- Email: [email protected]
Adjustment and income distribution in an agricultural economy: A general equilibrium analysis of Cameroon
World Development, Vol. 24, No. 6, pp. 1003-1013, 1996 Copyright 0 1996 Elsevier Science Ltd Printed in Great Britain. All rights reserved 0305-750X/96 $15.00 + 0.00
Pergamon
SO305-750X(%)000162
Adjustment and Income Distribution in an Agricultural
Economy: A General Equilibrium Analysis of Cameroon
NANCY BENJAMIN* U.S. International Trade Commission, Washington, DC, U.S.A. Summary. adjustment
To distinguish policy packages that have favorable characteristics for both structural and income distribution, this paper uses static and dynamically optimizing simulations of a
Computable General Equilibrium (CGE) model based on detailed country and household data from Cameroon. Model results indicate that public spending to reduce transport and marketing costs are beneficial for the poor and for economic growth, Copyright 0 1996 Elsevier Science Ltd
even at the expense
1. INTRODUCTION Structural adjustment refers broadly to the economic reorientation suited to coping with external shocks such as losses in terms of trade or in the availability of foreign loans. The policies used to encourage economic restructuring often include reduction and reallocation of government spending, changes in the structure of trade and excise taxes, and changes in the real exchange rate. While such adjustment plans consist of national-level policies, their effects are felt at the household level, particularly in terms of the distribution of income. Since many of the countries requiring structural adjustment are poor, observers have asked whether adjustment policies conflict or complement these countries’ poverty alleviation goals. The purpose of this paper is to analyze the implications for income distribution of alternative structural adjustment paths in a low-income, agricultural economy. We do so by performing counterfactual simulations on a general equilibrium model of a particular country, Cameroon. During the second half of the 198Os, Cameroon was subjected to three external shocks: oil prices declined, agricultural prices fell and Cameroon’s currency, the CFA Franc, being tied to the French Franc, appreciated substantially relative to the US dollar. There was no question that the coun-
try needed structural adjustment. But there are alternative tax and expenditure policies that could generate a given external balance, each with different implications for the distribution of income. In this paper, we examine these options and their implications, holding the level of current account reduction fixed, i.e., con-
of competing
public investments.
trolling for the overall level of adjustment required by the economy. In order to analyze structural adjustment and income distribution, we need a framework which can capture both the macroeconomic effects of external shocks and adjustment (i.e., shifts in the current account balance, etc.) and the microeconomic factors which determine income distribution - factor markets, product markets and household behavior. In addition, analysis of this type requires simulation of alternative paths or, in other words, a model which permits counterfactual simulatioirs. Both these criteria are met by computable general equilibrium (CGE) models which have a long history of analyzing income distribution issues (Adelman and Robinson, 1978; de Melo and Robinson, 1982; Bourguignon, Branson, and de Melo, 1992). Section 2 of this paper provides a thumbnail sketch of the data and CGE model used to analyze income distribution and adjustment in Cameroon. In section 3, we report on simulation results which reveal both the impact of external shocks as well as the implica-
*Research for this paper was undertaken as part of the Cornell University Food and Nutrition Policy Program project on Adjustment and Income Distribution in sub-Saharan Africa. I am grateful to Shantayanan Devarajan, Paul Dorosh, David Sahn and two anonymous referees for helpful comments on earlier drafts. The views expressed in the paper are the author’s own and not necessarily those of the US International Trade Commission. Final revision accepted: January 2,1996. 1003
WORLD DEVELOPMENT
1004
tions of alternative adjustment scenarios for the distribution of income among different household groups. Section 4 contains some concluding remarks.
2. DATA AND MODEL DESCRIPTION (a) Data sources The data used in the model come from a 1984-85 Social Accounting Matrix (SAM) described in Gauthier and Kyle (1990). The SAM starts with the aggregate balances given in the national accounts, adds sectoral input-output relations from the Cameroon input-output table, and disaggregates factor payments, household income and consumption according to data from the Cameroon household survey. Results of the household survey are detailed in Blandford and Lynch (1990) and Lynch (1991). The data allow for distinction in the model of seven household types. There are several agricultural households distinguished by region and income level, rural nonfarm households, and urban rich and poor. The overwhelming majority of the poor in Cameroon are rural, Most of the poor are farmers; the northern farmers are the poorest household group. Since the northern region is furthest from the capital and from major ports, northern farmers are isolated from other markets. They grow fewer cash crops and have less access to markets for food and consumer goods in other parts of the country (Blandford and Lynch, 1990).
(b) Model description The CGE model used for policy simulation is described in detail in Benjamin and Lee (1991). At this point, only certain model features are summarized for the purpose of understanding experiment results. A full set of equations is listed in the appendix. (i) Production and trade Production in each sector is characterized by Cobb-Douglas production functions with four labor types (rural and urban unskilled, skilled, and highly skilled) and sector-specific capital that can be augmented only between periods. Domestic goods are imperfect substitutes for imports in the same sector. Consumers purchase a CES aggregate of imported and domestic goods based on the relative price. Similarly, the price of exports can differ from world prices since Cameroon’s exports are considered distinguishable from world goods in the sector. Domestic producers will also respond, according to a constant elasticity transformation function that governs the rate at which domestic production can be transferred to export markets.
Each sector’s expenditures on trade and transport are accounted for in a manner like other intermediate purchases, but kept separate so that these services can be addressed by policy measures. For example, if the government spends more on road maintenance, the cost of operating and insuring transport equipment should fall, but not the cost of other services. Similarly, if the government works to improve storage facilities and reduce waste, this may reduce marketing margins but not other service inputs. (ii) Consumer demand Consumption behavior is governed by CES utility functions for each household. Consumption shares by commodity by household are taken from the household survey data. Income elasticities by good by household were also derived from the survey data. The methodology and results are in Lynch (1991). Given these component values of the utility function, minimum subsistence consumption levels are determined as a residual. The crossprice elasticities are a function of these minimum consumption levels. (iii) Macroeconomic closure Unless noted otherwise, the macroeconomic closure used in model experiments involves a fixed net flow of foreign exchange between Cameroon and the rest of the world. The numeraire of the model is the world price level or, equivalently, the price of foreign exchange. Thus the nominal exchange rate is fixed, but the domestic price level can vary against the world price level, and in this way the real exchange rate varies. The domestic price level is made up of the prices of pure nontraded goods and the prices of imperfect substitutes for world goods. The static general equilibrium simulation model solves for a single period based only on current values of variables. (No account is taken of the burdens of future debt repayment, for example.) The model can be run over several periods by incorporating each period’s investment less depreciation into the following period’s capital stock and updating total labor supply. In most cases, the results of model experiments are measured at the end of five years. All investment must be incorporated with the capital stock of the 11 sectors if it is to affect future output. The role of public investment can be seen in the flow of investment resources to public services and to the other public manufacturing sectors. Later, in an experiment, government investment will be directed toward one of the private sectors to simulate a change in public investment policy. Any decline in government tax revenue, such as from agriculture export taxes, will diminish government contributions to aggregate saving. Thus a tax cut (or public spending increase) transfers revenue from aggregate savings to private income, where only a fraction will be saved.
ADJUSTMENT
AND INCOME DISTRIBUTION
This simulation model allows us to pose the counterfactual question: Is adjustment to a negative shock better for low income groups than nonadjustment? It also allows us to compare the outcomes of different adjustment policies to determine which are best for income distribution. In addition, it allows the distinction of changes in the relative income positions of different households versus changes in absolute income levels. Given the concern for the welfare of poor households, policy makers must beware of improvements in relative income distribution that nevertheless generate unacceptable outcomes for the poor. (iv) Dynamic extension In addition to the simulation model, there is also a dynamic optimizing version of the model designed to check the growth characteristics of strategies whose distributional characteristics are tested with the simulation model. The dynamic model has the same sectoral and trade structure but lacks the detail related to income distribution. The model includes perfect foresight and maximizes the present discounted value of the utility of consumption over a 20-year period. All dynamic models face the problem of terminal conditions. To be solved, the model must define a steady state. In this case the steady state involves constant ratios of investment to capital stock and foreign debt to GDP. The 20-year period is necessary to establish a reasonably smooth growth path to this steady state. Several macro aggregates that are fixed in the static model - such as the time path of foreign capital flows, government consumption expenditures, household savings rates - are endogenous in the dynamic model. Clearly, solutions to the static model simulations are not optimal. Therefore, policy experiments tested with the simulation model cannot be replicated exactly with the optimizing model. The structure of an optimal growth path delineated by the dynamic model, however, can help in judging the growth characteristics of strategies mapped out by the simulation model. It allows us to consider optimal adjustment strategies while the simulation model can indicate their likely impact on income distribution.
3. SIMULATION RESULTS (a) Portrait of adjustment To set the stage for analysis of adjustment strategies, we first describe the effects of the most striking aspects of Cameroon’s terms-of-trade shocks using model simulations. High foreign exchange inflow during the oil boom certainly helped raise overall income levels. But one aspect of the foreign exchange windfall has strong implications for income distribution: real exchange rate appreciation.’ As mentioned earlier, spending an
1005
oil windfall domestically can bias relative prices in favor of nontraded goods and against agricultural exports. Thus, at the time negative terms-of-trade shocks hit Cameroon, poor farmers were already in a relatively disfavored position. But we can further infer that, faced with a decline in terms of trade, it is better for farmers’ relative income if Cameroon adjusts to the shock than if it increases foreign borrowing and does not adjust. Table 1 shows household income changes from a simulated drop in oil revenues. This is modeled as a capital outflow since Cameroon’s oil sector is an enclave with few production links with the rest of the economy (Benjamin, Devarajan and Weiner, 1989). In this case Cameroon’s capital outflow is increased by US$ 100 million. As expected, everyone’s income falls. But adjusting to achieve a larger current account surplus requires a decline in domestic prices against world prices, and this real depreciation helps farmers and their relative income position. Given the decline in oil income, the Cameroon government eventually had to cut back on spending. To evaluate the distributional effects of less public spending we need to consider that funds released by government cutbacks can be used in different ways. For example, it can all be used to accommodate greater capital outflow, or, at a given level of capital flows, it can all be added to domestic investment funds, or any combination of the two. To examine the two extremes, Table 2 shows first, the results of repeating the base run with a lower level of public consumption and second, a repeat of the foreign capital outflow reported in Table 1 with an accommodating decrease in public spending. In the second of these experiments, investment in each year is set at the baserun levels and public spending must adjust to the capital outflow. Of course, the results with higher capital Table 1. Differences from the base run in percentage growth overfive years
Foreign capital outflow Income for: Poor N. farmers Poor S. fanners Rich N+S farmers Nonfann rural Poor Nonfarm rural rich Urban poor Urban rich CPI
-6.9 -9.1 -8.8 -11.3 -11.2 -11.3 -11.0 -7.2
AP mfrs minus @ag AP mfrs minus AP ag in the “no-shock” case Source: Model simulations.
0.02 9.0
WORLD DEVELOPMENT
1006
Table 2. Differencesfrom the base run in percentage growth overfive years Lower public spending
Base run Income for: Poor N. fanners Poor S. farmers Rich N+S farmers Nonfarm rural poor Nonfarm rural rich urban poor Urban rich CPI
Higher capital outflow
2.0
4.5
0.9 1.3 -0.2 0.0 XI.1 0.6 a.01
-8.1 -7.4 -11.7 -11.4 -11.6 -10.4 -7.8
8.0
I.0
AP mfrs minus m ag AP mfrs minus m ag in the “no-shock” case
9.0
Source: Model simulations.
in both tables show lower income levels than the case with base level capital flows. But in checking only for the impact of reduced government spending, we see the impact is small, largely because public and private spending do not have sufficiently different effects on factor incomes. The contribution of lower public spending to real depreciation gives rise to the small distributional effects observed. outflow
(b) Alternative adjustmentpaths In evaluating adjustment strategies it helps to consider structural adjustment and income distribution as targets and that policy should aim at both targets. The targets should first be defined so that policies can be tested relative to them. Measurements of income distribution include indices based on income levels, such as poverty indices, and measurements of relative income positions comparing them to some egalitarian standard. For the purpose of this paper we consider only changes in relative incomes and record an “improvement” in income distribution when the income growth of the poorest households is high relative to that of the richest households. Moreover, since structural adjustment involves economic reorientation to absorb terms of trade shocks with minimal contraction of imports and growth, we consider the adjustment target to be foreign exchange generation or, in other words, importing capacity. There is some problem, however, with trying to hit two targets with one model. Typically, adjustment policies are compared in terms of how much foreign exchange they can “free up” or allow the economy to
do without. Such comparison requires a model closure where the international capital account is left open; each policy leads to a different level of the current account and the capital account accommodates this level. As shown above, however, different amounts of capital inflow lead to different domestic income levels. This complicates analysis of income distribution effects. While policies can be compared on their results for relative income distribution, such comparison is not always practical for the policy maker. Differences of income levels of several percentage points across adjustment plans will probably dominate any concern over differences in relative incomes. What is more likely is that policy makers have a target level for the current account or for income growth (or contraction) and are interested in the distributional consequences of alternative ways of hitting that target. For such an exercise, the model’s foreign capital account should be held at a constant level consistent with the target. In the base data, Cameroon transfers out US$162 million. This negative foreign exchange “shock” is held constant across policy experiments so that any improvement in income distribution is accomplished at a given level of capital outflow. In this way we hold the adjustment target constant and seek the greatest improvement in income distribution. Then the adjustment and growth characteristics of these policies can be examined. For example, any distribution-changing policy that succeeds in raising total income is consistent with a growth strategy, since the income expansion is achieved without reducing capital outflow and without removing any structural rigidities. Moreover, such a policy will generate forces in the same direction over a wide range of adjustment targets. Once we know which adjustment plans are also good distribution policies, we want to know which of the good distribution policies make good adjustment plans. To this end, the optimizing model can be used to check the compatibility between optimal growth strategies and those that improve income distribution. In considering policies that most affect income distribution, there is solid evidence supporting the importance of the agricultural terms of trade or the relative price of agricultural goods to manufactures (Adelman and Robinson, 1978). Since, in Cameroon, most of the poor are rural farmers, it is not surprising that the agricultural terms of trade would remain significant in experiments with the Cameroon model. But this raises the question: What most affects the agricultural terms of trade? The next set of experiments addresses this question beginning with policies intended to improve the relative price of agricultural output.
(c) Lower rnunufacturing
tariffs
Consider first a policy that could lower the price of
ADIIJSTMENT
AND INCOME DISTFUBUTION
manufactures, a significant intermediate and investment good for agriculture sectors. Since a large share of manufactures are imported, an effective instrument for reducing their cost is to cut the import tariff of 25% in half. With cheaper manufactures imports, the cost of production is lower for all other sectors. The results in Table 3 show that this effectively improves the agricultural terms of trade and the relative income of poor farmers. The decline in tariff revenue cuts income to the government, however, which saves all above a constant consumption level, and diverts these flows to others who will provide fewer savings to finance investment. While demand for manufactures is aided by the lower price, the slump in investment reduces a major source of demand for this sector. The policy leads to an overall decline in income. Such negative macroeconomic trends are not unlikely when trade liberalization is undertaken in a country with a fixed exchange rate. Overall efficiency gains are insufficient to compensate for the loss of competitiveness in one sector and the loss of tariff revenue.
1007
investment funds are again reduced from higher public expenditures on the subsidies. The agriculture subsidy produces favorable results for income distribution. There are, however, obvious problems with subsidizing domestic prices above world prices. As a practical matter it is difficult to maintain such price wedges when there are porous borders with neighboring countries which are exporting and importing the same commodities as Cameroon. In addition, since the early work with CGE models on income distribution, the biggest change has been the pressures of structural adjustment. Developing economies are increasingly required to conform to world prices for tradable goods. Under this constraint, there is often little scope either for raising the price of primary products or for lowering the price of manufactured inputs. This directs attention to improving the operation of domestic markets so that income can rise for the poor without using extensive trade taxes.
(e) Trade and transport margin subsidy
(d) Agricultural subsidy Suppose instead there is a subsidy designed to raise the producer price of agricultural goods. In this case the food and cash-crop sectors have their indirect taxes set to zero and they receive a 10% export subsidy. The subsidy is even more successful in improving the agricultural terms of trade and the relative income of farmers, while raising income overall. This time the favored sectors generate income for households with high budget shares for the sectors’ output. Therefore income expansion is possible even though
One possibility is to approach a problem identified by several observers of Cameroon: the large gap between wholesale and retail agriculture prices (World Bank, 1989). Indeed the data for Cameroon indicate that many sectors are making large purchases of trade and transport or marketing margin services. Without inducing any sudden productivity increases in trade and transport services, the model can simulate the effects of a government subsidy of marketing margins. Thus all sectors need as many services as before, but the government pays for 20% of the total package
Table 3. Differencesfrom the base run in the tetminal year (in percentage) Experiments
Income of: Poor N. farmers Poor S. farmers Rich N+S farmers Nonfarm rural poor Nonfarm rural rich Urban poor Urban rich CPI
Lower mfrs tariff
Ag subsidy
Margin subsidy (gov services)
Margin subsidy (construction)
-3.0 -4.1 4.0 -5.2 -5.1 -5.2 -5.0 4.3
5.6 2.4 2.9 -0.7 4l.9 -0.7 -1.1 2.1
5.1 1.5 2.0 -1.6 -1.1 -1.6 0.1 -1.6
7.8 2.7 3.7 -2.2 -1.3 -2.2 1.3 -2.3
2.0
3.0
1.5
AP mfrs minus APag AP mfrs minus AP ag in the “no-shock”
5.0 case
Source: Model simulations.
9.0
WORLD DEVELOPMENT
1008
of margin services. The subsidy forces government spending to increase by about 30%. If the structure of these new payments are the same as those for other publicly provided services, the impact on income distribution is as given in the column labeled Margin Subsidy (Gov Services) in Table 4. With government covering some of the cost of margin services, the private cost of delivering goods to final consumers is reduced. It is possible for producer prices to rise and consumer prices to fall. In this case the agricultural terms of trade improves as does the relative income of farmers. So reducing the burden of margin services is relatively more important for farmers whereas the decline in aggregate investment (due to higher public current expenditures) hurts them relatively less. There is an aggregation problem in that payments for marketing margins are observed for entire sectors in the model, but are actually concentrated on transported or marketed output. Likewise, the benefits of the margin subsidy affect value added throughout the sector, though in reality it will affect some producers more than others. What can be determined is that the household groups in the model that gain from the subsidy are those whose income sources are concentrated in sectors where marketing margins are high on average. In model experiments we can only suppose that government expenditures can reduce private marketing costs by some amount. The productivity of such a subsidy is a key parameter in these results. Model results indicate that if public spending could lessen
Table 4. D@erencesfrom the base run in terminal year (in percentage) Five-year run after two-year contraction Margin subsidy Margin subsidy (gov services) (construction) Income for: Poor N. farmers Poor S farmers Rich N+S farmers Nonfarm rural poor Nonfarm rural rich Urban poor Urban rich CPI
5.4 1.6 2.2 -1.7 -1.2 -1.7 0.1 -1.8
7.9 2.7 3.7 -2.3 -1.4 -2.2 1.3 -2.4
9.0
7.0
AP mfrs minus Nag AP mfrs minus AP ag in the “no-subsidy” case Source: Model simulations.
15.0
private marketing margin expenses, it would be good for income distribution. Given some possibility for success, the government may be able to reduce the effective cost of margin services to private sectors through various types of public expenditures. In the last experiment it was assumed that all new expenditures were made on public services. To test one other extreme, suppose that marketing margins can be subsidized through public expenditures on construction. As indicated earlier, current public spending on maintenance of roads and storage facilities can reduce trade and transport costs. Results from this experiment, assuming again a 20% subsidy, are listed under Margin Services (construction) in Table 4. The change in the composition of government spending accentuates the results of the earlier margin subsidy experiment. The gap between agriculture and manufactures prices is even smaller than with the direct agriculture subsidy. From the four experiment results presented so far, we can see that directing public expenditures toward reducing marketing margins can be similarly or more effective in reaching distribution and adjustment targets as direct approaches to the agricultural terms of trade. It could be argued that the role of trade and transport services - all nontraded activities -is exaggerated in the base data because 1984-85 reflects the peak absorption of oil revenues, which tends to inflate the price of nontraded goods relative to traded. After a few years of economic contraction and nontraded good price declines, the value of subsidizing margin services could be lower. This possibility is tested by repeating the margin subsidy experiment on a new base-year solution created from the original by contracting foreign capital for two years. After two years of contraction, the relative price of nontraded goods does decline. The results in Table 4 show, however, that the subsidy policies do not lose their effectiveness, even when applied to a deflated base-year solution. Clearly the size of margin services is not simply a matter of relative prices but involves the fundamental structure of the economy. These results differ from those in studies focused on increasing productivity by expanding public infrastructure (for example, Feltenstein and Morris, 1990). In such cases, the cost of producing infrastructure may be well known but its effectiveness in raising private productivity is simply imposed exogenously. With the Cameroon data, we know the size of the complete package of public and private trade and transport services purchased by each sector. The economy-wide value of reducing these costs can be determined and the expense deducted from the government account, but this does not determine the value of government spending on infrastructure. The last two modeling exercises look at the potential for investment schemes to contribute some of the effects generated by the current subsidies simulated above.
ADIUSTblENT
AND INCOME DISTRIBUTION
(f) Dynamic model simulation Reducing transport margins may be a productive way of raising the relative incomes of the poor in Cameroon, but is it also a good growth and adjustment strategy? By using the dynamic optimizing model we can check the consistency of this policy with a growth-maximizing investment strategy. The model finds the level and structure of investment that maximizes the present discounted value of the utility of consumption; thus it maximizes economic growth. Foreign borrowing and domestic saving are endogenous, but foreign debt must largely be repaid in order to meet the terminal condition on the ratio of foreign debt to GDP. The results in Table 5 compare the structure of the capital stock in the base year with the capital structure at the end of the optimal 20-year investment path. The results from this model show that to maximize growth, the share of capital in sectors producing trade and transport services must be increased (Table 5). Thus the productivity of investing in these areas is made apparent by the optimal growth path. It should be noted that the sectors with growing capital shares start with only average rates of return in the base year. Other sectors with higher returns lose capital share. Thus the optimizing model recognizes the bottleneck in services and the long-run value of expanding capacity in those sectors. While the optimizing model shows the growth advantage of investing in sectors providing marketing margin services, it does not guarantee that the income distribution results shown above will be sustained under an investment scheme as well as under direct current expenditures. If the government adds its own purchases of capital to those made by the service-supplying sectors, the added investment should bring down the cost of trade and transport services, though
1009
not as much as a direct subsidy. This approach has a more favorable outcome for total capital stock, but is less effective on income distribution compared to the current subsidy. The investment instrument is even more blunt if it brings down the price of all output to all users of the sectors producing margin services. It is possible to target margin services with an investment strategy by first, measuring the degree to which additional investment in the relevant sectors lowers their output price and then, combining such investment with a compensating tax on all uses other than marketing margins. Table 6 gives the results of this experiment where the amount of directed investment is equal to the amount spent on direct margin subsidies in the earlier experiments. The decline in price for margin services is 5% while the price for all other uses of these sectors remains very close to its base-run values. If the directed investment expenditures could be targeted toward adding only trade and transport capital stock, the impact on margin service prices would be greater.
5. POLICY CONCLUSIONS Evaluating the impact of adjustment programs on income distribution requires a general equilibrium framework for analysis and detailed data on the stmcture of the economy. For Cameroon these data show the sources of income of different households and the importance of trade and transport services for products produced by the poor. Given these two features, general equilibrium analysis demonstrates the value of lowering the cost of marketing margin services in improving the relative income of the poorest households. This finding has several implications: for adjustment strategy, for investment plans, and for exchange rate policy.
Table 5. Composition of capital stock in a dynamic, 20-year optimizing scenario Sectoral share of total capital stock in: (in percentage) Year 0 (Base) Year 20 1. Traditional food crops 2. Traditional export crops 3. Forestry 4. Agriculture modem sector 5. Food industries - private 6. Food industries - public 7. Manufacturing - private 8. Manufacturing - public 9. Construction 10. Services -private 11. Services -public Source: Model simulations.
7.5 5.9 3.6 0.9 4.9 0.6 33.0 5.3 10.7 21.0 6.3
8.0 5.1 3.9 0.8 3.4 0.4 26.9 4.2 14.5 25.4 7.3
WORLD DEVELOPMENT
1010
Table
6. Differences from the base run in percentage growth overfive years Targeted investment in margin services
Income for: Poor N. farmers Poor S. farmers Rich N+S farmers Nonfarm rural poor Nonfarm rural rich Urban poor Urban rich CPI
2.6 0.4 0.1 -2.1 -2.0 -2.0 -1.1 -0.4
AP mfrs minus mag AP mfrs minus AP ag in the “no-shock” case
7.0 9.0
Source: Model simulations.
Structural adjustment shocks place producers of traded goods under greater pressure from world prices, the producers of nontraded goods less so. The Cameroon data and model results indicate that those who profit from domestic trading routes are not the poor. Farmers gain more from having transport costs reduced than they lose in income from employment in the supplying sectors. Therefore, adjustment strategies should examine the possibility of reducing costs in sectors least pressured by world prices, but also how those sectors affect the employment of the poor and the price of goods produced by the poor.2 Public investment plans should also be considered in terms of their impact on income distribution. Model experiments show that poor households are favored by a policy where the government allocates funds away from investment into current subsidies of marketing margin services. This indicates that poor households do not gain much directly from investment expenditures.3 Nor, after five years, are the benefits of a higher capital stock sufficient to outweigh the advantage of lower margin costs for farmers. When the direct margin subsidy is replaced with capital investment to lower margin costs, the benelicial effects are diminished. Various factors combine
to lessen the effect of investment spending. One, indicated above, is the lack of employment of the poor in investment goods sectors. Another is the low capital intensity of the service sectors. This generates a low average productivity of capital and indicates that the greater efficiency gain is in lowering labor costs. Finally, model experiments only simulate the effects of adding capital to existing service technologies. They do not measure the benefits of adding new roads where none exists. The most common characteristic of poor farmers, however, is limited access to markets allowing cash sales of output and a variety of foods for purchase. This indicates the need for infrastructure investments to lessen the isolation of certain regions. Model simulations do show the value to all farmers of reducing the gap between wholesale and retail prices, even after two years of simulated macroeconomic contraction. This shows that high marketing margins arise from structural characteristics of the market, and not only from temporary conditions in relative prices. Given these features of the economic structure in Cameroon, however, the role of the exchange rate cannot be ignored. Poor farmers are engaged in producing both traded and (largely) nontraded goods, and exchange rate changes can have varying degrees of effectiveness in improving the market for cash crops. But the importance of certain nontraded services in total intermediate costs for farmers guarantees the value of real depreciation in improving income distribution in Cameroon. The general equilibrium analysis in this study enables us to simulate the distributional consequences of several public tax and spending policies and exchange rate policy. It also allows us to pinpoint those features of the economy having strong effects on income distribution. Once discovered, these features indicate fruitful areas for further investigation, such as: the market organization of transport and distribution services, the dependence of these services on public infrastructure, or the propensity of different farming households to market surplus output as transport costs fall. With such investigations we can look ahead to how structural adjustment programs may affect sensitive elements of the economy, and thus design reforms so that adjustment can proceed more smoothly.
NOTES 1. The impact of real exchange rate changes on income distribution and poverty is discussed in Kanbur (1987).
This conclusion further enhances the role of detailed analysis of distributional effects.
2. The impact of adjustment strategies on the nonpoor but powerful can affect the political feasibility of these strategies, according to de Janvry, Fargeix and Sadoulet (1991).
3. Blejer and Guerrero (1990) find similar results for the Philippines.
ADJUSTMENT
AND INCOME DISTRIBUTION
1011
REFERENCES Adelmau, I., and S. Robinson, Income Distribution Policies in Developing Countries: A Case Srudy of Korea (Stanford, CA: Stanford University Press, 1978). Benjamin, Nancy, and Yoon Lee, “Cameroon CGE models,” Mimeo (Ithaca: Cornell Food and Nutrition Policy Programs, (Cornell University, 1991). Photocopy. Benjamin, N., S. Devarajan, and R. Weiner, ‘The Dutch Disease in a developing country: Oil reserves in Cameroon,” Journal ofDevelopment Economics, Vol. 30 (January 1989), pp. 71-92. Blandford, David, and Sarah Lynch, “Structural Adjustment and the Poor in Cameroon,” Cornell Food and Nutrition Policy Program, Mimeo (Ithaca, NY: Cornell University, 1990). Blejer, M., and I. Guerrero, “The impact of macroeconomic policies on income distribution: An empirical study of the Philippines,” Review of Economics and Stafistics, Vol. LXXII, No. 3 (August 1990). Bourguignon, F., W. Branson, and J. de Melo, “Adjustment and Income Distribution: A Micro-Macro Model for Counterfactual Analysis.” Journal of Development Economics, Vol. 38, No. 1 (January 1992), pp. 1740. Devarajan, S., “Cameroon’s oil boom of 1978-85,” Mimeo (Washington, DC: The World Bank, 1991). Feltenstein, Andrew, and Stephen Morris, “Fiscal stabiliza-
APPENDIX:
tion and exchange rate instability: A theoretical approach and some policy conclusions using Mexican data,” Journal of Public Economics, Vol. 42 (August 1990). pp. 329-356. Gauthier, Madeleine, and Steven Kyle, “A Social Accounting Matrix for Cameroon,” Cornell Food and Nutrition Policy Program Working Paper No. 4 (Ithaca, NY: Cornell University, 1990). de Janvry, Alain, Andre Fargeix, and Elisabeth Sadoulet, “The political feasibility of rural poverty reduction.” Journal of Development Economics, Vol. 37 (November 1991), pp. 351-367. Kanbur, S. M. Ravi, “Measurement and alleviation of poverty,” IMFStafl Papers, Vol. 34, No. 1 (March 1987), pp. 60-85. Lynch, Sarah, “Income distribution, poverty and consumer preferences in Cameroon,” Ph.D. Dissertation (Ithaca, NY: Cornell University, 1991). de Melo, J., and S. Robinson, ‘“Trade adjustment policies and income distribution in three archetype developing economies,” Journal of Development Economics, Vol. 10 (1982), pp. 67-92. World Bank, “Cameroon: Country economic memorandum” (Washington, DC: The World Bank, 1989).
DEFINITIONS
AND EQUATIONS Export elasticity of demand Private marginal propensity to consume Minimum consumption level Government consumption share Export duty rates Tariff rates Indict tax rate Depreciation rates Proportion of the average wage by labor category and by sector Shares of investment by sector of destination Ratios of inventory investment to gross out-
(a) Indices
iJ if im 1 h a na agcY nacy r
sectors exportable sectors importable sectors Labor categories Household categories Agricultural sectors nonagricultnral sectors Agricultural capital income categories Nonagricultural capital income categories Time period
(b) Parameters maq maf? PC St
sub,
Marketing
margin as a rate of domestic out-
put Marketing margin as a rate of imports Private consumption share of public goods Subsidy rate on marketing margin Subsidy rates on private consumption of public goods CES utility (Armington) function shift parameter CES utility (Armington) function share parameter CES utility (Armington) function exponent Elasticity of substitution CET function shift parameter CET function share parameter CET function exponent Elasticity of transfonnation
a,, h I.I
tt;
PC, sub, -58
P,I
put Input-output coefficient Capital composition coefficient Share of marketing margin expenditures Aggregate private savings rate Private consumption shares of public goods by household group Subsidy rates on private consumption of public goods by household group Private savings rate by household category Private consumption shares by sector and by household group
(c) Variables
p< PQ PE, PWE,
Price of composite good i Price of domestic sales Domestic price of exports World market price of exports
WORLD DEVELOPMENT
1012
(d) Equations in the model
DKTOT DEPRECIA CTOT,
Domestic price of imports Domestic supply price World market price of imports World market price of exports Value added price Average wage rate by labor category Exchange rate Value of marketing margins Total value of marketing margins Price of trade margins Aggregate price index Return to capital Sector share of investible funds Composite goods supply Domestic output Domestic sales of domestic output Imports Exports Sector specific capital Inventory investment demand Employment by labor category and by sector Labor supply Private consumption demand for good i Consumption demand for good i by household group Total private consumption demand Public consumption demand for good i Total public consumption demand Investment demand by sector of origin Government revenue Total income of private sector Foreign savings Total savings Volume of investment by sector of destination Total investment Total depreciation charges Private consumption demand by household group Income by household category Total wage of agri sectors by labor category Total wage of non-agri sectors by labor cate-
MMTOT = C, MM,
Yh TWAG, TWNA,
c, L,, = I&
(13)
gory Total Total Total Total
(iv) Intermediate demand W, = Z, a, XD,
(14)
(v) Consumer demand C,, = 7,* + pih (CTOTIP,) + PC,, pi” GDTOT ( 1 - sub,,)
(15)
CTOT, = (1 - s,,) Yh
(lo)
PM px, PMW, PWE, PVA, WA, ER MM MMTOT PC Pindx 4 ISHR, 4 XD,
L.4 C, Ci,, CTOT G, GDTOT Z, RG Y F s Dk,
TWAGL TWNAL Tw, TWL
wage wage wage wage
of all agri sectors of all non-agri sectors of ag and non-ag sectors of all sectors and all labor cate-
Y/l YSUM HHSA V, THHSAV LSUPD,,,
gory Labor income by household and by labor category Depreciation of agricultural sectors Depreciation of non-ag sectors Agricultural capital income by household sector Non-ag capital income by household sector Total labor income by household category Total private income by household category Total private income Household savings by household group Total household savings Growth rates of labor supply by labor cate-
GDTOTUPD,
gory Growth rate sof government
YH,,, DEPRAG DEPRNA CAPY,,, CAPY,,,, YHL,
consumption
(i) Import Demand X, = A,C [6, M,+”+ (1 - Si) XXDiPi]-“+’
(1)
PJ, = PDJXD,
(2)
PM, M,
(M,JXXD,) = (PDjPM,p’
(S/l - 6Joi
(3)
PM,=PW,(l+?m,)ER+mum,(l-st)
(4)
(ii) Exporfs E, = E0 (PWE,IPE,) q
(5)
XD, = A,T [“IE,o’+ (1 - yi) XXDi(,]“+,
(6)
PX, XD; = PD, XXD, + PE, E,
(7)
(EJXD,) = (PEjPD,)+'l [(1 - yJr]+”
(8)
PWE, ER PE;= 1 + te,
(9)
(iii) Domestic supply of goods and demand for labor KC..% , XD, = A&;p” L,,&’ I+*i L4,~4,
(10)
a,, = 1 - x, Q,, and A, is a constant. PVA, = PD, - C, P, a,i - td, - pt marxi
(11)
PVA, @XD,laL,,) = WDIST,, WA, 1 = 1,2,3,4
(12)
pt = z:, no, P, tf, = rf0, MMTOT MM, = (1 - st) (marx, XDi + marm, M, PWM, ER)
(vi) Income distribution TWAG, = Za WA, WDIST,,, XLE,,, TU’NA, = Zn. WA, WDIST,,, TWAGL = XI TWAG, TWNAL = Z, TWNAl TW, = TWAG, + TWNA, TWL=C,Tw,
XLE,,,
ADJUSTMENT
YH,, = Tw, YSHARE,,
AND INCOME DISTRIBUTION
(17)
DK, PK, = ISHR, KIO, (S - 03-T)
DEPRAG = ca DEPR, PK, KS,
DKTOT = 2, DK,
DEPRNA = c,
ST, = Vi XDi
DEPR,
PK,
KS,
YAGCAP, = (& PVA, X0, - TWAG, -DEPRAG) = (C., PVA., XD.,.._- TWNA, “.-
I’D, E,
CAPYk,,
- DEPRNA)
=F
(24)
ER (1 + rei) (viii) Supply- demand equilibn’um X, = W, + &,C,, + Z, + ST, + G, + tt, i = 1,
YHLh = xi YH,,
(22)
(23)
xi PW, M, YNACAP, x CAPY,,,,
1013
., n
(25)
Equations used to update variables exogenously Ls ,.,+I= Ls,, (1 + x, ~UPD,,,)
Y, = YHL, + YAGCAPh + YNACAP, YSUM = & Y,,
KS,,,+, = KS,,, (1 - DEPR,) + DKi HHSAV, = CAPY, Y, GDTOT,+, = GDTOT, (1 + z, GDTOTUPD,) THHSAV = & HHSAV,, Objective function of dynamic model: DEPRECIA
= DEPRAG + DEPRNA
(18)
g’
(1 + r) -‘c:*/(l - E) + 7-l (1 + r) -Tc:“/(l
-E)
P ,*=xiP,Pi where (vi) Government demand G, = pi” GDTOT (1 - (I,, PC,, (1 - sub&))
(19)
c,=gcf
(vii) Investment demand S = s xi PVA, XDi + RG - pi
(20)
r = social rate of discount E = intertemporal substitution elasticity T = time horizon, after which the economy is assumed to be on its steady-state growth path.
Gi + DEPR + F*ER
RG = x:i td, PVA, XD, + tmi PM, + te,PD, E, R; = XL+ PVA, (1 - Xi ALPHLJKS, ISHR, = (R/KSi)I(xj R, KS,)
(21)
Pergamon
SO305-750X(%)000162
Adjustment and Income Distribution in an Agricultural
Economy: A General Equilibrium Analysis of Cameroon
NANCY BENJAMIN* U.S. International Trade Commission, Washington, DC, U.S.A. Summary. adjustment
To distinguish policy packages that have favorable characteristics for both structural and income distribution, this paper uses static and dynamically optimizing simulations of a
Computable General Equilibrium (CGE) model based on detailed country and household data from Cameroon. Model results indicate that public spending to reduce transport and marketing costs are beneficial for the poor and for economic growth, Copyright 0 1996 Elsevier Science Ltd
even at the expense
1. INTRODUCTION Structural adjustment refers broadly to the economic reorientation suited to coping with external shocks such as losses in terms of trade or in the availability of foreign loans. The policies used to encourage economic restructuring often include reduction and reallocation of government spending, changes in the structure of trade and excise taxes, and changes in the real exchange rate. While such adjustment plans consist of national-level policies, their effects are felt at the household level, particularly in terms of the distribution of income. Since many of the countries requiring structural adjustment are poor, observers have asked whether adjustment policies conflict or complement these countries’ poverty alleviation goals. The purpose of this paper is to analyze the implications for income distribution of alternative structural adjustment paths in a low-income, agricultural economy. We do so by performing counterfactual simulations on a general equilibrium model of a particular country, Cameroon. During the second half of the 198Os, Cameroon was subjected to three external shocks: oil prices declined, agricultural prices fell and Cameroon’s currency, the CFA Franc, being tied to the French Franc, appreciated substantially relative to the US dollar. There was no question that the coun-
try needed structural adjustment. But there are alternative tax and expenditure policies that could generate a given external balance, each with different implications for the distribution of income. In this paper, we examine these options and their implications, holding the level of current account reduction fixed, i.e., con-
of competing
public investments.
trolling for the overall level of adjustment required by the economy. In order to analyze structural adjustment and income distribution, we need a framework which can capture both the macroeconomic effects of external shocks and adjustment (i.e., shifts in the current account balance, etc.) and the microeconomic factors which determine income distribution - factor markets, product markets and household behavior. In addition, analysis of this type requires simulation of alternative paths or, in other words, a model which permits counterfactual simulatioirs. Both these criteria are met by computable general equilibrium (CGE) models which have a long history of analyzing income distribution issues (Adelman and Robinson, 1978; de Melo and Robinson, 1982; Bourguignon, Branson, and de Melo, 1992). Section 2 of this paper provides a thumbnail sketch of the data and CGE model used to analyze income distribution and adjustment in Cameroon. In section 3, we report on simulation results which reveal both the impact of external shocks as well as the implica-
*Research for this paper was undertaken as part of the Cornell University Food and Nutrition Policy Program project on Adjustment and Income Distribution in sub-Saharan Africa. I am grateful to Shantayanan Devarajan, Paul Dorosh, David Sahn and two anonymous referees for helpful comments on earlier drafts. The views expressed in the paper are the author’s own and not necessarily those of the US International Trade Commission. Final revision accepted: January 2,1996. 1003
WORLD DEVELOPMENT
1004
tions of alternative adjustment scenarios for the distribution of income among different household groups. Section 4 contains some concluding remarks.
2. DATA AND MODEL DESCRIPTION (a) Data sources The data used in the model come from a 1984-85 Social Accounting Matrix (SAM) described in Gauthier and Kyle (1990). The SAM starts with the aggregate balances given in the national accounts, adds sectoral input-output relations from the Cameroon input-output table, and disaggregates factor payments, household income and consumption according to data from the Cameroon household survey. Results of the household survey are detailed in Blandford and Lynch (1990) and Lynch (1991). The data allow for distinction in the model of seven household types. There are several agricultural households distinguished by region and income level, rural nonfarm households, and urban rich and poor. The overwhelming majority of the poor in Cameroon are rural, Most of the poor are farmers; the northern farmers are the poorest household group. Since the northern region is furthest from the capital and from major ports, northern farmers are isolated from other markets. They grow fewer cash crops and have less access to markets for food and consumer goods in other parts of the country (Blandford and Lynch, 1990).
(b) Model description The CGE model used for policy simulation is described in detail in Benjamin and Lee (1991). At this point, only certain model features are summarized for the purpose of understanding experiment results. A full set of equations is listed in the appendix. (i) Production and trade Production in each sector is characterized by Cobb-Douglas production functions with four labor types (rural and urban unskilled, skilled, and highly skilled) and sector-specific capital that can be augmented only between periods. Domestic goods are imperfect substitutes for imports in the same sector. Consumers purchase a CES aggregate of imported and domestic goods based on the relative price. Similarly, the price of exports can differ from world prices since Cameroon’s exports are considered distinguishable from world goods in the sector. Domestic producers will also respond, according to a constant elasticity transformation function that governs the rate at which domestic production can be transferred to export markets.
Each sector’s expenditures on trade and transport are accounted for in a manner like other intermediate purchases, but kept separate so that these services can be addressed by policy measures. For example, if the government spends more on road maintenance, the cost of operating and insuring transport equipment should fall, but not the cost of other services. Similarly, if the government works to improve storage facilities and reduce waste, this may reduce marketing margins but not other service inputs. (ii) Consumer demand Consumption behavior is governed by CES utility functions for each household. Consumption shares by commodity by household are taken from the household survey data. Income elasticities by good by household were also derived from the survey data. The methodology and results are in Lynch (1991). Given these component values of the utility function, minimum subsistence consumption levels are determined as a residual. The crossprice elasticities are a function of these minimum consumption levels. (iii) Macroeconomic closure Unless noted otherwise, the macroeconomic closure used in model experiments involves a fixed net flow of foreign exchange between Cameroon and the rest of the world. The numeraire of the model is the world price level or, equivalently, the price of foreign exchange. Thus the nominal exchange rate is fixed, but the domestic price level can vary against the world price level, and in this way the real exchange rate varies. The domestic price level is made up of the prices of pure nontraded goods and the prices of imperfect substitutes for world goods. The static general equilibrium simulation model solves for a single period based only on current values of variables. (No account is taken of the burdens of future debt repayment, for example.) The model can be run over several periods by incorporating each period’s investment less depreciation into the following period’s capital stock and updating total labor supply. In most cases, the results of model experiments are measured at the end of five years. All investment must be incorporated with the capital stock of the 11 sectors if it is to affect future output. The role of public investment can be seen in the flow of investment resources to public services and to the other public manufacturing sectors. Later, in an experiment, government investment will be directed toward one of the private sectors to simulate a change in public investment policy. Any decline in government tax revenue, such as from agriculture export taxes, will diminish government contributions to aggregate saving. Thus a tax cut (or public spending increase) transfers revenue from aggregate savings to private income, where only a fraction will be saved.
ADJUSTMENT
AND INCOME DISTRIBUTION
This simulation model allows us to pose the counterfactual question: Is adjustment to a negative shock better for low income groups than nonadjustment? It also allows us to compare the outcomes of different adjustment policies to determine which are best for income distribution. In addition, it allows the distinction of changes in the relative income positions of different households versus changes in absolute income levels. Given the concern for the welfare of poor households, policy makers must beware of improvements in relative income distribution that nevertheless generate unacceptable outcomes for the poor. (iv) Dynamic extension In addition to the simulation model, there is also a dynamic optimizing version of the model designed to check the growth characteristics of strategies whose distributional characteristics are tested with the simulation model. The dynamic model has the same sectoral and trade structure but lacks the detail related to income distribution. The model includes perfect foresight and maximizes the present discounted value of the utility of consumption over a 20-year period. All dynamic models face the problem of terminal conditions. To be solved, the model must define a steady state. In this case the steady state involves constant ratios of investment to capital stock and foreign debt to GDP. The 20-year period is necessary to establish a reasonably smooth growth path to this steady state. Several macro aggregates that are fixed in the static model - such as the time path of foreign capital flows, government consumption expenditures, household savings rates - are endogenous in the dynamic model. Clearly, solutions to the static model simulations are not optimal. Therefore, policy experiments tested with the simulation model cannot be replicated exactly with the optimizing model. The structure of an optimal growth path delineated by the dynamic model, however, can help in judging the growth characteristics of strategies mapped out by the simulation model. It allows us to consider optimal adjustment strategies while the simulation model can indicate their likely impact on income distribution.
3. SIMULATION RESULTS (a) Portrait of adjustment To set the stage for analysis of adjustment strategies, we first describe the effects of the most striking aspects of Cameroon’s terms-of-trade shocks using model simulations. High foreign exchange inflow during the oil boom certainly helped raise overall income levels. But one aspect of the foreign exchange windfall has strong implications for income distribution: real exchange rate appreciation.’ As mentioned earlier, spending an
1005
oil windfall domestically can bias relative prices in favor of nontraded goods and against agricultural exports. Thus, at the time negative terms-of-trade shocks hit Cameroon, poor farmers were already in a relatively disfavored position. But we can further infer that, faced with a decline in terms of trade, it is better for farmers’ relative income if Cameroon adjusts to the shock than if it increases foreign borrowing and does not adjust. Table 1 shows household income changes from a simulated drop in oil revenues. This is modeled as a capital outflow since Cameroon’s oil sector is an enclave with few production links with the rest of the economy (Benjamin, Devarajan and Weiner, 1989). In this case Cameroon’s capital outflow is increased by US$ 100 million. As expected, everyone’s income falls. But adjusting to achieve a larger current account surplus requires a decline in domestic prices against world prices, and this real depreciation helps farmers and their relative income position. Given the decline in oil income, the Cameroon government eventually had to cut back on spending. To evaluate the distributional effects of less public spending we need to consider that funds released by government cutbacks can be used in different ways. For example, it can all be used to accommodate greater capital outflow, or, at a given level of capital flows, it can all be added to domestic investment funds, or any combination of the two. To examine the two extremes, Table 2 shows first, the results of repeating the base run with a lower level of public consumption and second, a repeat of the foreign capital outflow reported in Table 1 with an accommodating decrease in public spending. In the second of these experiments, investment in each year is set at the baserun levels and public spending must adjust to the capital outflow. Of course, the results with higher capital Table 1. Differences from the base run in percentage growth overfive years
Foreign capital outflow Income for: Poor N. farmers Poor S. fanners Rich N+S farmers Nonfann rural Poor Nonfarm rural rich Urban poor Urban rich CPI
-6.9 -9.1 -8.8 -11.3 -11.2 -11.3 -11.0 -7.2
AP mfrs minus @ag AP mfrs minus AP ag in the “no-shock” case Source: Model simulations.
0.02 9.0
WORLD DEVELOPMENT
1006
Table 2. Differencesfrom the base run in percentage growth overfive years Lower public spending
Base run Income for: Poor N. fanners Poor S. farmers Rich N+S farmers Nonfarm rural poor Nonfarm rural rich urban poor Urban rich CPI
Higher capital outflow
2.0
4.5
0.9 1.3 -0.2 0.0 XI.1 0.6 a.01
-8.1 -7.4 -11.7 -11.4 -11.6 -10.4 -7.8
8.0
I.0
AP mfrs minus m ag AP mfrs minus m ag in the “no-shock” case
9.0
Source: Model simulations.
in both tables show lower income levels than the case with base level capital flows. But in checking only for the impact of reduced government spending, we see the impact is small, largely because public and private spending do not have sufficiently different effects on factor incomes. The contribution of lower public spending to real depreciation gives rise to the small distributional effects observed. outflow
(b) Alternative adjustmentpaths In evaluating adjustment strategies it helps to consider structural adjustment and income distribution as targets and that policy should aim at both targets. The targets should first be defined so that policies can be tested relative to them. Measurements of income distribution include indices based on income levels, such as poverty indices, and measurements of relative income positions comparing them to some egalitarian standard. For the purpose of this paper we consider only changes in relative incomes and record an “improvement” in income distribution when the income growth of the poorest households is high relative to that of the richest households. Moreover, since structural adjustment involves economic reorientation to absorb terms of trade shocks with minimal contraction of imports and growth, we consider the adjustment target to be foreign exchange generation or, in other words, importing capacity. There is some problem, however, with trying to hit two targets with one model. Typically, adjustment policies are compared in terms of how much foreign exchange they can “free up” or allow the economy to
do without. Such comparison requires a model closure where the international capital account is left open; each policy leads to a different level of the current account and the capital account accommodates this level. As shown above, however, different amounts of capital inflow lead to different domestic income levels. This complicates analysis of income distribution effects. While policies can be compared on their results for relative income distribution, such comparison is not always practical for the policy maker. Differences of income levels of several percentage points across adjustment plans will probably dominate any concern over differences in relative incomes. What is more likely is that policy makers have a target level for the current account or for income growth (or contraction) and are interested in the distributional consequences of alternative ways of hitting that target. For such an exercise, the model’s foreign capital account should be held at a constant level consistent with the target. In the base data, Cameroon transfers out US$162 million. This negative foreign exchange “shock” is held constant across policy experiments so that any improvement in income distribution is accomplished at a given level of capital outflow. In this way we hold the adjustment target constant and seek the greatest improvement in income distribution. Then the adjustment and growth characteristics of these policies can be examined. For example, any distribution-changing policy that succeeds in raising total income is consistent with a growth strategy, since the income expansion is achieved without reducing capital outflow and without removing any structural rigidities. Moreover, such a policy will generate forces in the same direction over a wide range of adjustment targets. Once we know which adjustment plans are also good distribution policies, we want to know which of the good distribution policies make good adjustment plans. To this end, the optimizing model can be used to check the compatibility between optimal growth strategies and those that improve income distribution. In considering policies that most affect income distribution, there is solid evidence supporting the importance of the agricultural terms of trade or the relative price of agricultural goods to manufactures (Adelman and Robinson, 1978). Since, in Cameroon, most of the poor are rural farmers, it is not surprising that the agricultural terms of trade would remain significant in experiments with the Cameroon model. But this raises the question: What most affects the agricultural terms of trade? The next set of experiments addresses this question beginning with policies intended to improve the relative price of agricultural output.
(c) Lower rnunufacturing
tariffs
Consider first a policy that could lower the price of
ADIIJSTMENT
AND INCOME DISTFUBUTION
manufactures, a significant intermediate and investment good for agriculture sectors. Since a large share of manufactures are imported, an effective instrument for reducing their cost is to cut the import tariff of 25% in half. With cheaper manufactures imports, the cost of production is lower for all other sectors. The results in Table 3 show that this effectively improves the agricultural terms of trade and the relative income of poor farmers. The decline in tariff revenue cuts income to the government, however, which saves all above a constant consumption level, and diverts these flows to others who will provide fewer savings to finance investment. While demand for manufactures is aided by the lower price, the slump in investment reduces a major source of demand for this sector. The policy leads to an overall decline in income. Such negative macroeconomic trends are not unlikely when trade liberalization is undertaken in a country with a fixed exchange rate. Overall efficiency gains are insufficient to compensate for the loss of competitiveness in one sector and the loss of tariff revenue.
1007
investment funds are again reduced from higher public expenditures on the subsidies. The agriculture subsidy produces favorable results for income distribution. There are, however, obvious problems with subsidizing domestic prices above world prices. As a practical matter it is difficult to maintain such price wedges when there are porous borders with neighboring countries which are exporting and importing the same commodities as Cameroon. In addition, since the early work with CGE models on income distribution, the biggest change has been the pressures of structural adjustment. Developing economies are increasingly required to conform to world prices for tradable goods. Under this constraint, there is often little scope either for raising the price of primary products or for lowering the price of manufactured inputs. This directs attention to improving the operation of domestic markets so that income can rise for the poor without using extensive trade taxes.
(e) Trade and transport margin subsidy
(d) Agricultural subsidy Suppose instead there is a subsidy designed to raise the producer price of agricultural goods. In this case the food and cash-crop sectors have their indirect taxes set to zero and they receive a 10% export subsidy. The subsidy is even more successful in improving the agricultural terms of trade and the relative income of farmers, while raising income overall. This time the favored sectors generate income for households with high budget shares for the sectors’ output. Therefore income expansion is possible even though
One possibility is to approach a problem identified by several observers of Cameroon: the large gap between wholesale and retail agriculture prices (World Bank, 1989). Indeed the data for Cameroon indicate that many sectors are making large purchases of trade and transport or marketing margin services. Without inducing any sudden productivity increases in trade and transport services, the model can simulate the effects of a government subsidy of marketing margins. Thus all sectors need as many services as before, but the government pays for 20% of the total package
Table 3. Differencesfrom the base run in the tetminal year (in percentage) Experiments
Income of: Poor N. farmers Poor S. farmers Rich N+S farmers Nonfarm rural poor Nonfarm rural rich Urban poor Urban rich CPI
Lower mfrs tariff
Ag subsidy
Margin subsidy (gov services)
Margin subsidy (construction)
-3.0 -4.1 4.0 -5.2 -5.1 -5.2 -5.0 4.3
5.6 2.4 2.9 -0.7 4l.9 -0.7 -1.1 2.1
5.1 1.5 2.0 -1.6 -1.1 -1.6 0.1 -1.6
7.8 2.7 3.7 -2.2 -1.3 -2.2 1.3 -2.3
2.0
3.0
1.5
AP mfrs minus APag AP mfrs minus AP ag in the “no-shock”
5.0 case
Source: Model simulations.
9.0
WORLD DEVELOPMENT
1008
of margin services. The subsidy forces government spending to increase by about 30%. If the structure of these new payments are the same as those for other publicly provided services, the impact on income distribution is as given in the column labeled Margin Subsidy (Gov Services) in Table 4. With government covering some of the cost of margin services, the private cost of delivering goods to final consumers is reduced. It is possible for producer prices to rise and consumer prices to fall. In this case the agricultural terms of trade improves as does the relative income of farmers. So reducing the burden of margin services is relatively more important for farmers whereas the decline in aggregate investment (due to higher public current expenditures) hurts them relatively less. There is an aggregation problem in that payments for marketing margins are observed for entire sectors in the model, but are actually concentrated on transported or marketed output. Likewise, the benefits of the margin subsidy affect value added throughout the sector, though in reality it will affect some producers more than others. What can be determined is that the household groups in the model that gain from the subsidy are those whose income sources are concentrated in sectors where marketing margins are high on average. In model experiments we can only suppose that government expenditures can reduce private marketing costs by some amount. The productivity of such a subsidy is a key parameter in these results. Model results indicate that if public spending could lessen
Table 4. D@erencesfrom the base run in terminal year (in percentage) Five-year run after two-year contraction Margin subsidy Margin subsidy (gov services) (construction) Income for: Poor N. farmers Poor S farmers Rich N+S farmers Nonfarm rural poor Nonfarm rural rich Urban poor Urban rich CPI
5.4 1.6 2.2 -1.7 -1.2 -1.7 0.1 -1.8
7.9 2.7 3.7 -2.3 -1.4 -2.2 1.3 -2.4
9.0
7.0
AP mfrs minus Nag AP mfrs minus AP ag in the “no-subsidy” case Source: Model simulations.
15.0
private marketing margin expenses, it would be good for income distribution. Given some possibility for success, the government may be able to reduce the effective cost of margin services to private sectors through various types of public expenditures. In the last experiment it was assumed that all new expenditures were made on public services. To test one other extreme, suppose that marketing margins can be subsidized through public expenditures on construction. As indicated earlier, current public spending on maintenance of roads and storage facilities can reduce trade and transport costs. Results from this experiment, assuming again a 20% subsidy, are listed under Margin Services (construction) in Table 4. The change in the composition of government spending accentuates the results of the earlier margin subsidy experiment. The gap between agriculture and manufactures prices is even smaller than with the direct agriculture subsidy. From the four experiment results presented so far, we can see that directing public expenditures toward reducing marketing margins can be similarly or more effective in reaching distribution and adjustment targets as direct approaches to the agricultural terms of trade. It could be argued that the role of trade and transport services - all nontraded activities -is exaggerated in the base data because 1984-85 reflects the peak absorption of oil revenues, which tends to inflate the price of nontraded goods relative to traded. After a few years of economic contraction and nontraded good price declines, the value of subsidizing margin services could be lower. This possibility is tested by repeating the margin subsidy experiment on a new base-year solution created from the original by contracting foreign capital for two years. After two years of contraction, the relative price of nontraded goods does decline. The results in Table 4 show, however, that the subsidy policies do not lose their effectiveness, even when applied to a deflated base-year solution. Clearly the size of margin services is not simply a matter of relative prices but involves the fundamental structure of the economy. These results differ from those in studies focused on increasing productivity by expanding public infrastructure (for example, Feltenstein and Morris, 1990). In such cases, the cost of producing infrastructure may be well known but its effectiveness in raising private productivity is simply imposed exogenously. With the Cameroon data, we know the size of the complete package of public and private trade and transport services purchased by each sector. The economy-wide value of reducing these costs can be determined and the expense deducted from the government account, but this does not determine the value of government spending on infrastructure. The last two modeling exercises look at the potential for investment schemes to contribute some of the effects generated by the current subsidies simulated above.
ADIUSTblENT
AND INCOME DISTRIBUTION
(f) Dynamic model simulation Reducing transport margins may be a productive way of raising the relative incomes of the poor in Cameroon, but is it also a good growth and adjustment strategy? By using the dynamic optimizing model we can check the consistency of this policy with a growth-maximizing investment strategy. The model finds the level and structure of investment that maximizes the present discounted value of the utility of consumption; thus it maximizes economic growth. Foreign borrowing and domestic saving are endogenous, but foreign debt must largely be repaid in order to meet the terminal condition on the ratio of foreign debt to GDP. The results in Table 5 compare the structure of the capital stock in the base year with the capital structure at the end of the optimal 20-year investment path. The results from this model show that to maximize growth, the share of capital in sectors producing trade and transport services must be increased (Table 5). Thus the productivity of investing in these areas is made apparent by the optimal growth path. It should be noted that the sectors with growing capital shares start with only average rates of return in the base year. Other sectors with higher returns lose capital share. Thus the optimizing model recognizes the bottleneck in services and the long-run value of expanding capacity in those sectors. While the optimizing model shows the growth advantage of investing in sectors providing marketing margin services, it does not guarantee that the income distribution results shown above will be sustained under an investment scheme as well as under direct current expenditures. If the government adds its own purchases of capital to those made by the service-supplying sectors, the added investment should bring down the cost of trade and transport services, though
1009
not as much as a direct subsidy. This approach has a more favorable outcome for total capital stock, but is less effective on income distribution compared to the current subsidy. The investment instrument is even more blunt if it brings down the price of all output to all users of the sectors producing margin services. It is possible to target margin services with an investment strategy by first, measuring the degree to which additional investment in the relevant sectors lowers their output price and then, combining such investment with a compensating tax on all uses other than marketing margins. Table 6 gives the results of this experiment where the amount of directed investment is equal to the amount spent on direct margin subsidies in the earlier experiments. The decline in price for margin services is 5% while the price for all other uses of these sectors remains very close to its base-run values. If the directed investment expenditures could be targeted toward adding only trade and transport capital stock, the impact on margin service prices would be greater.
5. POLICY CONCLUSIONS Evaluating the impact of adjustment programs on income distribution requires a general equilibrium framework for analysis and detailed data on the stmcture of the economy. For Cameroon these data show the sources of income of different households and the importance of trade and transport services for products produced by the poor. Given these two features, general equilibrium analysis demonstrates the value of lowering the cost of marketing margin services in improving the relative income of the poorest households. This finding has several implications: for adjustment strategy, for investment plans, and for exchange rate policy.
Table 5. Composition of capital stock in a dynamic, 20-year optimizing scenario Sectoral share of total capital stock in: (in percentage) Year 0 (Base) Year 20 1. Traditional food crops 2. Traditional export crops 3. Forestry 4. Agriculture modem sector 5. Food industries - private 6. Food industries - public 7. Manufacturing - private 8. Manufacturing - public 9. Construction 10. Services -private 11. Services -public Source: Model simulations.
7.5 5.9 3.6 0.9 4.9 0.6 33.0 5.3 10.7 21.0 6.3
8.0 5.1 3.9 0.8 3.4 0.4 26.9 4.2 14.5 25.4 7.3
WORLD DEVELOPMENT
1010
Table
6. Differences from the base run in percentage growth overfive years Targeted investment in margin services
Income for: Poor N. farmers Poor S. farmers Rich N+S farmers Nonfarm rural poor Nonfarm rural rich Urban poor Urban rich CPI
2.6 0.4 0.1 -2.1 -2.0 -2.0 -1.1 -0.4
AP mfrs minus mag AP mfrs minus AP ag in the “no-shock” case
7.0 9.0
Source: Model simulations.
Structural adjustment shocks place producers of traded goods under greater pressure from world prices, the producers of nontraded goods less so. The Cameroon data and model results indicate that those who profit from domestic trading routes are not the poor. Farmers gain more from having transport costs reduced than they lose in income from employment in the supplying sectors. Therefore, adjustment strategies should examine the possibility of reducing costs in sectors least pressured by world prices, but also how those sectors affect the employment of the poor and the price of goods produced by the poor.2 Public investment plans should also be considered in terms of their impact on income distribution. Model experiments show that poor households are favored by a policy where the government allocates funds away from investment into current subsidies of marketing margin services. This indicates that poor households do not gain much directly from investment expenditures.3 Nor, after five years, are the benefits of a higher capital stock sufficient to outweigh the advantage of lower margin costs for farmers. When the direct margin subsidy is replaced with capital investment to lower margin costs, the benelicial effects are diminished. Various factors combine
to lessen the effect of investment spending. One, indicated above, is the lack of employment of the poor in investment goods sectors. Another is the low capital intensity of the service sectors. This generates a low average productivity of capital and indicates that the greater efficiency gain is in lowering labor costs. Finally, model experiments only simulate the effects of adding capital to existing service technologies. They do not measure the benefits of adding new roads where none exists. The most common characteristic of poor farmers, however, is limited access to markets allowing cash sales of output and a variety of foods for purchase. This indicates the need for infrastructure investments to lessen the isolation of certain regions. Model simulations do show the value to all farmers of reducing the gap between wholesale and retail prices, even after two years of simulated macroeconomic contraction. This shows that high marketing margins arise from structural characteristics of the market, and not only from temporary conditions in relative prices. Given these features of the economic structure in Cameroon, however, the role of the exchange rate cannot be ignored. Poor farmers are engaged in producing both traded and (largely) nontraded goods, and exchange rate changes can have varying degrees of effectiveness in improving the market for cash crops. But the importance of certain nontraded services in total intermediate costs for farmers guarantees the value of real depreciation in improving income distribution in Cameroon. The general equilibrium analysis in this study enables us to simulate the distributional consequences of several public tax and spending policies and exchange rate policy. It also allows us to pinpoint those features of the economy having strong effects on income distribution. Once discovered, these features indicate fruitful areas for further investigation, such as: the market organization of transport and distribution services, the dependence of these services on public infrastructure, or the propensity of different farming households to market surplus output as transport costs fall. With such investigations we can look ahead to how structural adjustment programs may affect sensitive elements of the economy, and thus design reforms so that adjustment can proceed more smoothly.
NOTES 1. The impact of real exchange rate changes on income distribution and poverty is discussed in Kanbur (1987).
This conclusion further enhances the role of detailed analysis of distributional effects.
2. The impact of adjustment strategies on the nonpoor but powerful can affect the political feasibility of these strategies, according to de Janvry, Fargeix and Sadoulet (1991).
3. Blejer and Guerrero (1990) find similar results for the Philippines.
ADJUSTMENT
AND INCOME DISTRIBUTION
1011
REFERENCES Adelmau, I., and S. Robinson, Income Distribution Policies in Developing Countries: A Case Srudy of Korea (Stanford, CA: Stanford University Press, 1978). Benjamin, Nancy, and Yoon Lee, “Cameroon CGE models,” Mimeo (Ithaca: Cornell Food and Nutrition Policy Programs, (Cornell University, 1991). Photocopy. Benjamin, N., S. Devarajan, and R. Weiner, ‘The Dutch Disease in a developing country: Oil reserves in Cameroon,” Journal ofDevelopment Economics, Vol. 30 (January 1989), pp. 71-92. Blandford, David, and Sarah Lynch, “Structural Adjustment and the Poor in Cameroon,” Cornell Food and Nutrition Policy Program, Mimeo (Ithaca, NY: Cornell University, 1990). Blejer, M., and I. Guerrero, “The impact of macroeconomic policies on income distribution: An empirical study of the Philippines,” Review of Economics and Stafistics, Vol. LXXII, No. 3 (August 1990). Bourguignon, F., W. Branson, and J. de Melo, “Adjustment and Income Distribution: A Micro-Macro Model for Counterfactual Analysis.” Journal of Development Economics, Vol. 38, No. 1 (January 1992), pp. 1740. Devarajan, S., “Cameroon’s oil boom of 1978-85,” Mimeo (Washington, DC: The World Bank, 1991). Feltenstein, Andrew, and Stephen Morris, “Fiscal stabiliza-
APPENDIX:
tion and exchange rate instability: A theoretical approach and some policy conclusions using Mexican data,” Journal of Public Economics, Vol. 42 (August 1990). pp. 329-356. Gauthier, Madeleine, and Steven Kyle, “A Social Accounting Matrix for Cameroon,” Cornell Food and Nutrition Policy Program Working Paper No. 4 (Ithaca, NY: Cornell University, 1990). de Janvry, Alain, Andre Fargeix, and Elisabeth Sadoulet, “The political feasibility of rural poverty reduction.” Journal of Development Economics, Vol. 37 (November 1991), pp. 351-367. Kanbur, S. M. Ravi, “Measurement and alleviation of poverty,” IMFStafl Papers, Vol. 34, No. 1 (March 1987), pp. 60-85. Lynch, Sarah, “Income distribution, poverty and consumer preferences in Cameroon,” Ph.D. Dissertation (Ithaca, NY: Cornell University, 1991). de Melo, J., and S. Robinson, ‘“Trade adjustment policies and income distribution in three archetype developing economies,” Journal of Development Economics, Vol. 10 (1982), pp. 67-92. World Bank, “Cameroon: Country economic memorandum” (Washington, DC: The World Bank, 1989).
DEFINITIONS
AND EQUATIONS Export elasticity of demand Private marginal propensity to consume Minimum consumption level Government consumption share Export duty rates Tariff rates Indict tax rate Depreciation rates Proportion of the average wage by labor category and by sector Shares of investment by sector of destination Ratios of inventory investment to gross out-
(a) Indices
iJ if im 1 h a na agcY nacy r
sectors exportable sectors importable sectors Labor categories Household categories Agricultural sectors nonagricultnral sectors Agricultural capital income categories Nonagricultural capital income categories Time period
(b) Parameters maq maf? PC St
sub,
Marketing
margin as a rate of domestic out-
put Marketing margin as a rate of imports Private consumption share of public goods Subsidy rate on marketing margin Subsidy rates on private consumption of public goods CES utility (Armington) function shift parameter CES utility (Armington) function share parameter CES utility (Armington) function exponent Elasticity of substitution CET function shift parameter CET function share parameter CET function exponent Elasticity of transfonnation
a,, h I.I
tt;
PC, sub, -58
P,I
put Input-output coefficient Capital composition coefficient Share of marketing margin expenditures Aggregate private savings rate Private consumption shares of public goods by household group Subsidy rates on private consumption of public goods by household group Private savings rate by household category Private consumption shares by sector and by household group
(c) Variables
p< PQ PE, PWE,
Price of composite good i Price of domestic sales Domestic price of exports World market price of exports
WORLD DEVELOPMENT
1012
(d) Equations in the model
DKTOT DEPRECIA CTOT,
Domestic price of imports Domestic supply price World market price of imports World market price of exports Value added price Average wage rate by labor category Exchange rate Value of marketing margins Total value of marketing margins Price of trade margins Aggregate price index Return to capital Sector share of investible funds Composite goods supply Domestic output Domestic sales of domestic output Imports Exports Sector specific capital Inventory investment demand Employment by labor category and by sector Labor supply Private consumption demand for good i Consumption demand for good i by household group Total private consumption demand Public consumption demand for good i Total public consumption demand Investment demand by sector of origin Government revenue Total income of private sector Foreign savings Total savings Volume of investment by sector of destination Total investment Total depreciation charges Private consumption demand by household group Income by household category Total wage of agri sectors by labor category Total wage of non-agri sectors by labor cate-
MMTOT = C, MM,
Yh TWAG, TWNA,
c, L,, = I&
(13)
gory Total Total Total Total
(iv) Intermediate demand W, = Z, a, XD,
(14)
(v) Consumer demand C,, = 7,* + pih (CTOTIP,) + PC,, pi” GDTOT ( 1 - sub,,)
(15)
CTOT, = (1 - s,,) Yh
(lo)
PM px, PMW, PWE, PVA, WA, ER MM MMTOT PC Pindx 4 ISHR, 4 XD,
L.4 C, Ci,, CTOT G, GDTOT Z, RG Y F s Dk,
TWAGL TWNAL Tw, TWL
wage wage wage wage
of all agri sectors of all non-agri sectors of ag and non-ag sectors of all sectors and all labor cate-
Y/l YSUM HHSA V, THHSAV LSUPD,,,
gory Labor income by household and by labor category Depreciation of agricultural sectors Depreciation of non-ag sectors Agricultural capital income by household sector Non-ag capital income by household sector Total labor income by household category Total private income by household category Total private income Household savings by household group Total household savings Growth rates of labor supply by labor cate-
GDTOTUPD,
gory Growth rate sof government
YH,,, DEPRAG DEPRNA CAPY,,, CAPY,,,, YHL,
consumption
(i) Import Demand X, = A,C [6, M,+”+ (1 - Si) XXDiPi]-“+’
(1)
PJ, = PDJXD,
(2)
PM, M,
(M,JXXD,) = (PDjPM,p’
(S/l - 6Joi
(3)
PM,=PW,(l+?m,)ER+mum,(l-st)
(4)
(ii) Exporfs E, = E0 (PWE,IPE,) q
(5)
XD, = A,T [“IE,o’+ (1 - yi) XXDi(,]“+,
(6)
PX, XD; = PD, XXD, + PE, E,
(7)
(EJXD,) = (PEjPD,)+'l [(1 - yJr]+”
(8)
PWE, ER PE;= 1 + te,
(9)
(iii) Domestic supply of goods and demand for labor KC..% , XD, = A&;p” L,,&’ I+*i L4,~4,
(10)
a,, = 1 - x, Q,, and A, is a constant. PVA, = PD, - C, P, a,i - td, - pt marxi
(11)
PVA, @XD,laL,,) = WDIST,, WA, 1 = 1,2,3,4
(12)
pt = z:, no, P, tf, = rf0, MMTOT MM, = (1 - st) (marx, XDi + marm, M, PWM, ER)
(vi) Income distribution TWAG, = Za WA, WDIST,,, XLE,,, TU’NA, = Zn. WA, WDIST,,, TWAGL = XI TWAG, TWNAL = Z, TWNAl TW, = TWAG, + TWNA, TWL=C,Tw,
XLE,,,
ADJUSTMENT
YH,, = Tw, YSHARE,,
AND INCOME DISTRIBUTION
(17)
DK, PK, = ISHR, KIO, (S - 03-T)
DEPRAG = ca DEPR, PK, KS,
DKTOT = 2, DK,
DEPRNA = c,
ST, = Vi XDi
DEPR,
PK,
KS,
YAGCAP, = (& PVA, X0, - TWAG, -DEPRAG) = (C., PVA., XD.,.._- TWNA, “.-
I’D, E,
CAPYk,,
- DEPRNA)
=F
(24)
ER (1 + rei) (viii) Supply- demand equilibn’um X, = W, + &,C,, + Z, + ST, + G, + tt, i = 1,
YHLh = xi YH,,
(22)
(23)
xi PW, M, YNACAP, x CAPY,,,,
1013
., n
(25)
Equations used to update variables exogenously Ls ,.,+I= Ls,, (1 + x, ~UPD,,,)
Y, = YHL, + YAGCAPh + YNACAP, YSUM = & Y,,
KS,,,+, = KS,,, (1 - DEPR,) + DKi HHSAV, = CAPY, Y, GDTOT,+, = GDTOT, (1 + z, GDTOTUPD,) THHSAV = & HHSAV,, Objective function of dynamic model: DEPRECIA
= DEPRAG + DEPRNA
(18)
g’
(1 + r) -‘c:*/(l - E) + 7-l (1 + r) -Tc:“/(l
-E)
P ,*=xiP,Pi where (vi) Government demand G, = pi” GDTOT (1 - (I,, PC,, (1 - sub&))
(19)
c,=gcf
(vii) Investment demand S = s xi PVA, XDi + RG - pi
(20)
r = social rate of discount E = intertemporal substitution elasticity T = time horizon, after which the economy is assumed to be on its steady-state growth path.
Gi + DEPR + F*ER
RG = x:i td, PVA, XD, + tmi PM, + te,PD, E, R; = XL+ PVA, (1 - Xi ALPHLJKS, ISHR, = (R/KSi)I(xj R, KS,)
(21)