Directory UMM :Journals:Journal Of Policy Modeling:Vol22.Issue6.2000:

(1)

A General Equilibrium Analysis of the Effect of

Macroeconomic Adjustment on Poverty in Africa

Paul A. Dorosh, International Food Policy Research Institute

David E. Sahn, Cornell Food and Nutrition Policy Program

Using CGE models for four countries (Cameroon, The Gambia, Madagascar, and Niger), this paper examines the consequences of macropolicy reform on real incomes of poor households in sub-Saharan Africa. The simulations suggest that, compared to alterna-tive policy options, trade and exchange rate liberalization tends to benefit poor households in both rural and urban areas—as rents on foreign exchange are eliminated, demand for labor increases, and returns to tradable agriculture rise. The small magnitudes of the gains in average real incomes of poor household groups modeled suggest that macropolicy reform alone will not be sufficient in the short run to significantly reduce poverty in Africa. 2000 Society for Policy Modeling. Published by Elsevier Science Inc.

1. INTRODUCTION

Has economic policy reform in sub-Saharan Africa worsened income distribution and exacerbated poverty? These questions remain the subject of considerable debate. Critics of the reform process argue that orthodox adjustment policies, including ex-change rate devaluation and reduced government spending, have resulted in a disproportionately large burden for the poor (Cornia, Jolly, Stewart, 1987; Taylor, 1988; UNECA, 1989). Other analysts

Address correspondence to Paul A. Dorosh, IFPRI, 2033 K St., NW, Washington, DC 20006–1002, USA.

Paul A. Dorosh is currently a Research Fellow with the International Food Policy Research Institute. This work was conducted while he was an Associate Professor at Cornell University. David E. Sahn is Professor of Economics at Cornell University.

This research was supported by a grant from the United States Agency for International Development. We wish to thank Irma Adelman and Erik Thorbecke for helpful comments.

Received March 1997; final draft accepted December 1997.

Journal of Policy Modeling22(6):753–776 (2000)

(2)

of the adjustment process, while acknowledging that reforms have often failed to generate real economic growth,1 _{suggest that, in}

general, the poor (and especially the rural poor) marginally bene-fited from many adjustment measures (World Bank, 1994; Sahn, Dorosh, and Younger, 1996).

In assessing the effects of reform on the poor, as well as other groups of households, it is crucial to separate the results of adjust-ment policy from the impacts of terms-of-trade shocks and other factors linked to the economic crises that necessitated adjustment measures. Therefore, in this paper we use computable general equilibrium models to simulate the effects of policy reform on real incomes of various household groups from four African countries: Cameroon, The Gambia, Madagascar, and Niger.2 _{We focus on}

two major broad types of adjustment policies—trade and exchange rate liberalization, and reductions in government spending—and present counterfactual simulations that elucidate important path-ways by which policy reforms affect real incomes of poor house-holds in the African context.

As a prelude to the model simulations, we briefly discuss key aspects of the economic structures and adjustment processes in the four countries in Section 2. A brief description of the models follows in Section 3. Section 4 then presents the results of coun-terfactual simulations of alternative adjustment policies. Conclud-ing remarks are in Section 5.

2. COUNTRY CONTEXT

The four countries modeled in this paper, while sharing some broad economic characteristics common to most countries in sub-Saharan Africa (such as a high share of labor employed in agricul-ture, dependence on primary commodity exports for foreign ex-change, and relatively small industrial sectors), differ considerably

1_{Poor macroeconomic performance in the wake of economic reform programs has been}

attributed to a number of factors including lack of effective implementation of policy reforms, adverse terms-of-trade shocks, and other structural factors impeding growth.

2_{CGE models have been used to examine the sectoral and macroeconomic impacts of}

(3)

MACROECONOMIC ADJUSTMENT AND POVERTY 755

in terms of the structure of production and household expendi-tures, the size of the external shocks they faced in the 1980s, and the policy options available.3

As members of the CFA franc zone, both Cameroon and Niger maintained fixed nominal exchange rates relative to the French franc until January 1994. Adverse terms-of-trade shocks (and for Niger, adverse weather shocks) were major factors contributing to the economic crises in both countries, as well. The differences between the countries in levels of income, natural resource endow-ments, and government policies are greater than the similarities, however. Cameroon’s petroleum exports and overall good natural resource endowment make it one of the wealthier countries in sub-Saharan Africa, with a GDP per capita of nearly U.S. $1,000 in 1990. Unlike most of the other countries in the region, the early 1980s were good years in Cameroon. However, when the price of oil fell in the second half of the decade (resulting in a 50% decline in Cameroon’s terms of trade between 1984–85 and 1988–89; Table 1), the Cameroonian government had little re-course but to adopt a stabilization and adjustment program. In contrast, Niger’s landlocked location in the Sahel contributes to the marked instability in rainfall, degradation of the fragile soils, and extreme dependence on uranium exports for foreign exchange and government revenues. These factors, in combination with a policy of deficit spending, contributed to an economic crisis in the early 1980s as uranium prices fell and rainfall declined.

In common with Niger, the economic crises in The Gambia and Madagascar, each with a GDP of around one-quarter that of Cameroon, hit early. In both cases, the crisis was largely attribut-able to the fiscal deficits that resulted from a heavy investment push at the end of the 1970s. A decline in the world price of groundnuts, The Gambia’s major commodity export, coupled with drought in the first half of the 1980s, reduced export earnings and domestic food production. Similarly, a decline in the world price of coffee contributed to a 25 percent decline in Madagascar’s terms of trade in the early 1980s, and reduced Madagascar’s foreign exchange earnings just as world credit markets tightened and commercial loans on foreign-financed investment projects came

3_{A more detailed discussion of the economic crises and subsequent reforms undertaken}

(4)

Table 1: Budget Surplus/Deficit, Terms of Trade, and Real Exchange Rate, 1975–1979 to 1990

Budget surplus/deficit Terms of trade Real exchange rate (% GDP) (19875100) (19875100)

Cameroon

1975–1979 20.1a _170.9 _—

1980–1983 20.9 146.3 76.3

1984–1985 1.3 142.5 79.0

1986–1987 21.4 99.2 94.7

1988–1989 21.1 71.0 92.8

1990–1991 — 89.7 92.5

The Gambia

1975–1979 27.1 145.8 —

1980–1983 27.4 116.8 129.9

1984–1985 22.9 120.8 126.3

1986–1987 24.1 102.8 97.4

1988–1989 23.9 121.0 105.8

1990–1991 20.4 89.6 97.2

Madagascar

1975–1979 25.6 117.8 —

1980–1983 29.7 88.9 181.8

1984–1985 23.6 96.3 159.7

1986–1987 23.4 115.9 102.0

1988–1989 23.8 105.7 84.7

1990–1991 23.1 84.5 87.5

Niger

1975–1979 24.8 150.2 —

1980–1983 212.0 119.7 138.5

1984–1985 28.3 125.1 120.3

1986–1987 29.5 108.5 105.1

1988–1989 210.1 83.1 90.9

1990–1991 210.4 77.8 84.8

Sources: World Bank (1992); World Bank (1995); IMF (various years).

a_{1977–1979 only.}

due. The financial crises that ensued in The Gambia and Madagas-car contributed to the need for stabilization efforts and subse-quently set the stage for adjustment efforts to deal with the under-lying impediments to growth, including reform of exchange rate and pricing policy.

(5)

Instead, we use the four models to show how external conditions and the policy framework result in different growth and distribu-tional outcomes in the four countries with diverse economic struc-tures and different abilities to absorb and respond to changing external conditions. That is, we choose to simulate the same shocks and same policy responses to enable cross-country comparisons. Use of four different models also provides an implicit sensitivity analysis to changes in model structure and parameters.

3. THE MODELS

The four CGEs, following Dervis, de Melo, and Robinson (1982), share many of the same characteristics.4 _{Production is}

disaggregated into major agricultural, industrial, and tertiary sub-sectors in a manner reflecting the importance of various subsub-sectors in each country (Appendix Table 1A), with value added modeled as a CES (constant elasticity of substitution) function of labor of various skill types and sector-specific fixed capital. Capital includes formal and informal nonagricultural capital, and land disaggre-gated by ecological region (for Cameroon, Madagascar, and Niger) and the income level (poor/nonpoor) of farm households. Incomes of households are determined according to each household’s fixed share of returns to factors of production. We model incomes net of household transfers, except in The Gambia, where they are fixed in local currency terms.5 _{Consumption of each commodity}

is alternatively modeled as a fixed value share of total expenditures (The Gambia and Niger) or using a linear expenditure system (LES) specification (Cameroon and Madagascar).

4_{The structures of the CGE models are presented in Condon, Dahl, and Devarajan}

(1987), Benjamin (1993), and Subramanian (1996) for Cameroon; Dorosh and Lundberg (1996) for The Gambia; Dorosh (1996) for Madagascar; and Dorosh, Essama-Nssah, and Samba-Mamadou (1996) for Niger. See also Shoven and Whalley (1992) for a detailed discussion of the data needs and the implications of alternative equation specifications in the CGE models of this type.

5_{Private transfers, particularly international transfers, were small in all four countries}

(6)

In all four models, prices and wages clear markets,6_{labor supply}

is fixed,7_{imports and exports are less than perfect substitutes for}

local goods, and government spending is exogenous. Investment in the models is savings driven, with total savings equal to the sum of households’ savings (a linear function of households’ in-comes) and government savings (revenues less recurrent expendi-tures) determined by total savings. Foreign savings (foreign capital inflow) is held fixed across the model simulations presented here, so as to facilitate comparison of income levels across simulations.8

No attempt is made to model monetary or financial variables. Given these market clearing assumptions, the models do not accurately reflect short-term (less than 1 year) consequences of adjustment, but may be taken as indicative of medium-term adjust-ment to the shocks and policy changes simulated. Sizeable evi-dence suggests, however, that markets, particularly labor markets in African economies generally perform well, and that wages are flexible (except in the public sector) (Lavy and Newman, 1989; Colclough, 1991; Hoddintot, 1993; Horton, Kanbur, and Ma-zumdar, 1994). In Ghana, less than 10 percent of government rede-ployees were unemployed 1 year after redeployment (Younger, 1996), though in Guinea, male government redeployees appeared to queue for formal sector wage jobs (Mills and Sahn, 1993).

The major differences between the models lie in the structures of the economies modeled. Petroleum exports dominate Cameroon’s foreign exchange earnings, but the sector uses little domestic labor or other resources. The large groundnut export trade, the reexport trade, small manufacturing base, and reliance on noncompetitive imports make the Gambian economy very sensitive to changes in the real exchange rate and foreign capital inflows. Rice dominates food production, consumption, and trade in Madagascar, and ex-port crops account for 61 percent of total exex-port earnings. In Niger, much of private investment is in the livestock sector, forging a direct link between national savings and rural incomes.

6_{In the simulations of foreign exchange rationing below, an implicit tariff is modeled}

that raises the price of imports in the open (parallel) market and brings about equilibrium in the foreign exchange market.

7_{In an earlier version of the simulations of adjustment to terms of trade shocks presented}

here, a non-zero labor supply elasticity was used to simulate underemployment (Dorosh and Sahn, 1993). This alternative specification does not lead to qualitatively different results.

(7)

In presenting the simulation results, we focus on GDP growth, investment, and foreign savings, as well as income distribution. To facilitate comparison, our functional income distribution distin-guishes between four groups: urban nonpoor, urban poor, rural nonpoor, and rural poor. These groups are aggregated, according to the level of income, from those found in the income distribution of the underlying Social Accounting Matrices (SAMs).9_Their

in-comes per capita in local currency are found in Appendix Table 2A, along with shares of revenue from factors of production and transfers.10

4. ADJUSTING TO THE ECONOMIC CRISIS

Each of the four countries in this study suffered a terms-of-trade shock, concurrent with a tightening of world capital markets and foreign aid inflows. We, therefore, frame our analysis around the policy response to the adverse shocks experienced by most countries in sub-Saharan Africa in the late 1970s and early 1980s. In particular, to facilitate comparison between countries, the same terms-of-trade shock is modeled for each country: a 10-percent decline in the world price of the major export good.11 _In

addition, foreign capital inflows are reduced by an amount equal to 10 percent of base year exports.

4A. Simulating Alternative Policies

In the early stages of confronting their economic crises, many African countries avoided necessary adjustments in nominal and real exchange rates by rationing foreign exchange, often through the imposition of import quotas, a policy we refer to as de facto adjustment. In Simulation I, we model this policy alternative by an implicit tariff on imports (equivalent to a premium on foreign

9_{For details on the SAMs, see Dorosh and Essema-Nssah (1991), Dorosh et al. (1991),}

Gauthier and Kyle (1991), and Jabara, Lundberg, and Sireh Jallow (1992).

10_{Note that in the Gambia and Niger models, the urban poor are classified according}

to their incomes relative to other urban households rather than vis a` vis a national poverty standard. Thus, average incomes of the urban poor in these countries are higher than average incomes of the rural nonpoor.

11_{World prices were reduced for the following export commodities: Cameroon (export}

(8)

exchange for imports).12 _{With the quota rents accruing to}

high-income urban households. The level of the implicit tariff is set so as to keep the real exchange rate fixed at the preshock level, thus highlighting the role of real exchange rate depreciation (or the lack of it) in adjustment.

In Simulation II, we model adjustment with a liberalized foreign exchange regime, allowing the real exchange rate to depreciate in the wake of the shock. In reporting the results, we compare the simulated outcomes from Simulation II (adjustment through real devaluation) with Simulation I (de facto adjustment). Next, we examine variants on Simulation II by allowing the real ex-change rate to adjust while reducing government spending (Simu-lation III). Finally, we model the imposition of taxes on foreign trade to help maintain government revenues in the face of weaken-ing world prices for exports (Simulation IV). In this last simulation, no exogenous change is made in government expenditures. Be-cause Simulations III and IV are variants of the real exchange rate devaluation simulation, these results are presented relative to Simulation II.

In all cases, the results are anchored to the same macroeconomic constraint: the current account deficit is held constant across the simulations. Because each of the simulations are designed to meet the same macroeconomic adjustment objectives, the relative out-comes are comparable across countries in terms of the policies undertaken and, within a country, in terms of the macroeconomic target that alternative means of adjustment will achieve.

Simulation I: de factoadjustment. Table 2 shows the effects of the external shocks with de facto adjustment. Not surprisingly, real GDP falls sharply in all countries (1.8 to 3.6%) as lower export earnings and foreign capital inflows reduce aggregate demand and incomes. By construction of the simulations, there is no change in the real exchange rate. The premium on foreign exchange that helps reduce import demand and equilibrate the external accounts ranges from 14.2 percent in Niger to 35.1 percent in Madagascar. Structural characteristics of the economies, including the size of the external shock relative to investment spending and overall GDP, explain the variations in foreign exchange premiums across countries. For all four countries, the external shock (measured in terms of lost foreign exchange earnings) is between 9.7 and 12.0

(9)

MACROECONOMIC

ADJUSTMENT

AND

POVERTY

761

Table 2: Simulation I: De Facto Adjustment—Impose Import Quotas

Percentage changea

Cameroon The Gambia Madagascar Niger

Real GDP 21.83 23.55 22.32 22.94

Consumption/GDP 22.09 22.76 22.49 21.33

Total investment/GDP 20.89 26.37 21.03 23.78

Government recurrent expenditures/GDP 0.00 0.00 0.00 0.00

Government revenue/GDP 20.73 20.97 20.88 20.12

Exports/GDPb

22.21 0.75 20.98 21.25

Imports/GDPb

23.36 24.83 22.18 23.45

Change in foreign savings/base year exports 210.00 210.00 210.00 210.00

Prices

Foreign exchange premium (level in percent) 26.83 16.97 35.10 14.22

Real exchange rate 0.00 0.00 0.00 0.00

Real wage rates

Skilled labor 29.97 211.01 219.58 211.00

Semiskilled labor 29.08 27.87 217.33 —

Unskilled labor 212.02 24.64 28.27 25.44

Real incomes

Urban nonpoor 21.14 4.32 16.33 14.33

Urban poor 28.81 28.75 213.51 26.68

Rural nonpoor 210.09 26.51 25.61 24.09

Rural poor 210.04 25.71 25.25 24.67

Small farm—export-oriented 25.77 24.17

Total 23.22 23.78 23.23 21.52

Source: Model simulations.

a_{Percentage change relative to base SAM.}

(10)

percent of the initial level of imports. However, for The Gambia (the most open economy considered), the ratio of the shock to GDP is 6.3 percent, compared to 1.7 to 3.5 percent for the other three countries. Given the large reduction in foreign savings and export earnings, total savings in The Gambia declines steeply, and along with it, total investment and demand for services of the labor-intensive construction sector. Real wages for both skilled and unskilled urban workers fall, helping to reduce the price of hotel and other services and permitting an increase in tourism exports. Thus, the loss of foreign exchange revenues is partially offset, so that a premium of only 17.0 percent is sufficient to restore equilibrium in foreign exchange markets. In Niger, the initial foreign exchange shock is also relatively large, 3.5 percent of GDP, so that investment (and consumer) demand for livestock falls. As a result, the exportable surplus of livestock rises, softening the decline in foreign exchange availability, so that the simulated foreign exchange premium is only 14.2 percent. In contrast, with Madagascar’s very closed economy, the shock is a small 1.7 percent of GDP, so there is little impact on savings (and thus total invest-ment). Because investment demand for manufactured imports (which accounts for a large share of total imports) changes little, the bulk of the adjustment must be borne by a reduction in con-sumer demand. To achieve this shift in demand, a large increase (35.1 percent) in the exchange rate premium (and the price of imported manufactured goods) is required.

The price elasticity of the export good suffering the simulated terms-of-trade shock also significantly affects the size of the pre-mium by influencing he extent of the loss of foreign exchange earnings through decreased volume of exports. In Niger, uranium export supply is very inelastic so the 10-percent decline in the price of uranium reduces total uranium revenues by only 11.7 percent. The 10-percent decline in world prices of Madagascar’s export crops (coffee, vanilla, cloves) reduces export revenues from these crops by 17.1 percent.

(11)

urban high-income households actually see a rise in their real incomes in this scenario, however, because they collect the rents associated with the import quotas. Of course, in practice, it is likely that only a small number of the urban elite would gather the lion’s share of the rents. Urban nonpoor households not receiving rents suffer a decline in real incomes in this scenario.

Simulation II: Adjustment through real exchange rate deprecia-tion:in the second simulation we allow the real exchange rate to adjust freely to the negative terms-of-trade shocks to restore the equilibrium in the external accounts. Although most nominal deval-uations are actually accompanied by changes in fiscal policies (e.g., cuts in government spending) or trade policies (e.g., reductions in trade taxes), such policies remain unchanged in this simulation.

Compared with thede factoadjustment simulation (Simulation I), liberalization of foreign exchange markets leads to a real exchange rate depreciation of 15.7 percent in Cameroon, 4.3 percent in The Gambia, 11.5 percent in Niger, and 9.2 percent in Madagascar (Table 3).13 _{Real GDP is also slightly higher (by 0.2 to 1.3%) in}

all countries. The depreciation of the real exchange rate spurs exports by 3.2 and 3.5 percent in Niger and The Gambia, respec-tively, and by 9.0 percent in Cameroon and 10.4 percent in Mada-gascar. This increase in export earnings permits a higher level of imports as well.

But what of the effect of such policies on income distribution? As intimated above, much of the early writing on the deleterious impact of adjustment focused on the harmful effects of exchange rate adjustments on the poor, who were assumed to be net consum-ers of tradable staple foods and net suppliconsum-ers of labor, and for whom a combination of greater unemployment and lower wages would have adverse consequences. In practice, the results indicate that with the removal of the quota rents, the urban nonpoor households suffer a loss in real incomes (by 4.6 to 20.4%) despite the overall increase in economic activity and an increase in real wages for skilled labor (Table 4). Lower incomes for the urban

13_{Although nominal exchange rates vis-a`-vis the French franc were fixed for CFA}

(12)

P.

A.

Dorosh

and

D.

E.

Sahn

Table 3: Simulation II: Real Devaluation

Real GDP 0.68 1.25 1.28 0.23

Consumption/GDP 0.85 1.13 0.74 20.50

Exports/GDPb _2.25 _2.04 _1.16 _0.68

Imports/GDPb _2.25 _2.04 _1.16 _0.68

Prices

Foreign exchange premium (level in percent) 0.00 0.00 0.00 0.00

Real wage rates

Skilled labor 6.88 8.62 18.32 5.43

Unskilled labor 9.77 1.88 6.97 2.90

Real incomes

Urban nonpoor 215.92 24.61 220.36 218.92

Urban poor 5.51 8.37 11.56 2.94

Rural nonpoor 7.91 2.88 4.23 2.13

Rural poor 7.39 2.78 4.53 2.69

Total 1.26 1.84 0.84 21.71

a_{Percentage change relative to Simulation I.}

(13)

Table 4: Simulation II: Income Breakdown (Percentage Change)

Urban Urban Rural Rural

nonpoor Poor nonpoor Poor Total

Cameroon

Formal skilled labor 0.761 1.137 0.855 0.209 0.718

Formal semi-skilled labor 1.198 1.751 1.597 0.402 1.280

Formal unskilled labor 0.229 1.538 0.641 0.679 0.540

Informal labor 0.158 1.891 2.916 7.108 2.881

Land 0.000 0.000 0.909 0.335 0.553

Capital 1.076 0.000 0.293 0.000 0.455

Government transfers 0.357 0.000 0.045 0.000 0.123

Firm transfers 0.831 0.000 0.226 0.000 0.352

Rents 220.770 0.000 0.000 0.000 25.716

Subtotal 216.160 6.317 7.482 8.733 1.186

Consumer price effect 0.286 20.756 0.398 21.231 0.072

Total 215.770 5.513 7.910 7.394 1.259

The Gambia

Skilled labor 3.005 3.476 0.238 0.105 1.833

Unskilled labor 2.059 4.012 0.845 1.260 2.075

Largeholder land 0.003 0.000 0.280 0.000 0.060

Smallholder land 0.000 0.005 0.000 0.032 0.009

Informal capital 2.561 0.678 0.202 0.171 1.039

Housing 0.056 0.068 0.237 0.299 0.156

Transfers 0.118 0.103 0.936 0.361 0.345

Rents 211.683 0.000 0.000 0.000 23.623

Subtotal 23.880 8.341 2.738 2.228 1.893

Total 24.614 8.369 2.881 2.777 1.836

Madagascar

Skilled labor 5.863 0.000 0.000 0.000 1.028

Semi-skilled labor 0.198 7.761 1.007 0.000 1.203

Unskilled labor 0.036 2.350 2.105 4.607 2.765

Largeholder land 0.000 0.000 1.809 0.000 0.574

Smallholder land 0.000 0.000 0.000 0.796 0.317

Dividends 6.040 0.000 0.000 0.000 1.059

Rents 235.000 0.000 0.000 0.000 6.253

Subtotal 223.536 10.254 5.200 5.472 0.824

Total 220.358 11.550 4.226 4.528 0.838

Niger

Skilled labor 3.380 0.000 0.000 0.000 0.678

Unskilled labor 0.138 1.731 1.592 2.658 1.658

Large holder land 0.000 0.000 0.584 0.000 0.207

Small holder land 0.000 0.085 0.000 0.159 0.061

Rents 223.534 0.000 0.000 0.000 24.722

Subtotal 219.803 2.426 2.670 2.887 21.800

Total 218.919 2.934 2.127 2.687 21.815

(14)

nonpoor, who have relatively high marginal propensities to save, help offset the gain in savings of other households so that total savings (and total investment) rise relatively little (and actually falls slightly in Cameroon).

Most important, the average real incomes of the poor, including the rural poor who have lowest per capita incomes, rise with real exchange rate depreciation. This increase results because the removal of distortions in the prices of importables leads to a more efficient allocation of resources: demand for unskilled labor increases, as do returns to informal sector capital, benefitting poor households in both urban and rural areas. Urban poor households also benefit from lower prices of importable goods they consume. For example, the reductions in the household consumer price indices for urban poor households in Madagascar and Niger raise their real incomes by 4.16 and 1.10 percent, respectively (Table 4). A major factor contributing to increased real wage payments to unskilled labor for rural households is increased profitability of agricultural traded goods, due to the depreciation of the real exchange rate. Real incomes of small farmers who cultivate export crops in Madagascar increase by 5.0 percent compared to the average country-wide increase of only 0.84 percent.14,15

Simulation III: Reductions in government spending:In Simula-tion III, real government expenditures are cut by 10 percent to increase government savings (reduce the government deficit) and

14_{Our modeling assumptions regarding household transfers have little impact on these}

results. Not explicitly incorporating international transfers to households in the Cameroon, Madagascar, and Niger models is equivalent to assuming that all international transfers accrue to the urban nonpoor, and that these transfers are fixed in dollar terms. National household survey data from Madagascar (Government of Madagascar, 1993), in fact, show that reported international transfers are both small and concentrated among the urban nonpoor, with per capita transfers equal to FMG 54,522 (1.1% of expenditures) for the richest 5 percent of households and only FMG 1,747 (0.2% of expenditures) for the poorest 40 percent. Data in Ghana show a similar pattern. (Note that these national household surveys each captured about one-quarter of private household transfers shown in the balance of payments.)

15_{The result that real exchange rate depreciation tends to benefit the poor is consistent}

(15)

permit an increase in private investment (Table 5).16 _{Because, in}

this, and the two following simulations, the real exchange rate is allowed to adjust without an imposition of quotas, results of these runs are thus compared with the results of Simulation II. The biggest effect of reducing the size of government is that it allows investment to increase by 4.5 percent (Cameroon) to 20.3 percent (Niger). Government recurrent expenditures are very large rela-tives to private investment in Madagascar and Niger, so the change in investment is greater in these two countries. Real GDP falls slightly in The Gambia, Madagascar, and Niger, and the real ex-change rate depreciates slightly in all countries (0.3 to 0.8%). Gov-ernment revenues fall in Niger as income taxes on wages of skilled government workers (an important source of revenues) decline.

In general, urban households are hurt more by the decline in government expenditures because government employment is concentrated in urban areas. The increase in investment spending tends to help offset this decline, however. The effect on rural households varies across countries: rural households enjoy slight increases in real incomes in The Gambia and Madagascar, mainly due to effects of the small depreciation of the real exchange rate in boosting incomes for producers of tradable commodities.

Simulation IV: Maintaining government revenues through in-creased trade taxes:in Simulation IV, in addition to allowing the real exchange rate to adjust, taxes on foreign trade are increased uniformly by 10 percent in an attempt to maintain government revenues in spite of the shock of lower world prices for exports and reduced foreign capital inflows.17

Greater trade taxes in this simulation raise real tax revenues by 8.4 to 32.5 percent, thus increasing the pool of total savings in the economy (Table 6). Real investment increases sharply in all four economies. Taxing trade, however, introduces distortions in the economy by reducing incentives for both exports and imports and leading to shifts in resources away from production of ex-portables and towards imex-portables and, to a lesser extent, non-traded goods. Real GDP thus falls by 0.1 to 2.5 percent in this

16_{Note that in the models, total investment is determined simply by the availability of}

total savings; thus, investment increases automatically with a cut in government expendi-ture, which increases government (and total) savings in the economy.

17_{Export taxes are imposed on the export sectors listed in footnote 4 except for The}

(16)

P.

A.

Dorosh

and

D.

E.

Sahn

Table 5: Simulation III: Real Devaluation with Reduction in Government Spending

Real GDP 0.03 20.38 20.20 20.18

Consumption/GDP 20.09 20.32 0.00 21.13

Exports/GDPb _0.07 _0.16 _0.10 _0.12

Imports/GDPb _0.07 _0.16 _0.10 _0.12

Real wage rates

Skilled labor 21.97 20.36 28.32 211.40

Unskilled labor 0.16 20.02 0.64 0.31

Real incomes

Urban nonpoor 0.11 20.76 0.93 27.66

Urban poor 20.78 20.97 23.12 20.96

Rural nonpoor 20.28 0.14 0.04 20.66

Rural poor 20.11 0.05 0.56 20.44

Total 20.16 20.42 0.00 21.75

a_{Percentage change relative to Simulation II.}

(17)

MACROECONOMIC

ADJUSTMENT

AND

POVERTY

769

Table 6: Simulation IV: Real Devaluation with Increased Taxes on Foreign Trade

Real GDP 20.66 22.47 21.12 20.09

Consumption/GDP 21.72 25.74 22.18 22.19

Exports/GDPb

21.12 23.23 21.00 20.80

Imports/GDPb

21.12 23.23 21.00 20.80

Real wage rates

Skilled labor 21.96 212.97 28.01 23.00

Unskilled labor 27.65 24.08 24.99 22.86

Real incomes

Urban nonpoor 0.49 212.12 21.93 24.12

Urban poor 20.50 211.73 26.99 23.11

Rural nonpoor 23.76 24.02 22.14 22.35

Rural poor 24.54 23.34 22.98 22.77

Total 22.92 28.28 23.05 22.87

a_{Percentage change relative to Simulation II.}

(18)

simulation. In Cameroon, The Gambia, and Madagascar the real exchange rate depreciates because the negative effect of raising taxes on agricultural exports in reducing supply of foreign ex-change is relatively more important than the effect of the increased import tariffs on reducing demand for foreign exchange.

Real incomes of all household groups except the urban nonpoor in Cameroon fall, despite the increase in investment spending. De-clines in urban incomes are generally larger than deDe-clines in rural incomes, as reduced economic current domestic production leads to declines in returns to informal and formal sector capital. Farmers producing export crops tend to suffer larger declines in real incomes than other small farmers, because of the decline in the real domestic price of export crops due to the increased export taxes.

5. CONCLUSIONS

The impact of economic reforms on growth and income distribu-tion has been one of the most keenly debated subjects of the past decade. The simulations presented above suggest a number of general conclusions relevant to this debate. First, the terms-of-trade shocks that were so pervasive in Africa-reduced real incomes for most household groups, especially the producers of commodi-ties with falling world prices. The failure to reform policy and allow the exchange rate to adjust, but instead employ rationing and quotas to cope with imbalances in the external accounts, however, was to the clear detriment of the poor. The beneficiaries of maintaining an overvalued exchange rate and related policy distortions were the urban nonpoor. The political economy of distortions favoring the urban elite, the most influential interest group, without doubt explains why such policies prevailed for so long without being challenged, even in the face of obvious eco-nomic decline.

(19)

overall equity. Lower income urban groups, however, tend to suffer more through the cut in recurrent expenditures.

Fourth, the models indicate that there is considerable latitude for achieving welfare objectives without compromising macrosta-bility and long-term growth. For example, the continued availabil-ity of concessional financing is important to maintain aggregate incomes. However, efforts must be made to ensure that foreign borrowing is used in ways to limit adverse effects of appreciation of the real exchange rate. This includes channeling aid to meet the needs of the poor, directly raising consumption of, and invest-ments in, the poor. Equally important is limiting the real apprecia-tion itself through offsetting macroeconomic and trade policies, such as more rapid trade liberalization and greater fiscal restraint. By preventing a drop in the relative price of traded to non-traded goods, incentives for agricultural production can be maintained and rural incomes enhanced. In this way, incomes of small farmers will generally be shielded from adverse movements in relative prices. While avoiding appreciations of the real exchange rate is a very indirect way of affecting income distribution, it is potentially the most effective and most far-reaching economic policy for rais-ing real incomes of the large numbers of poor in sub-Saharan Africa in the short to medium run.

Fifth, the composition of investment goods, including human capital, is a critical determinant of the income distribution effects when total savings and investment in the economy change. In most countries, investment goods are predominantly imported or produced in urban areas, so that rural households enjoy little direct benefit from investment booms. Thus, greater consideration needs to be given to the destination of investment by sector, as well as the factor intensity of the production technology, which affects income distribution once the new capital is in place.

Finally, the magnitude of the effects and the differential impact of policies on various income groups in the four countries illustrate the importance of proper country-specific policy analysis. Like-wise, the indirect effects of certain policies often outweigh the direct, or expected, effects, further increasing the importance of examining policies in a general equilibrium framework.

REFERENCES

(20)

Benjamin, N.C., Devarajan, S., and Weiner, R.J. (1989) The ‘Dutch’ Disease in a Devel-oping Country.Journal of Development Economics30:71–92.

Bourguignon, F., Branson, W.H. and de Melo, J. (1989a) Macroeconomic Adjustment and Income Distribution: A Macro-Micro Simulation Model. Technical Papers No. 1. Paris: OECD Development Centre.

Bourguignon, F., Branson, W.H. and de Melo, J. (1989b) Adjustment and Income Distribu-tion: A Counterfactual Analysis. NBER Working Paper, Cambridge, MA. Colclough, C. (1991) Wage Flexibility in sub-Saharan Africa: Trends and Explanations.

InTowards Social Adjustment: Labor Market Issues in Structural Adjustment(G. Standing and V. Tokman, Eds.). Geneva: ILO.

Condon, T., Dahl, H. and Devarajan, S. (1987) Implementing a Computable General Equilibrium Model on GAMS: The Cameroon Model. DRD Discussion Paper, Report No. DRF290, Washington DC: The World Bank.

Cornia, G.A., Jolly, R. and Stewart, F., Eds. (1987) Adjustment with a Human Face: Protecting the Vulnerable and Promoting Growth.Oxford: Clarendon Press. de Melo, J., and Robinson, S. (1982) Trade Adjustment Policies and Income Distribution

in Three Archetype Developing Economies.Journal of Development Economics 10:67–92.

Dervis, K., de Melo, J. and Robinson, S. (1982)General Equilibrium Models for Develop-ment Policy.Washington, DC: The World Bank.

Dorosh, P.A. (1996) Rents and Exchange Rates: Redistribution through Trade Liberaliza-tion in Madagascar. InEconomic Reform and the Poor in Africa(D.E. Sahn, Ed.). Oxford: Oxford-Clarendon.

Dorosh, P.A., Bernier, R.E., Randrianarivony, A.R. and Rasolomanana, C. (1991) A Social Accounting Matrix for Madagascar: Methodology and Results. Working Paper No. 6, Ithaca, NY: Cornell Food and Nutrition Policy Program.

Dorosh, P.A., and Essama-Nssah, B. (1991) A Social Accounting Matrix for Niger: Method-ology and Results. Working Paper No. 18, Ithaca, NY: Cornell Food and Nutrition Policy Program.

Dorosh, P.A., Essama-Nssah, B., and Samba-Mamadou, O. (1996) Terms of Trade and the Real Exchange Rate in the CFA Zone: Implications for Income Distribution in Niger. InEconomic Reform and the Poor in Africa(D.E. Sahn, Ed.). Oxford: Oxford-Clarendon.

Dorosh, P.A., and Lundberg, M.K.A. (1996) More than just Peanuts (Groundnuts): Aid Flows and Policy Reform in The Gambia. InEconomic Reform and the Poor in Africa(D.E. Sahn, Ed.). Oxford: Oxford-Clarendon.

Dorosh, P.A., and Sahn D.E. (1993) A General Equilibrium Analysis of the Effect of Macroeconomic Adjustment on Poverty in Africa. Working Paper No. 39, Ithaca, NY: Cornell Food and Nutrition Policy Program.

Gauthier, M., and Kyle, S. (1991) A Social Accounting Matrix for Cameroon. Working Paper No. 4, Ithaca, NY: Cornell Food and Nutrition Policy Program.

Government of Madagascar. 1993.Enqueˆte permanente aupre`s des me´nages 1993(data tapes). Antananarivo, Madagascar.

Hoddintot, J. (1993) Wages and Unemployment in Urban Coˆte d’Ivoire. Working Paper No. 93.3. Oxford: Centre for the Study of African Economies.

(21)

International Monetary Fund (IMF). (various years)Government Financial Statistics Year-book.Washington, DC: IMF.

Jabara, C.L., Lundberg, M. and Jallow, A.S. (1992) A Social Accounting Matrix for The Gambia. Working Paper No. 20, Ithaca, NY: Cornell Food and Nutrition Policy Program.

Lavy, V., and Newman, J. (1989) Wage Rigidity: Micro Evidence on Labor Market Adjust-ment in the Modern Sector.The World Bank Economic Review3:97–117. Mills, B., and Sahn, D.E. (1993) Is There Life After Public Service: The Fate of Retrenched

Workers in Conakry, Guinea. Working Paper No. 42. Ithaca, NY: Cornell Food and Nutrition Policy Program.

Radelet, S. (1993) The Gambia’s Economic Recovery: Policy Reforms, Foreign Aid, or Rain?Journal of Policy Modeling15:251–276.

Sahn, D.E., Ed. (1994)Adjusting to Policy Failure in African Economies. Ithaca, NY: Cornell University Press.

Sahn, D.E., Dorosh, P.A. and Younger, S.D. (1996) Exchange Rate, Fiscal and Agricultural Policy in Africa: Does Adjustment Hurt the Poor?World Development24:719–747. Schneider, H., Weekes-Vagliani, W., Groppo, P., Lambert, S., Suwa, A. and Nguyen Tinh,

N. (1992)Adjustment and Equity in Coˆte d’Ivoire.Paris: OECD.

Shoven, J.B., and Whalley, J. (1992)Applying General Equilibrium.Cambridge: Cambridge University Press.

Subramanian, S. (1996) Vulnerability to Price Shocks under Alternative Policies in Camer-oon. InEconomic Reform and the Poor in Africa(D.E. Sahn, Ed.). Oxford: Oxford-Clarendon.

Taylor, L. (988)Varieties of Stabilization Experience: Towards Distribution, and Growth in the Developing World.Cambridge: MIT Press.

Tyler, G.J., and Akinboade, O. (1992) Structural Adjustment and Poverty: A Computable General Equilibrium Model of the Kenyan Economy.Oxford Agrarian Studies 20:51–61.

United Nations Economic Comission for Africa (UNECA). (1989)Statistics and Politics— ECA Preliminary Observations on the World Bank Report.Addis Ababa, Ethiopia: Author.

World Bank, The. (1992)African Development Indicators.Washington, DC: International Bank for Reconstruction and Development/The World Bank.

World Bank, The. (1994)Adjustment in Africa: Reforms, Results, and the Road Ahead. A World Bank Policy Research Report.New York: Oxford University Press. World Bank, The. (1995)World Tables.Washington, DC: The World Bank.

(22)

Table A1: Trade Levels and Parameters: Cameroon, The Gambia, Madagascar, and Niger

Elasticity of Production Imports Exports substitution

Cameroon

(billion 1984/85 CFA francs)

Food Agriculture 588.33 14.93 6.09 1.2/1.5a

Export Agriculture 425.06 14.53 169.19 0.8/1.2

Forestry 167.05 0 31.52 0.4/1.2

Estate Sector 32.88 1.12 13.09 0.8/1.2

Food Proc., private 288.54 24.65 17.82 0.9/0.9

Food Proc., public 34.15 3.05 2.21 0.9/0.9

Manufacture, private 891.56 598.35 95.65 1.1/0.7

Manufacture, public 162.22 31.79 8.97 1.1/0.7

Construction 503.83 0 0 0.4/0.4

Services, private 2009.35 21.92 71.41 0.4/0.5

Services, public 244.09 3.97 12.93 0.4/0.4

Public Admin. 372.68 0 0 0.4/0.4

Oil 915.70 13.60 721.43 3.0/2.0

Total 6635.44 727.91 1150.31 —

The Gambia

(million 1989/90 dalasis)

Groundnuts 199.4 0.0 67.5 2.0

Rice 48.3 91.3 0.0 2.0

Coarse Grains 114.1 30.8 0.0 2.0

Fruits, Vegetables 103.3 25.8 32.7 2.0

Livestock, Forestry 201.6 15.6 22.2 0.9

Groundnut Products 117.0 0.0 77.1 2.0

Manufacturing 231.0 14.1 16.5 0.9

Construction 183.9 0.0 0.0 0.9

Transport, Communications 292.0 0.0 43.2 0.9

Informal Trade 71.5 0.0 0.0 0.4

Formal Trade 253.7 0.0 0.0 0.4

Re-Exports 723.1 0.0 723.1 2.0

Private Services 390.8 0.0 207.3 0.7

Public Services 231.8 0.0 0.0 0.4

Urban Housing 95.0 0.0 0.0 0.4

Rural Housing 89.2 0.0 0.0 0.4

Non-competitive Imports 0.0 1191.0 0.0 2.0

Total 3345.9 1368.5 1189.4 —

(23)

Table A1: Continued

Elasticity of Production Imports Exports substitution

Madagascar

(billion 1984 FMG)

Paddy 168.2 0.0 0.0 2.0

Other food crops 237.9 3.7 2.0 0.9

Export crops 43.3 0.0 35.2 5.0

Industrial crops 14.2 0.0 0.0 2.0

Livestock 250.1 0.0 10.1 0.9

Energy 80.2 66.3 7.4 0.9

Milled rice 170.0 18.0 0.0 5.0

Processed food 322.2 8.6 12.9 0.9

Textiles 73.5 11.3 6.7 0.9

Manufactures 122.6 130.9 1.8 0.7

Construction 88.2 0.0 0.0 0.4

Transport 266.6 24.5 33.8 0.4

Commerce 346.8 0.0 0.0 0.4

Private services 298.0 26.0 3.4 0.4

Public administration 180.4 0.0 0.0 0.4

Total 2662.1 289.2 113.5 —

Niger

(billion 1987 FCFA)

Cereals 70.93 11.61 1.62 2.0

Export crops 23.82 0.21 14.12 2.0

Other food crops 56.84 6.44 2.71 0.9

Livestock 84.77 1.50 11.57 2.0

Forestry products 23.38 0.41 0.09 0.9

Mining 91.19 6.17 85.51 2.0

Meat 53.93 0.12 0.11 0.9

Processed food 17.19 20.02 1.35 0.9

Manufactures 84.55 141.14 13.91 0.7

Construction 55.83 0.00 0.00 0.4

Commerce 165.83 0.00 0.00 0.4

Transport./communic. 56.60 4.85 0.00 0.4

Private services 76.92 5.75 0.00 0.4

Public services 82.12 0.00 0.00 0.4

Total 943.92 198.21 131.00 —

Sources: Cameroon—Subramanian (1996); The Gambia—Dorosh and Lundberg (1996); Madagascar—Dorosh (1996); Niger—Dorosh et al. (1996).

a_{For Cameroon, the two numbers shown are the elasticity of substitution for imports}

(24)

Urban Urban Rural Rural nonpoor poor nonpoor poor

Cameroon

Formal skilled labor 0.149 0.165 0.125 0.030

Formal semi-skilled labor 0.261 0.282 0.259 0.063

Formal unskilled labor 0.071 0.354 0.149 0.154

Informal labor 0.022 0.198 0.307 0.731

Land 0.000 0.000 0.065 0.022

Capital 0.342 0.000 0.069 0.000

Government transfers 0.046 0.000 0.004 0.000

Firm transfers 0.108 0.000 0.022 0.000

Total 1.000 1.000 1.000 1.000

Per capita income 795 232 497 296

(‘000 1985 FCFA)

Population (percent) 10.4 0.7 49.4 39.5

The Gambia

Skilled labor 0.408 0.416 0.057 0.025

Unskilled labor 0.245 0.421 0.442 0.666

Largeholder land 0.002 0.000 0.151 0.000

Smallholder land 0.000 0.009 0.000 0.067

Informal capital 0.250 0.060 0.049 0.041

Housing 0.067 0.072 0.095 0.121

Transfers 0.029 0.022 0.206 0.080

Total 1.000 1.000 1.000 1.000

(1990 Dalasis)

Madagascar

Skilled labor 0.492 0.000 0.000 0.000

Semi-skilled labor 0.017 0.491 0.068 0.000

Unskilled labor 0.007 0.323 0.306 0.673

Dividends 0.484 0.000 0.000 0.000

Total 1.000 1.000 1.000 1.000

(‘000 1984 FMG)

Niger

Skilled labor 0.769 0.000 0.000 0.000

Unskilled labor 0.057 0.572 0.526 0.883

Formal capital 0.000 0.000 0.000 0.000

Total 1.000 1.000 1.000 1.000

(‘000 1987 FCFA)