula- tion period), these matrices are left constant after having been corrected for changes in the structure and level of indirect taxation between 1969 and 1973. For this correction, four categories of indirect taxes are con- sidered separately: (a) export duties, (b) subsidies, (c) import duties and (d) other indirect taxes. For the first two categories we have data on 1973 revenues by sector, for the other two we only have total revenues. One crude assumption has to be introduced in the absence of price deflators for gross output by sector and that is that the price deflators (1969=1) for in- direct taxes and gross output are equal to the one for value added (at factor cost). It is further assumed that the incidence of those taxes for which we 1/ The National Accounts are in 1966 prices and had, therefore, to be con- verted. 2/ These three matrices appear in Appendix B.  - 7 - have no sectoral breakdown (i.e. (c) and (d) above) has not changed in the sense that the ratio of these taxes to gross output (in constant prices) changes for each sector by the same percentage. For each sector we then derive an estimate BT2 of customs duty payments and BT3 of other indirect taxes (net of subsidies) both in current prices. The new coefficients B2i and 33i are then calculated as the ratio between tax payments and gross out- put (both i n current prices) and the rest of column i in the matrices D and B is adjusted so that it still sums to one, preserving the original ratios: (16) B2i = BT2i/(Ni PXi) (i=1,...,ll) (17) B3i - BT3i/(XNi PXi) (i=1,.., ,1) (18) COR.Ri = (1-32i - 33i)/(I-Bi - B3i69) (i=1, **l (19) Dii = C0RR- Dji (i=, ,ll; j=l,.,ll) (20) Bk.