Apn4S ijuea plioA v DDL1 r U} °g S31d° tmu o00r a0400 sos uausnwTpv U Making Adjustment Work for the Poor A Framework for Policy Reform in Africa The World Bank Washington, D.C. Copyright i 1990 The International Bank for Reconstruction and Development/The World Bank 1818 H Street, N.W. Washington, D.C. 20433, U.S.A. All rights reserved Manufactured in the United States of America First printing September 1990 This is a study by the staff of the SDA Unit at the World Bank, and the judgments in it do not necessarily reflect the views of the Board of Executive Directors or the governments they represent. The material in this publication is copyrighted. Requests for permission to reproduce portions of it should be sent to Director, Publications Department, at the address shown in the copyright notice above. The World Bank encourages dissemination of its work and will normally give permis- sion promptly and, when the reproduction is for noncommercial purposes, with- out asking a fee. Permission to photocopy portions for classroom use is not required, though notification of such use having been made will be appreciated. The complete backlist of publications from the World Bank is shown in the annual Index of Publications, which contains an alphabetical title list and indexes of subjects, authors, and countries and regions. The latest edition is available free of charge from the Publications Sales Unit, Department F, The World Bank, 1818 H Street, N.W., Washington, D.C. 20433, US.A., or from Publications, The World Bank, 66, avenue d'Iena, 75116 Paris, France. Library of Congress Cataloging-in-Publication Data Making adjustment work for the poor: a framework for policy reform in Africa. P. cm. "A World Bank study." Includes bibliographical references and index. ISBN 0-8213-1640-0 1. Africa-Economic policy. 2. Income distribution-Africa. 3. Poor-Africa. I. World Bank. HC800.A55256 1990 362.5'096-dc2O 90-45406 CIP This volume has been prepared by Tony Addison, Lionel Demery, Marco Ferroni, and Jeffery Round, under the overall guidance of Christiaan Grootaert, Tim Marchant, and Michel Noel. It has benefited greatly from comments received at several workshops and seminars on the conceptual framework of the Social Dimensions of Adjustment Pro- gram involving donors, policymakers, and researchers, including those held at Paris (February 1989), Arusha (March 1989), and Dakar (March 1989). The helpful com- ments on early drafts received from ILO, UNICEF, IFAD, UNDP, UN Statistical Office, UN office at Vienna, an Expert Panel set up by the Norwegian Ministry of Developmnent Cooperation, ODA, and SIDA are gratefully acknowledged. The vol- ume has also benefited from conmments received from Ramesh Chander, Francis Colaco, Jean-Marie Cour, Lual Deng, Jean-Luc Dubois, Nancy Gillespie, Ravi Kanbur, Steve O'Brien, Alexander Shakow, Antoine Simonpietri, Mahfouz Tadros, Bjorn Wold, and other Bank staff. The technical support of IFAD is especially acknowledged. An in-kind contribution by IFAD to the SDA program, consisting of a conceptual paper focusing on smallholders, was an important input into this work. Foreword This document is produced as part of the work in the integration of social and economic policies undertaken under the Social Dimensions of Ad- and programs. justment (SDA) Program in Africa. This initia- This document deals with this last track and tive is cosponsored by the United Nations Devel- represents the state of present thinking on these opment Programme (UNOP), the African Devel- issues, as well as contributions by the authors. opment Bank, and the World Bank, and is also The broad theoretical underpinnings of the SDA supported by several bilateral and multilateral policy approach are explained. This sets the aid agencies, with the World Bank as executing boundaries for investigative studies on the social agency. To date, thirty-two governments have dimensions of adjustment and provides the nec- asked to participate in this program. essary economic rationale for empirical work to The program aims at thoroughly integrating be both consistent across countries and helpful social dimensions into economic and financial for policy design. The document then applies decisionmaking. To do so, it pursues several this conceptual framework in a real-world set- parallel tracks: ting by outlining an empirical framework for data * first, within each country, a special institution analysis. The objective of this presentation is to building effort is being launched to strengthen provide guidance to researchers and practitio- national capability for policy analysis and project ners in assembling and analyzing the necessary identification, preparation, and implementation. data in order to achieve the objectives of the SDA This includes coordination mechanisms, training initiative. The document then explores the major of nationals, and some limited foreign technical policy issues that must be faced by governments assistance to integrate social dimensions in the design of * second, country-specific studies dealing with their structural adjustment programs and devel- poverty alleviation and economic policy arebeing opment plans. It should be perceived as the start undertaken. These would serve as an inventory of of an ongoing examination of these issues, an existing knowledge, a baseline for monitoring examination of great urgency in Africa, but of future change, and a foundation for proposals for truly universal interest. specific actions It is important for readers to know, however, * third, developing an adequate data base for that in parallel to pursuing these questions, the analysis of social issues and formulating so- SDA initiative also supports governments in the cioeconomic policy through monitoring surveys formulation of social action programs to respond and integrated household surveys to the pressing needs they face. These social * fourth, and perhaps most important, by im- action programs are based on inadequate data proving our understanding of the links among bases in many cases. They represent, however, conceptual, empirical, and policy issues involved the best judgments that can be brought to bear on v these issues in the given country and are in no integrated. They are managed in the countries way a substitute for the parallel in-depth work and at the Bank by the same teams, and the syn- being done at the country level or at the concep- ergies will undoubtedly become more apparent tual level. over time. With such efforts, the thorough inte- Ultimately, as these various streams (institu- gration of the social dimensions into economic tion building, conceptual thinking, data base de- and financial policies will gradually become a velopment, and social action programs) are being reality. We hope that this document will prove a implemented, they will need to become ever more useful contribution on this long and arduous road. IsmaYl Serageldin Director Technical Department Africa Region vi Table of contents Foreword v Executive summary 1 1 Introduction 7 The meaning of adjustment 7 The policy problem 8 An outline of the book 9 2 Macroeconomic disequilibrium 10 The causes of macroeconomic disequilibrium 10 Policy sequences in Africa 11 An analytical framework 12 3 Adjustment 20 Adjustment policies in Africa and their macro effects 20 An orthodox framework for analyzing the meso effects 22 An alternative framework: adjustment with unemployment 30 The growth factor 33 4 The determinants of welfare 36 Individual welfare 36 The economy of the household 38 Analyzing household behavior 40 The distribution of welfare within households 43 Household expenditures 46 Interactions between households 47 The meaning and measurement of poverty 49 v. 5 The effects of destabilization and adjustment on households 51 Household incomes under macroeconomic disequilibrium 51 A framework for considering the household effects of adjustment 54 Other aspects of household welfare effects 57 6 Data and analysis implications 59 Structural adjustment, income distribution, and data analysis 59 Statistical initiatives in Africa 60 A hierarchical information system 62 Information requirements at the macro level 65 Information requirements at the meso level 70 Information requirements at the micro level 74 Towards macro-meso-rnicro analysis 80 Macroeconomic modeling 81 Macro-meso modeling 82 Meso-micro analysis 85 7 Strategic policy issues 87 Social interventions and economic distortions 87 Implications of the conceptual framework 89 Poverty and vulnerability 90 Defining target groups for policy purposes 90 8 Designing poverty-sensitive adjustment programs 93 Overall fiscal and monetary balance 93 The public finance strategy 97 Monetary and financial policy 100 Exchange rate policies 103 Trade policies 105 Policies toward self-employment 109 Policies toward wage employment 114 Concluding perspective 120 Appendix: The social accounting matrix 122 Endnotes 128 Bibliography 133 viii Executive summary Adjustment is best considered as the implementa- tion, and institutional reforms. Trade policies, tion of comprehensive reforms of macro and mi- public enterprise and agricultural policies account cro policies, in response to various shocks, and to for about two thirds of the total number of policy rectify inappropriate past policies that have conditions set under adjustment lending in 15 hampered economic performance. These shocks African countries recently reviewed by the World have adversely affected the whole range of eco- Bank. However, while 70 percent of exchange rate nomic policy objectives, including the balance of policy conditions were fully met, only 57 percent payments, price stability, full employment, eco- of agricultural policy conditions and 55 percent of nomic growth, the protection of the environment, trade policies were met in practice. Policies di- and equity/poverty reduction. Shocks, whether rected at changing relative prices tended to be internal or external, affect all policy objectives, more successfully implemented than those re- and not simply the balance of payments, price quiring institutional change. stability, and growth which are the traditional In considering the social effects of adjustment, it focus. We are here concerned with how adjust- is necessary to establish analytically the links be- ment programs have affected the wider social tween the macro economy on the one hand and objectives of governments, and especially poverty the micro economy of households and enterprises reduction. on the other. This link between the macro and the While policy responses to shocks have been micro, termed the "meso" (or middle), comprises: varied and complex, there are basically three markets (both product and factor, formal and courses of action that governments have taken: informal); and infrastructure, this consisting of finance theexternal currentaccountdeficit through economic infrastructure (roads, communications, capital inflows; reduce the deficit to the level of irrigation) and social infrastructure (health and available external capital through tightening capi- education services). The objective is to establish tal and trade restrictions; or implement policy how the macroeconomic processes initiated un- adjustments to restore a viable balance of pay- der adjustrnent programs affecthouseholds, which ments position through export expansion and requires an empirical understanding of mesoeco- import substitution. Most adjustment programs nomic linkages. in Africa consist in varying degrees of a currency The standard international trade model, based devaluation, a reduction in the government defi- on the Meade-Salter-Swan model is a useful start- cit in order to bring expenditures into line with ing point for analyzing these issues. Consider a real resources, a liberalization of domestic prod- small, open economy which is a price taker in uct markets, bringing domestic prices into line international product markets. If the domestic with world prices, trade and exchange liberaliza- relativepriceofexportablestoimportables(PX/P.) 1 is fixed, they can be combined into a composite policy are modified significantly. If wages in the commodity - tradables. The domestic price of nontradables sector are subject to nonmarket de- tradables (P) will be determined by the prevailing termination (as for example when nontradables world price and the exchange rate. Since comprise mainly public sector services), workers nontradablesarebydefinitionnottradedinworld there may experience a cut in their real wage, and markets, their price (P.) is determined entirely by the tradable/nontradable (informal/formal) wage domestic supply and demand. gap will narrow. If adjustment leads only to a Four major effects of an expansionary monetary reduction in aggregate economic activity, how- and fiscal policy can be derived. First, it will lead ever, the wage gap may widen. Finally, the to a current account balance of payments deficit, infrastructural effects of adjustment are likely to since theincreased demand generated for tradables be critical in Africa. The evidence suggests that will raise imports, and direct exports to the home price responsiveness in the region is significantly market. Second, the excess demand for nontrad- slower than elsewhere in the developing world, ables will raise P., so that P./Pt will rise, inducing and this is undoubtedly due to the deteriorationof resource transfers into nontradables. Thirdly, the economic and social infrastructure. factor prices will change in response to these There are a number of ways, therefore, in which sectoral shifts: in the short run, real wages will fall the household economy, and the welfare of its if workers consume sufficient nontradables; in the members, can potentially be favorably or unfa- long run, they will fall if the tradables sector is vorably affected by adjustment policies and the relatively labor intensive. Finally, increased relatedmesoevents. Oneadverseeventcansetoff government expenditures may affect the economic a whole chain of unfavorable events. For example and social infrastructure, depending on the nature losing a job, in leading to loss of income, may of the fiscal expansion. Macroeconomic result in asset sales in order to protect consump- disequilibrium has affected all the main elements tion, resulting in a fall in self-employment. If of the meso economy. If a terms of trade decline is consumption levels fall, nutrition and health sta- superimposed on a situation of general fiscal ex- tus can deteriorate, labor productivity will fall as pansion, the effects on the meso economy noted a consequence, and further damage to employ- above are modified. Notably, with P./P. falling, ment prospects will occur. But the reverse of this resource transfers will tend tobe out of exportables is that a suitable policy intervention can set in and into importables and nontradables. Moreover, motion favorable effects for household welfare. if governments raise Pm further through tariffs or These vicious and virtuous circles suggest that import controls, resources will be drawn from households can move from better- to worse-off exportables and nonprotected importables into states in relatively short periods, and that some nontradables and protected importables. policy interventions can be effective in setting In a simnilar fashion, the model can be used to households on upward "income escalators." show the meso effects of adjustment policy. First, However, welfare outcomes depend on the there are effects on product markets. A combina- household's behavioral response to the external tion of exchange rate adjustment and fiscal/mon- shocks. In order to make a meaningful assessment etary contraction will initiate fundamental relative of how shocks and adjustment programs affect price movements in favor of tradables, causing welfare, models are needed which explain the resource switching into that sector. There may decisionmaking process of the household, and also be changes in relative prices within the thus predict its response to variations in the pa- tradables category. Second, the adjustment-policy rameters of its economic environment. These package will affect labor markets, the outcome models should capture African households' dual depending on the time perspective taken, and on roles asboth unitsof consumptionand production, the structure of the market. In a competitive labor and the interdependence of the household's pro- market the movement in the real wage depends duction and consumption decisions. Such house- onworkers'consumptionpropensitiesintheshort hold models should also explicitly recognize so- run, and on relative factor intensities in the long cially determined constraints on the activities of run. The theory is strictly agnostic about how real women. These create allocative inefficiencies, wages will change - it is simply an empirical since the labor resources of the household are not matter. If the labor market (or a section of it) is not allocated in accordance with its members' mar- competitive, the real-wage effects of a switching ginal productivities. 2 Households will be affected as buyers and sell- Markets are not generally observable as physical ersinproductmarkets,ashiring in (or out) services entities and therefore information about how they from labor markets, and by the economic and function must be derived indirectly. There is the social infrastructural changes that are induced important additional problem of identifying the through destabilization and adjustment. How existence and operation of parallel markets. By precisely poor households are affected obviously their very nature, information on these might be depends on two key sets of factors: the nature of difficult to obtain and, of course, one might expect the mesoeconomic changes that are induced; and adjustment to reduce parallel market activities the characteristics of the household - whether it overall. Information at the micro level is almost is a net buyer or seller of tradable goods, which entirely derived from the household surveys. markets (official/parallel) it trades in, whether it While the ultimate interest is in individual well- hires in labor services, or depends on its own labor being, for survey purposes the household is to be in generating primary income, whether it has chosen as the sampling unit, in preference to access to schools and health services, and whether possible alternative social units. it is well served by roads and communications. The hierarchical information system is intended The role and importance of a national informa- to feed into a policy-orientated analytical frame- tion system becomes critical during the adjust- work that must be capable of (i) performing ment process. The policymaker requires relevant "counterfactual" experiments by tracing what and timely data in order to select the composition might have occurred had an alternative set of and sequencing of adjustment measures. Infor- policies been applied; and (ii) tracing through mation is required at each of the macro, meso and these effects to households. Since most models micro levels if policymakers are to know what which satisfy (i) are limited in the degree of disag- changes are affecting households over time. It is gregation that they can reasonably accommodate, important to view information requirements in they need to be supplemented by data analyses the context of a hierarchical information system, that satisfy (ii). Thus, the analytical strategy for where data and empirical analysis are explicitly techniques to investigate macro-meso linkages, considered at each of the three levels. The hier- through which counterfactual type experiments archical information system is a way of systemati- would be feasible, together with a survey-based cally viewing the body of data currently available, meso-micro analysis of household welfare. identifying gaps and deficiencies, and their rela- The key policy problem is how to assist target tionship to potential analytical tools and con- groups without at the same time causing distor- structs. The system must be capable of measuring tions in economic mechanisms. If the latter are the main national aggregates (such as national severe, neither economic recovery nor the poor product and income, the balance of payments and will be helped. The foundations of sustainable the various components of aggregate expendi- growth will never be secured if Africa's human tures) and subaggregates (such as sectoral output capital is allowed to "depreciate". Adjustment and sectoral allocations of productive resources). transmits its effects to target groups through the It must also be capable of tracing how the produc- meso economy: the fabric of markets and eco- tive activities which generate these aggregates, nomic infrastructure which links households to allocate the resulting incomes to the various fac- the wider economy. This highlights how the tors of production and on to the households which deterioration of Africa's meso economy has deter- possess them. In this way, the system should be mined the distribution of the costs and benefits of able to trace the process through which income adjustment policies. That there are gainers and flows are allocated to household groups. losers from policy reforms is an inescapable part Information requirements at the macro level of the adjustmnent process in its early stages, since include the basic macroeconomic, social and de- the structure of output and expenditures under- mographic data necessary to service the goes substantial change. But who gains, and who macroeconomicaccountsandtoderivecoremacro loses, is critical. indicators. The macroeconomic accounts consist Policy interventions concerned with the social of the national accounts, public finance, balance of dimensions of adjustment must recognize the twin payments and monetary data, all of which are problems of poverty and vulnerability. The former fundamental to the analysis of adjustment. Data concems the level of income (or expenditure) of a relating to meso-level activities are also critical. household in relation to some standard, while the 3 latter relates to the influence of external shocks on must be determined. These consist of the pro- the household and its ability to cope with them in grams which are most costeffective in reaching avoiding serious welfare loss. This suggests three the poor, and whose real levels of expenditure main groups for policy concern. The chronic poor must be maintained or raised. At the stage of whose situation is caused by multiple depriva- planning and priority setting, all expenditure cat- tions, such as low productivity due to poor health egories should be viewed as adjustable in order to and nutrition, poor access to productive assets, make room for core items. Two broad categories etc. This poverty is deep-rooted, existing before of core expenditure can be identified: economic the recent deteriorationin economic circumstances expenditures which raise the income-earning po- and the implementation of adjustment programs. tential of poor groups; and social expenditures The new poor who are above the poverty line prior which enhance their human capital. Economic to the shock and adjustment measures, but have expenditures directed at raising the productivity fallen into poverty as a result. And other vulner- of the poor include agricultural research and ex- able groups who remain above the poverty line but tension services, small irrigation schemes, feeder are severely affected by adjustment and therefore roads, credit and marketing services, and so on. merit policy consideration. Using the criteria of The rehabilitation of primary health and pri- poverty and vulnerability permits a further re- mary education is critical to protecting social ex- finement in the classification of households - penditures. An emphasis on rural and urban going frombroader socioeconomic groups to more clinics serving low-income groups, basic drugs, specific target groups. These are defined as those immunization, the screening of children and households whose poverty and/or vulnerability mothers for nutrition and health programs, and call for policy attention. the provision of safe water and improved sanita- The first step in ensuring that adjustment con- tion, are all health priorities. Likewise, the prior- tributes to social objectives is to orientate pro- ity in education is to renew the growth of primary grams towards supply-expansion through appro- school enrolments among poor children, and to priate policies and external finance. The strategy improve the quality of education by securing sup- towards overall fiscal and monetary balance is plies of basic educational materials. Replacing critical to this. Macro imbalance implies that do- cost-ineffective food subsidies by targeted food- mestic demand exceeds domestic supply, either interventions is complementary to the task of because of supply shocks, or because of reorientating social provisions. Combined nutri- overexpansionary domestic demand policies. If tion and health programs are one means to better uncorrected, this situation leads to a loss of for- targeting, and measures of selective subsidization eign exchange reserves and creditworthiness. Ad- should be explored. justment cannot therefore be postponed. One Monetary targets are one of the key instruments critical element of adjustment is how much room for attaining macro balance. While their primary for maneuver governments have in maintaining objective must be a reduction in inflation and the their fiscal deficits. They can finance deficits external deficit, there are opportunities in their either through printing money ('seignorage'), or application to favor borrowers among the poor. borrowing (from at home or abroad). Each of The application of credit ceilings on the public these has its own constraints. Given the shallow- sector in favor of credit to the private sector may ness of domestic financial markets, most African in itself push more funds towards the poor. Within governments have limited opportunity to borrow overall credit ceilings priority can be given to domestically. Given the inflationary effects of loans to smallholder agriculture. There is much seignorage, and the limits to foreign borrowing, evidence that poor producers can use more credit they generally have to reduce fiscal deficits dur- profitably, but that they are disadvantaged by ing adjustment, simply to restore macro balances. government interventions in credit markets which This obviously reduces their room for maneuver favor large borrowers, and by a lack of collateral to protect various components of expenditure which limits their access to banks. The former which are important for social policy. problem is reduced by financial liberalization This suggests that the composition of the fiscal which raises incentives to lend to smaller borrow- adjustment is critical for social policy objectives, ers. Credit to the poor is, in most cases, best and priorities need to be established with care. improvedbygroup-lendingschemes,whichhave First, a set of core expenditures benefiting the poor a high success rate, especially in reaching women. 4 Currencydevaluation is anothermajorinstrument The shift inoutput towardsexports and efficient of macro policy. It will affect poor households as import substitutes should promote a more labor- producers and consumers of tradables and intensive pattern of production than in the past. nontradables, and as wage earners. Many poor The employment gains from this (in terms of households in Africa are engaged in producing raised real wages and higher levels of employ- tradables, so there is every reason to believe that ment) are likely to be reaped over the medium- to they will benefit from a devaluation. It may be, long-term. In the short run, employment pros- however, that the price signals from a devaluation pects may not be so beneficial. It is likely that real never reach the household producers because of wages will fall, especially in the formal sector, and the imperfect functioning of product markets. If there may be increases in unemployment, mainly the market is monopsonistic, traders may not pass of a frictional nature, but also arising from factor on price increases to the farm gate. To enhance the market imperfections. Policies to ease these short- benefits of a devaluation to such households, run adverse effects are of two broad types: those complementary meso-level policies may be re- which improve the operation of labor markets, quired in order to improve the functioning of through, for example, raising the mobility of la- markets and to ensure that poor farmers benefit. bor, providing training and reemployment The improvement of the economic infrastructure schemes; and those which provide emergency will also enhance any beneficial effects of de- employment, frequently in public works. valuation in providing better access to favored There are at least two major problems in decid- product markets. If poor households produce ing on a conceptual framework for analyzing the nontradables (and especially if they use imported social effects of adjustment. First, the policy re- intermediate inputs such as fertilizers), they will search problem is itself a major undertaking. The be adversely affected by devaluation. The key to analytical challenges that are faced in establishing raising their incomes lies in creating alternative how macro- and sector-level policies affect house- production patterns which may require comple- holds and the well-being of individuals are seri- mentary policies, such as targeted extension ser- ous and should not be underestimated. One of the vices that aim to encourage farmers to adjust major outcomes of the theory is that there are no production, or credit services that may be essen- ready answers - the problem is highly complex tial if farmers are to move into the market-ori- and can only be resolved in the last analysis at the ented production of tradables. Through raising empirical level. And, of course, the empirical official food prices, devaluation can benefit the problems that are raised are just as challenging. rural poor but tends to hurt the urban poor, be- Thesecondproblemarisesfromtheheterogene- cause of the differences in their employment and ityof the Sub-Saharan African region to which this expenditure patterns. It is crucial that a macro framework must apply. It must at the same time policy (such as devaluation) which raises food be general enough to apply to all the diverse prices is coordinated with targeted food subsidies economic and social circumstances of the region, in order to dampen the effect on poor groups. and specific enough to be of use in guiding the Liberalizing import restrictions improves effi- policy and investigative initiatives at the country ciency and reduces general poverty through tilt- level. Because of this, we have persistently issued ing incentives back in favor of food and export warnings that the analytical framework is not agriculture, and by promoting labor-intensive in- meant to be strictlyapplied in each and every case. dustrialization. Since liberalization changes rela- For some cases, the assumptions we have made tive prices, similar considerations apply in tracing are a reasonable approximation, but for others, its poverty implications to those discussed for there may be a need for further refinement. Our devaluation. It affects the poor as producers, purpose is to establish a structured way of think- consumers and wage earners, and the net effect ing about the problem. At the heart of this struc- depends on the economic activities of poor house- ture is the simple device of dividing the research holds. Similarly, complementary policies (such as task into two stages: first identifying macro-meso improving the functioning of product and credit interactions and then dealing with meso-micro markets, rural economic infrastructure, targeted effects. extension services, etc.) can enhance favorable The most important conclusion that the theory effects for the poor, and avoid any adverse reper- offers (apart, that is, from indicating that the sub- cussions. ject is inherently difficult, and that the theory is 5 inevitably inconclusive) is that adjustment poli- the theory is of little use in analyzing the myriad cies can set in motion changes in the economy, detailed changes that adjustment programs inevi- which have profound and pervasive effects on tably involve, it is at its most useful in uncovering markets, infrastructure and households. Attempts these more fundamental economywide effects. to raise household incomes which ignore (or even Without it, the policy maker would not be able to run counter) to these deep-seated changes, are see the wood for the trees. both misplaced and counterproductive. While 6 1 Introduction There are at least two important reasons for focus- tected, sustainable economic growth in many Af- ing on the social dimensions of adjustment in rican countries will itself be threatened. Africa at the present time. First, it is now some sixteen years since the first oil shock of the 1970s, The meaning of adjustment and since then the developing countries of the region have experienced economic turbulence, We begin by clarifying what is meant by adjust- shocks, and disequilibrium. After the commodity ment and adjustment policy in developing coun- price boom of the mid 1970s, most countries faced tries, a subject on which the literature is not always a persistent decline in their terms of trade, caused in agreement. The terms "adjustment," "structural in part by a general Organisation of Economic Co- adjustment," and "stabilization" are common- operation and Development (OECD) recession. place. Yet the use of these terms is often imprecise Interest rate hikes have compounded their debt and inconsistent. Adjustment is best considered problems, while internal shocks (notably the sus- as the implementation of comprehensive adjust- tained drought of the early 1980s) have consider- ments in macro and micro policies, both to re- ably weakened their capacity to deal with macro spond to various shocks and to rectify inappro- imbalances. In all of this, the poorer groups in priate policies that have hampered economic African societies have suffered; with little margin performance, the fundamental objective of which above subsistence, many of them have found little is to restore sustainable economic growth. The room to maneuver. shocks include both the environmental, such as Second, with a return to a longer-term policy adverse weather, and the external, for example, perspective, greater emphasis is now being given terms of trade declines. These shocks have ad- to restoring sustainable growth in these countries. versely affected the whole range of economic Given this perspective, it is now both feasible policy objectives, including the balance of pay- and essential to give more careful thought to the ments, price stability, full employment, economic social dimensions of adjustment - to consider growth, the protection of the environment, and how poorer groups in society can be encouraged equity/poverty reduction. Shocks, whether in- to participate in the recovery; to examine how the ternal or external, affect all policy objectives, and health and education of the societies are likely to not simply the balance of payments and price be affected; to investigate the effects on food se- stability, which is the traditional focus of con- curity and nutrition - in short, to trace how cern. We are concerned here with how the ad- structural adjustment policies will ultimately af- justment programs have affected the objective of fect the well-being of the populations involved. poverty alleviation and the wider social objectives The social dimension, however, is not only im- of governments. portant for equity. Unless human capital is pro- A useful summary of the "language of adjust- 7 ment" is to be found in a World Bank report erty effects as the economy assumes its new (1988a, p. 23), which distinguishes among the growth path. The challenge for the policymakers following: is to identify the set of adjustment policies that * stabilization: policies (generally relying on will induce the participation of the poor in the demand management) to achieve sustainable fis- process of recovery and growth. This essentially cal and balance of payments current account implies that an ex ante approach is to be pre- deficits and to reduce the rate of price inflation ferred, whereby policy interventions that enhance * structural adjustment: reforms of policies and social dimensions are identified as much as pos- institutions covering microeconomic (such as sible before a structural adjustment program is taxes and tariffs), macroeconomic (fiscal policy), applied - that is, at the policydesign stage. and institutional interventions; these changes are Adjustment programs are conceived primarily designed to improve resource allocation, increase in terms of theireconomywide effects-to achieve economic efficiency, expand growth potential, and macroeconomic targets. Many of the policy in- increase resilience to shocks struments have an essentially macroeconomic * adjustment: policies to achieve internal and profile-exchangerate policy, fiscaland monetary external balance and changes in the structure of policy, producer prices, and so on. The principal incentives and institutions, or both; where the problem is to establish the effects of these policies emphasis is on the former, it can be identified as on households, which interact in varying degrees stabilization, where on the latter, as structural with the economy at large. As a basis for policy adjustment appraisal, the SDA program must establish ana- * structural adjustment lending: World Bank lytically the links between the macro economy on lending that supports structural adjustment; this the one hand and the micro economy of house- generally provides import financing and is rela- holds and enterprises on the other. This link be- tively quick disbursing; it is possible that some tween the macro and the micro, termed the elements of a structural adjustment lending will ''meso,"1 comprises the following key elements: have stabilizing effects (for example, reducing * Markets, and other resource allocative the fiscal deficit) mechanisms, are the main links between the a sector adjustment lending: World Bank lending macro economy and individual households. These in a sector, focusing on institutional and mi- include product and factor markets, both official croeconomic distortions; this is also usually and parallel markets. Adjustment will alter the quickly dispersing. market conditions faced by households and en- Stabilization on the one hand and structural terprises through changing relative prices or the adjustment on the other should be regarded as quantities traded in the markets. It might also two types of adjustment response by African change the characteristics of marketing institu- governments. In restoring internal and external tions, and thereby influence the way in which equilibriums, stabilization is intended to realign market signals and opportunities are transmitted domestic absorption with domestic supply, to households. whereas structural adjustment is designed to * Economic infrastructure is a crucial part of the change the equilibrium configuration itself (Buiter meso economy in Sub-Saharan Africa. Its provi- 1986). sions are directly determined by macroeconomic policy, including public expenditures on physi- The policy problem cal infrastructure and support services. These can have a noticeable effect on the primary incomes The relation between structural adjustment and generated in the micro economy. In many cases poverty can be analyzed from two broad per- the effects of market changes on the incomes of spectives. It can refer to the "social" or "transi- individual households will be conditioned by tional" costs that are incurred as the economy is economic infrastructure. We include here the moved from one time path to another. With a provision of support services (such as agricul- longer-term perspective, the "steady-state" effects tural extension services) that influence the gen- of structural adjustment on poverty can be as- eration of primary incomes by households. sessed. The former deals with the costs incurred * Social infrastructure, consisting of health, by the various socioeconomic groups while the education, nutrition support, and other transfers, regime of controls is dismantled and new policy affects the returns households obtain through instruments applied. The latter assesses the pov- market opportunities. These services also have 8 direct effects on household welfare, especially in longer run (after allowing for various household the case of health services. responses). The objective, therefore, is to establish how the macroeconomic processes initiated under adjust- An outline of the book ment programs affect households, which requires an empirical understanding of the mesoeconomic We begin by examining the mesoeconomic ef- linkages. Conceptually, the SDA faces two major fects of destabilization (Chapter 2) and adjust- research challenges. First it must establish the ment (Chapter 3), on the basis of which the ef- main links between events in the macro economy fects on households will be assessed. Chapter 4 (notably internal and external shocks and the reviews the determinants of household welfare consequent adjustment programs) and the meso and the nature of the household economy. Chap- economy, the latter determining the intervening ter 5 brings the various threads together by as- variables that communicate these changes to the sessing how the mesoeconomic outcomes are households. Second, it must establish in what likely to influence household welfare. The em- ways these intervening mesoeconomic variables pirical implications of this conceptual framework affect individuals and households in both the are then reviewed in Chapter 6. Chapters 7 and 8 short run (during which households do not re- review the strategic policy issues that are involved spond to the mesoeconomic changes) and the and define the room for policy maneuver. 9 Macroeconomic disequilibrium We first consider the macro economy, and spe- * over-expansionary fiscal and monetary poli- cifically the economic imbalances that have occa- cies, in turn associated with narrow tax bases and sioned the whole adjustment process in African the poor performance of public enterprises countries. Our objective is to explain how the * domestic pricing policies biased against ag- various sources of macroeconomic imbalance riculture, especially with regard to producer might affect the "real" economies of Africa and prices, frequently underpinned by overvalued (in Chapter 5) how these changes are likely to currencies. influence households. Much of the literature on In addition, African economies have suffered adjustment and poverty ignores the distributive two types of shock: effects of the period of disequilibrium prior to the * external shocks, including higher oil prices, implementation of adjustment policies. The ef- lower prices for primary commodities, and in- fects of the adjustment on the incomes of various creases in the real interest rate on commercial groups cannot properly be explained without debts reference to how they fared during the disequili- * environmental shocks resulting from an in- brium phase. Understanding how these econo- creased variability in rainfall patterns.2 rnies are destabilized, and the consequent effects These four sets of factors have contributed to on households, is essential if observed changes in economic distress in almost all African countries, African poverty are to be understood. although their relative importance varies across the region and for individual countries at differ- The causes of macroeconomic disequilibrium ent times.3 Clearly, these movements in the terms of trade and the domestic economic strategy are In the 1980s African countries experienced large not unrelated. Buoyant commodity prices for current account deficits, high inflation rates, and many countries in the mid to late 1970s raised low growth. Indeed, in many countries these expectations about resource availability that were problems first emerged in the 1970s. Balance of not realized. Thus the decline in the terms of payments problems across the region are the most trade in the 1980s was in part the cause of over- immediate sign of economic difficulty since, in expansionary fiscal policy in many countries, since highly open economies, the external position is the policy was based on false expectations of crucial to overall economic performance. But all future price movements (as well as national and of these difficulties are visible symptoms of un- political ambition). derlying disequilibrium in the operation of their The region's net barter terms of trade fell economies. In many cases, these problems have sharply in the 1980s, but the decline was from arisen from inappropriate policies (World Bank 1984, historically high levels in 1980-81. The IDA-eli- p. 37; Zulu and Nsouli 1985, p. 1) consisting of: gible low-income countries of the region have 10 been affected most severely. They suffered a sharp justment. Each of the first two strategies contains decline in their terms of trade through the 1970s inherent weaknesses, however, that make them (mainly as a result of the oil price hikes), and incapable of producing solutions to the kinds of although the 1980s have brought a measure of problems experienced by African countries over stability for them, their terms of trade remain the last decade. Nevertheless, they only appear lower than in the past. Middle-income oil-im- to be alternatives to adjustment in the early stages porters enjoyed a rapid rise in their terms of of macroeconomic difficulty. But for various rea- trade in the 1970s, followed by a general decline sons, financing followed by exchange controls is in the 1980s. Although commercial borrowing often chosen as the policy response in the early cushioned these countries initially, the rapid de- stages. Only later is serious adjustment under- cline in finance availability has meant that more taken. Many African countries have taken up the fundamental policy adjustments have been re- three strategies in sequence: first additional ex- quired (see World Bank and UNDP 1989). ternal finance is sought, then exchange controls are tightened, and then concerted adjustment is Policy sequences in Africa implemented. Obviously this is a stylized picture of the process We have seen that the causes of economic diffi- that countries experience. In particular, the fi- culties in Africa involve a combination of domes- nancing and import-compression strategies often tic policy orientation and multiple shocks. Gov- overlap. In Sub-Saharan Africa the access of most emments have varied in their policy responses. non-oil-producing countries to international But while recent events have been complex, there commercial finance has been limited when com- are basically four courses of action that govern- pared with Asia and Latin America. Most low- ments take to deal with balance of payments income African countries reached the limits of problems: international borrowing early on in their present - finance the current account deficit through difficulties, and therefore quickly resorted to in- additional external capital inflows tensifying exchange controls. Although govern- * reduce the deficit to the level of available ments have sometimes applied adjustment mea- external capital inflows through tightening capi- sures in the early stages, these have often taken tal and trade restrictions second place to seeking finance and intensifying * undertake macroeconomic stabilization poli- exchange controls. Thus a devaluation may take cies to restore internal and external balances, place, but often it has been of insufficient size mainly through fast-acting restrictive monetary compared with the scale of the currency's and fiscal policies overvaluation. In addition, it is generally not * implement structural adjustment policies at coordinated with sufficient monetary restraint, both macro- and microeconomic levels, involv- so that the real exchange rate eventually resumes ing market and trade liberalization and institu- its upward trend because domestic inflation tional policy reforms. grows ahead of the world inflation rate.4 Examples Under the terms of the definitions introduced of African countries that have taken action quickly above, the last two are adjustment strategies. Each after an external shock are relatively few.5 So while of these strategies has implications for the other a stylized picture has been presented, it approx- main targets of macroeconomic policy - the in- imates the actual path that African governments flation rate and the growth rate. In addition each have followed (see Zulu and Nsouli 1985, p.3). has ramifications for employment and income A further complication is that a number of distribution and for the political economy within countries have started along the path, but have which governments operate. Their preferences then doubled back. They have gone through the affect the weight assigned to each of these policy sequence from financing to import restriction to objectives in the final choice of strategy. At the adjustment, but have then abandoned adjustment same time governments operate within con- and reverted back to earlier strategies. This has straints imposed by the structures of both the frequently been related to changes of govern- domestic economy and the international economy. ment - for example, those that occurred in These, too, affect the ranking of policy objectives. Uganda during the 1980s and in Ghana during The first two strategies - financing and ex- the 1970s. Temporary windfalls have also led to change controls - are often viewed as alterna- the relaxation of adjustment efforts begun in an tives to each other and to the third strategy, ad- earlier period - for example, the coffee price 11 boom in the mid-1970s contributed to the relaxa- trade policy measures in small, open economies tion of the adjustment efforts made by coffee- (for example, see Dornbusch 1980; Lal 1984,1986, growing countries such as Kenya after the first 1988; and Corden 1985; for applications to Afri- oil price shock. The boom in the copper price in can countries, see Collier 1988, Collier and Lal the same period delayed Zambia's adjustment to 1986, Collier et al. 1986, and Devarajan and de the unfavorable long-term trend in the world Melo 1987).7 We use this model to explore some copper price (World Bank 1981, p. 29). of the main consequences of destabilization and, In addition, the history of events over the last in Chapter 3, to trace the effects of macroeco- decade is not one of African countries having to nomic adjustment. adjust to a one-off shock only. In the 1980s African The framework that we present here has two countries have found their economies shocked features that bear mention at the outset. First, it from both the import side (including sharp in- assumes initially that the economies are at full creases in the prices of imported energy and in- employment and that there is sufficient wage- ternediate and final goods) and the export side price flexibility to maintain this status. It is debat- (lower commodity prices). Aside from these ex- able whether this assumption is appropriate for ternal shocks, many have suffered internal, envi- African cases and, in the last analysis, this can ronmental shocks to their agricultural sectors, only be decided empirically, on a country-by- often occurring simultaneously with the external country basis. Of course, alternative models are shocks. Adjustments dealing with the first round available that involve an explicit recognition of of shocks have had to be extended and widened structural rigidities (for example, Taylor 1983) or to cope with these new shocks. Consequently of quantity rations in product and factor markets macroeconomic targets that would have been (Cuddington , Johansson, and Ofgren 1984).8 achieved, given the measures applied, have Second, and perhaps more important, it is a com- sometimes not been realized because of a fresh paratively static equilibrium framework. It can deterioration in the world economy or new reveal something about the dynamics of change droughts.' Given the structure of African econo- in an economy responding to various shocks, but .-ties, delayed responses to the implementation it has to be admnitted that this is not its strength. of adjustment policies have been inevitable, so Its main advantage is that it is well understood, that a time lag exists between the start of a vigor- and it is a simple yet powerful tool for showing ous adjustment program and the realization of how policy instruments influence real economies. all its objectives. Finally, although large groups Before proceeding to the framework, some of countries in Africa have suffered from similar words of caution are in order. What is needed shocks at the same time (for example, the non-oil- ideally is a model that is applicable to all Sub- producers in 1978-80), others (the oil producers) Saharan African countries. Clearly, this is quite benefited during the same period. The gainers out of the question in any exact sense. What we and losers were reversed after the oil price began are obliged to present, therefore, is a framework to fall back in the mid-1980s, and countries such based on stylized facts, which, although unable as Nigeria are now having to implement adjust- to capture the full complexity of economic inter- ment programs. actions during adjustment, and certainly not the variety of economic structures and experiences An analytical framework in the region, will highlight critical relationships that apply to a greater or lesser extent in all coun- Our concern in this section is to trace analytically tries. As Collier (1988, p. 1) argues: how these destabilizing factors might influence the real economies of the region, particularly the At their best such models, by dispensing meso economies. In Chapter 5, we explore how with complex but minor detail, clarify ma- these mesoeconomic changes affect households. jor consequences of policies which might The approach adopted here is to explain macro- otherwise have been obscured. At their mesoeconomnic processes (at least initially) by us- worst they are a misleading caricature, dan- ing the standard international trade or "ortho- gerous because of their apparent rigour. dox" model based on the Meade-Salter-Swan, or "dependent-conomy," model. This has proved The orthodox model is presented to structure both popular and useful in analyzing the real our thinking about these issues - it is not in- economy implications of macroeconomic and tended to be strictly applied in each and every 12 case. There will certainly be departures from the to identify where such changes occur.10 A second real world, and some of these may be critical. The difficulty is that commodity classifications may actual mesoeconomic effects can only be deter- change geographically. A certain commodity may mined empirically. The general applicability of be importable at or near the port of entry, but as the model to the African situation is reviewed transportation costs increase its price in remoter below, however, and alternative approaches are areas, it may become entirely insulated from briefly discussed. world markets. This problem may have become more acute under adjustment, because transpor- Tradables and nontradables tation networks in Sub-Saharan Africa have de- teriorated, and costs have risen accordingly. At the core of dependent economy models is the There are sectors with outputs that clearly fall distinction between "tradable" and "nontradable" under the tradable label, such as production of goods and services. Nontradables are goods and cash crops for export. Similarly, many govern- services whose prices are determined by domes- ment services are unquestionably nontradable. tic supply and demand. This is due to the nature In between these pure cases, there lies a grey area of the good involved (for example, public ser- of conceptual ambiguity. Our preference is to vices, housing, and construction) or because include any sector that is protected by severe transport costs prohibit either the import or the import quotas under the nontradable banner, export of the good in question and insulate it since changes in the world price will leave do- from world markets. Tradable goods are those mestic prices unaffected and will only influence that cross frontiers and, in theory, their prices are the margins obtained by importers. Furthermore, determined directly by world market conditions, our inclination (at least for most countries in Sub- so that for a "small" economy9 tradable prices are Saharan Africa) is to treat food production as assumed to be exogenously given. One of the tradable."' Food, in contrast to manufactured more important problems encountered in using products, is rarely protected. In the last analysis, this type of classification of product markets is however, the categorization has to be country- that commodities can switch categories, fre- specific. quently in response to the type of policy change The definition of 'tradable" production is under investigation. The most important reasons bound to raise serious conceptual and empirical why goods are nontraded are commercial policy problems when applied to African economies. (for example, prohibition of imports) and trans- We employ the tradable-nontradable distinction portation costs. Taking the transportation markup here because it allows us to use a class of models to be q and the world price of a commodity to be that capture the main macroeconomic processes P*, the domestic price must be equal to or less powerfully and simply, and because it is essen- than P'*(1+q) for it to be exportable (assuming tial to an understanding of how macroeconomic no trade taxes or subsidies). Similarly, for the policy instruments will influence relative prod- commodity to be importable, its domestic price uct prices and resource allocation. In practice, must be greater than P*(1+q), as otherwise its im- goods have to be ranked in terms of their porter would not be able to compete with domes- "tradability," so that we can say that one good is tic suppliers. Thus we have a range of domestic more tradable than another, in that its domestic prices for which the commodity is nontradable price is open to greater influence by world prices.' - neither an exportable nor an importable. This range is simply given by A three-sector model Ps 5 P*I(1+q) < P,, < P*(l+q) 5 Pm. In order to illustrate how destabilization and ad- justment policies have important mesoeconomi,c The difficulty here is that a commodity price effects, consider a simple three-sector economy.'3 can cross these boundaries and move from being Aside from assuming that the country is a price nontradable to being either an exportable (if the taker in world markets (which is an assumption domestic price falls sufficiently) or an importable held throughout this section and the next), we (if the price rises). Although we assume away also begin by assuming that: changes in commodity classifications in the simple * two tradable commodities are produced, an analytical framework presented below, the exportable (X) and an importable (M). Their country-based empirical work would clearly have prices (P2 and P., respectively) are givenby world 13 markets, with Pm taken as the numeraire; we as- Ps IP, = eP,*IP. sume that the nominal rate of exchange (e) is fixed; P1 /Pm = PY*P *(1+k) a nontradable (N) is also produced, its price being Pn where e is the nominal exchange rate, k is the tar- @ Pn is determined by domestic supply and iff rate (or equivalent), the asterisk refers to world demand factors, the latter being partly depend- prices, and where there are no taxes or subsidies ent on monetary and fiscal policy - other things on exports. The two independent relative prices being equal, an expansionary fiscal-monetary are P IP11' and Pn /P3, the former depending on policy will raise P. the international terms of trade and on the trade * for simplicity, all three commodities are only restrictions imposed, while the latter will be de- for final consumption and are not used as inter- termined by the exchange rate and the domestic mediates in the production process; in addition, money supply (which determines the price level exportables are only consumed abroad, domestic of nontradables). consumption being confined to importables and Ps /PI and Pn /IP. are traced on the vertical and nontradables horizontal axes, respectively, in Figure 2.1. This * product and factor (labor and capital) mar- means that no change in PJnPm occurs for vertical kets are perfectly competitive, so that the econ- movements, and no domestic terms of trade omy is in equilibrium on its production frontier; changes occur horizontally. NN is a locus of val- product and factor prices are assumed (initially ues of these relative prices, which gives equilib- at least) to be flexible rium in the nontradables market. Along the curve, a capital is sector-specific in the short run, so the supply and demand for nontradables are that only labor reallocations can change the out- equal, so that put mix in the economy; in the long run, both capital and labor may be reallocated between DM(PnIPf,y,MIP) = Y(PxI/PM,PnIPv,K). sectors. With the terms of trade (P /Pm) fixed, export- In other words, the NN locus (which is in rela- ables and importables can be combined into a tive-price space) is drawn on the assumption that Hicksian composite commodity - tradables (T). real income (y) and real money balances (MIP) are This composite proves useful in analyzing the constant. Changes in any of these will shift the effects of policies (such as devaluation) that are curve. The NN curve can also shift in response to designed to change the real exchange rate (PJ/P,). movements in capital stock (Kn) into or out of the Once the external terms of trade change, or should sector. Thus, if capital is transferred out of the govermnents adjust trade interventions (changing tradables sector into nontradables, the NN curve PJP~), however, the composite breaks down. would shift to the left. To the right of NN, With the relative price of exportables and importables influenced by both external shocks and discretionary policy, the use of the tradables composite would be difficult to justify. Figure 2.1 In such a model, the domestic price of exportables (P.) will be determined by the world P/P,, price, the exchange rate, and any export taxes or subsidies imposed. Similarly, the importables EL% price (Pm) will be given by the world price, the exchange rate, and any import controls or tariffs in place. The price of nontradables (P.) will be determined by domestic demand and supply conditions. The demand for nontradables will L depend, in part, on fiscal and monetary policy, which will therefore play an important part in / determining P.. There are thus three relative prices defined in such a model, only two of which __N are independent: PJ/P Pn / Pm = Pn /eP3(1+k) 14 nontradables will be in excess supply and to the Figure 2.2 left, in excess demand. NN is positively sloped, and its slope will be greater than a ray through PJP-, the origin (Collier 1988, pp. 2-3). N' Similarly, the LL locus denotes values of these relative prices that give equilibrium in the money market, assuming that real income, asset demand, L / I and money supply are constant. That is, LL satis- , D fies MS =Md(P,R,fl) Ay,MMIe. 1 Again, with the locus drawn in relative-price ,' space, it will shift with changes in asset demand ,N" N /N' (A), real income (y), and money supply expressed p./p in foreign currency (MW/e) - e being the rate of exchange.'4 With the price level too high below LL,15 the money market is in excess demand, while exists, which is then disturbed by these two influ- above the locus, the market is in excess supply. If ences. money supply were to increase (caused either The deterioration in the terms of trade reduces endogenously by a balance of payments surplus real incomes (y), and thereby shifts the NN locus or exogenously through discretionary policy), the to N"N" (Figure 2.2). The initial movement of the LL curve would shift downward. As drawn, Fig- system depends on whether the deterioration is ure 2.1 shows that full equilibrium is at A, with due to an import or an export price shock. With both money and nontradables markets being in an import price shock (for example, with the two equilibrium. It follows (from Walras law) that oil price shocks), the economy will move to a there is zero excess demand in the tradables mar- position such as B (P,/PX will remain unchanged ket and that the balance of payments is in equilib- initially), while an export price shock would take rium. With this simple apparatus, we can now the economy to C (with P. IPM unchanged). As analyze how destabilization and adjustment may drawn, the market for nontradables is assumed influence these key relative prices. to move into excess supply. For purposes of exposition, we shall trace the At the same time, expansionary fiscal and changes in the economy following what may be monetary policies have shifted the LL curve to L'L'. considered typical destabilizing processes in Af- The expansion in real money balances will cause rica. Since the different sources of destabilization demand for nontradables to increase, and the NN have conflicting effects on the positions of the curve will shift back to the right to N'N'. As de- curves, the net outcome is theoretically ambigu- picted in Figure 2.2, the nontradables market is ous. We show for illustrative purposes one pos- now assumed to move into excess demand, with sible outcome. Country-specific experiences can the monetary expansion more than compensat- of course be accommodated in this framework ing for the deflationary effects of the terms of (see Korayem 1989). trade deterioration. Thus, with no change in dis- cretionary policy, the economy is jolted into a Terms of trade shocks and inappropriate macro disequilibrium situation, with Z signifying the policies: the case of financing combination of relative prices that would rees- tablish equilibrium in all markets, but with As we have already noted, excessive expansion disequilibrium prices (at B or C, depending on the of the public sector was associated with optimis- nature of the terms of trade shock) prevailing. tic expectations of resource availability in gen- Assuming no further changes in the terms of eral, and movements in the terms of trade in trade (or in government trade policy), the excess particular. To simulate this combination of a de- demand for nontradables at B or C will cause P, terioration in a country's terms of trade com- to rise until the equilibrium in the nontradables bined with an expansionary fiscal and monetary market is restored (at D)."6 This is partly because policy (which is financed through foreign bor- demand for nontradables will fall with the price rowing), we assume that an initial equilibrium rise, but also because factor reallocations from 15 exportables and importables will raise the sup- example, Latin America - where the distribu- ply of nontradables. tion of credit, although unequal, is more dis- Thus, the combination of a deterioration in the bursed, and thus the supply-side effects of credit country's terms of trade and an expansionary expansion can be expected to be greater. Insofar fiscal and monetary policy might be associated as formal credit is concentrated in the public sector with a movement in the key relative prices from and urban services and industries, which are A to D, through either B or C (depending on the mainly nontradable (see below), the effect of nature of the terms of trade shock). Full equilib- monetary expansion on the supply side of the rium, however, is not restored, because the money product market will reinforce the effect on the market is in excess supply and, by implication, demand side of the associated real appreciation there exists excess demnand for tradables and a to further shift output toward nontradables. balance of payments deficit. The combination of Drawing all the threads together, we may now these shocks, in taking the economy from A to D, summarize the effects of a terms of trade shock is seen therefore to lead to a decrease in Pr/P,m an combined with a fiscal and monetary expansion. increaseinP./P andadecreasein PJ/P.17 These It has caused:21 relative price changes will then signal resource * P.tPM to fall reallocations, specifically from exportables into * Pn to rise, so that PI/P, falls nontradables and importables. * resource reallocations from exportables to These shocks, however, will also induce changes nontradables and importables in factor markets as resource transfers take place * real wages to fall in the long run, with am- between the sectors. Following Edwards (1988a), biguous short-run effects. and assuming that exportables are the most labor These are some major mesoeconomic effects of intensive while importables are the least, we can destabilization on product and factor markets. trace the effect of a terms of trade shock on the There would also be effects on the other compo- factor markets. Recall that the shock results in a nents of the meso economy - namely the eco- decrease in Pr/P, and the domestic effects of this nomic and social infrastructure. Insofar as the may be fueled further by increasing import con- disequilibrium has been caused by expansionary trols, that will have the effect of decreasing the domestic fiscal and monetary policies in the relative price further.'8 With resource flows out countries concerned, there may be direct effects of the labor-intensive exportables sector and into on these elements. It is also possible that the capital-intensive importables, the terms of trade period of destabilization is associated with major shock will lead to long-term adjustments in fac- institutional changes - for example, in market- tor markets that raise the rental rate and lower ing - which have a powerful influence on the wages. Insofar as poorer households derive their meso economy. The fall in the price of export- income from selling their labor services, these ables may be aggravated by these changes if gov- shocks are likely to hit them particularly hard.19 emient marketing boards keep producer prices Edwards (1988a) also analyzes the short-run ef- below market levels asameans of raising revenue. fects, in which capital is sector-specific. In this These effects, of course, can only be determined case, a terms of trade deterioration will decrease at the empirical (country) level. the production of exportables, increase the pro- duction of importables, but have ambiguous ef- Effects of import compression fects on the production of nontradables. Similarly, the effect on the real wage is ambiguous and will The liniit to the financing strategy is reached when depend on the consumption bundle of workers. the supply of extemal finance becomes inadequate (A simple framework for analyzing these short- to meet the rising current account deficit. At this run effects is presented in Chapter 3.) point the government concerned may choose to Under this scenario direct effects on the supply implement an adjustment program. Many gov- side will also occur as the cost of credit falls with emments, however, resort instead to a reliance the expansion of the money supply, and producers on tightening import restrictions and exchange find it cheaper to borrow from the banking sys- controls as a balance of payments strategy. It is tem.20 Overall, the effect of monetary expansion the real economy effects of this strategy that con- on aggregate supply is likely to be small and cern us here. Both tariffs and quotas are generally highly concentrated in most African countries in place prior to the onset of severe balance of compared with other developing regions - for payments problems, since they are instruments 16 of import-substituting industrialization. In situa- Figure 23 tions of balance of payments crisis, resorting to quotas has been the main way of extending pro- N/ N- N/ tection, although tariff rates are also raised. In our simple three-sector model, a govern- ment can intervene by changing the level of trade restrictions, the rate of exchange, and the budget deficit and money supply. Figure 2.3 depicts the situation immediately following the destabiliz- ing events that were illustrated in Figures 2.1 and 2.2. To restore equilibrium, policy choices should involve some combination of fiscal and monetary contraction (shifting the LL curve upward to- ward D) and trade policy changes to decrease P/ 0 PM and P1 /Pm - that is, an increase in protection /P. that would raise Pm. One solution is at E, involv- ing total reliance on exchange rate and monetary policy, which would shift LL to L*L* and NN to the level of demand for tradables if the govern- N*N*.2 In this case, no import controls are im- ment maintains its expansionary fiscal and mon- posed, and no policy induced change is made in etary stance. In addition, if the government is P/Pm. This adjustment mechanism can readily forced by the foreign exchange shortage to restrict be analyzed using a two-sector dependent econ- importable intermediate goods, domestic pro- omy model, because the adjustment process leaves duction of both tradables and nontradables will Px'Pm unchanged and shifts the economy hori- fall; the size of the fall will depend on how the zontally in terms of Figure 2.3. Governments scarce import quotas are rationed among sectors may wish to adjust trade policies to restore a and the possibilities of substituting home-pro- compatible and sustainable combination of policy duced inputs for imports in the production pro- instruments. An increase in tariffs or import cess. Thus, in addition to the contraction of quotas would lower both PJPm and PnIP., and available goods from the cut in imports, aggre- move relative prices along the ray OD (say to F). gate domestic supply will also fall through the The degree of fiscal and monetary contraction reduction in domestic output of both tradables (shifting the NN and LL curves) necessary to re- and nontradables. store equilibrium will consequently be reduced, With aggregate demand continuing to grow taking the economy to F (rather than to E). and aggregate supply reduced, the inflation rate If government policy can be characterized as will increase, given that the tradables market can initiating a movement from D to F, it is clear that no longer clear itself through sucking in imports. it places much greater emphasis on a decline in As inflation accelerates, many countries tighten PxJPm (that is, an increase in import protection) price controls. Consumers accordingly find than the movement from D toE, which is achieved themselves rationed in the controlled markets. through increases in P1IPn and P,1mP (as a result But with monetary policy driving excess demand of devaluation and fiscal contraction). The move- in goods markets, equilibrium prices steadily ment from D to E would be associated with a move above controlled prices. Sellers and buyers general shift of resources out of nontradables into therefore have an incentive to establish parallel both exportables and importables, whereas the markets. With the imposition of import quotas movement from D to E implies resource reallo- and price controls, economic rents become in- cations from nontradables and exportables into creasingly important determinants of household the importables sector. incomes. Households with access to the increas- These import restrictions, however, will not ingly scarce commodities can make large gains as only affect relative prices in product markets. the prices of those goods rise in parallel markets. They can also result in quantitative adjustments Access to scarce consumer goods is generally in these markets that will have far-reaching ef- gained through rationed import licenses and fects on both the meso economy and on the smuggling goods from neighboring countries. households. Applying quotas to imports cuts the These are Directly Unproductive Profit (DUP) supply of tradable goods, but it does not reduce seeking activities, where DUPs are defined as 17 activities that give pecuniary gains to those en- ports, rationing the available supply among pro- gaged in them, but with no corresponding out- ducers. As import constraints tighten, capacity put of goods or services." Since these activities utilization and output fall. This is exacerbated by take resources away from productive activities, the high dependence of import-substituting in- they shrink the economy's production possibili- dustries on intermediate imports, the result of ties for such goods (Srinivasan 1985, p. 46). previous macro and trade policies. The rise in the overall inflation rate (as mea- Unless the government reverses its policy of sured by uncontrolled parallel market prices) will keeping producer prices low, real producer prices be accompanied by changes in relative prices continue to decline as inflation accelerates, thus (again measured through parallel markets). The reducing the incentive to produce cash crops. real exchange rate will continue to appreciate Moreover, even if producer prices are raised at because its course is driven by the excessively this stage, the supply response of smallholders expansionary monetary policy. Thus, the relative may be limited, because they are now rationed as price of nontradables to exportables will con- buyers in the consumer goods market. Extra cash tinue to move in favor of nontradable activities. no longer buys them "incentive goods." Imports Given the rise in the prices of consumer, capital, of intermediate inputs for farming are also ra- and intermediate goods and static nominal pro- tioned, and typically larger and better-off farm- ducer prices, the real price received by the farm ers gain more access to the rationed inputs than household will fall. While agriculture as a tradable smaller and poorer farmers (World Bank 1986a). activity will be disfavored under this policy sce- This, together with the fall in capacity utilization nario, within agriculture incentives tend to shift of domestic agroindustries and the effects on the toward food crops and away from export crops crop collection and transport systems, further through the operation of the parallel market. The disrupts agriculture. domestic parallel market is usually much larger In the labor market, the contraction of capacity for food crops than for nonfood export crops. utilization leads to short-time work and increas- Thus food producers, faced with low official ing redundancies. Up to this stage employees in prices, find it easier to redirect their sales to the the protected import-substituting industries have parallel market, while export crop producers, been relatively favored, particularly through the unless they can smuggle their produce to neigh- sharing out of excess profits into higher wages. boring countries with better prices, must con- The contraction of employment in the manufac- tinue to sell most of their output to the state at turing sector will put pressure on money-wage controlled prices. Thus, the structure of parallel levels, and the degree to which reduced labor market prices favors food over export produc- demand is translated into reduced employment tion, and farmers redirect their resources into or reduced wages will depend on the flexibility food. Since both food and export crops are grown of wages. Faced with demand constraints in the on most small farms, this is relatively easy for labor market, the urban unemployed will look them to do. Over time, with demand expanding for alternative incomes. With the contraction of as the budget deficit widens, the structure of prices formal sector job opportunities, new labor market between food and nonfood export crops on the entrants will look increasingly to the infornal parallel market will shift further toward food. sector for their survival. Employment in the in- This will compress the income differential enjoyed formal sector accordingly increases during such by export farmers over food farmers (Lele 1985). periods. Nevertheless, aggregate real incomes are With aggregate supply now restricted by the now falling with the decline in formal sector ac- tightening of import controls, excess money bal- tivity, so total sales of informal sector products ances build up as the government budget deficit tend to decline as well. Accordingly, informal continues to generate a monetary supply expan- producers find themselves competing for a sion in excess of money demand. As the current shrinking market, and average incomes in the account deteriorates further, the government, sector usually decline. unable to obtain financing, further restricts im- The contraction of gross domestic product ports. In the early stages of the crisis, priority has (GDP) reduces the government's revenue base, usually been given to imports of key intermedi- which in turn raises the public sector deficit, thus ate inputs. But as the external situation deterio- adding to inflationary monetary expansion. The rates further, a position is eventually reached revenue base also stagnates, because as market where cuts begin to be made in intermediate im- prices rise above controlled prices, a larger num- 18 ber of sales are conducted through parallel mar- falls; for example, schools and clinics cannot pro- kets. Incomes generated in parallel markets avoid vide their previous levels of education and health income tax. Moreover, sales taxes are usually set care because of the increasing scarcity of school as a given percentage of the controlled price level, books, drugs, and so forth. Like the strategy of so that the difference between the price on the financing the trade deficit, the option of reducing parallel market and the controlled price is effec- the deficit through cutting imports does not solve tively not taxed. the underlying disequilibrium and the economy With imports cut and domestic supply rationed, remains unstable. Unless corrected by sufficient governments increasingly find that they are un- policy adjustments, a spiral of falling output and able to maintain both economic and social infra- imports is generated that feeds on itself. structure. The effective supply of public goods 19 3 Adjustment In the previous chapter we saw how economies Adjustment policies in Africa and their macro can be destabilized by both inappropriate pol- effects icies and a variety of shocks. If governments rule out entirely the correction of policies in the form A number of recent studies (for example Balassa of an adjustment program, they havetwo choices: 1988a; Thomas and Chhibber 1989; World Bank either to find external financing for the trade 1988a, 1988b; and World Bank and UNDP 1989) deficit or to compress imports. Each of these ap- have made anassessment of structural adjustment proaches has implications for the meso economy, lending (SAL) and its impact in the developing which we explored in previous chapters. We now world. These studies provide useful information take up the story at the point at which the gov- on the experience of African countries with struc- emnment has decided to tackle the fundamental tural adjustment. The first and obvious question distortions underlying the internal and external that has to be addressed concerns the content of the imbalances through the implementation of a structural adjustment programs: what were the comprehensive adjustment program. policy instruments manipulated under the pro- As we have already noted, adjustment can be grams. In its review, the World Bank (1988a, p.88) achieved through stabilization, which mainly in- provides a useful summary, which is repeated in volves short-run demand management, and Table 3.1. through structural adjustment, which introduces There are three major implications of these re- microeconomic and mesoeconomic interventions sults. First, adjustment lending by the World Bank to the adjustment process, often involving insti- in the African countries reviewed has embraced tutional reforms. Stabilization and structural ad- a wide range of policy interventions, with energy justment are not alternative modes of adjustment, being the only sector not covered in Africa as and many countries implement stabilization much as elsewhere. More attention seems to have packages in agreement with the International been paid to public expenditures, public enter- Monetary Fund (IMF) and through structural ad- prises, agricultural policy, and (surprisingly) in- justment programs with World Bank support. dustrial policy in SAL conditionality in Africa. The latter essentially take a medium- to long- Although (for reasons given in the table) the data term policy perspective. Before considering how are difficult to interpret, the evidence also sug- these policy responses may affect the elements of gests that exchange rate interventions are more the meso economy, we briefly review the types of likely in adjustment policies in Africa than else- adjustment response that have been observed in where. Trade policies, public enterprises, and ag- Africa. ricultural policy account for 61.8 percent of the total number of policy conditions set under ad- 20 Table 3.1 The policy content of World Bank Finally, the World Bank study found that in lending operations most cases there were four or five key policy con- (percentage of total number of loans with conditions in various ditions in each structural adjustment program, policy areas) most of which concern trade policy (35 percent), SSA All countries public expenditure and fiscal policy (19 percent), public enterprise reforms (14 percent), and pric- 1. Exchange ratea 30.8 15.7 ing policy (especially agriculture and energy 2. Trade policies 76.9 78.4 pricing, comprising 14 percent). While the Bank 3. Fiscal policy 61.5 64.7 study reports only the figures for all 15 countries, 4. Budget/public expenditure 69.2 51.0 it is likely that similar orders of magnitude apply 5. Public enterprises 61.5 52.9 to Sub-Saharan African countries (except the role 7. industrial polcy 53.8 235.5 of energy pricing policies which were not a fea- 8. Energy policy 7.7 23.5 ture of most Sub-Saharan African programs). 9. Agricultural policy 76.9 49.0 10. Other 23.1 13.7 MACRO EFFEcTS. Before we apply our analytical framework to the mesoeconomic effects of these Note: Lending operations under structural adjustment and sector adjustment loans in Africa (Ghana, Kenya, Malawi, and Zambia) and policies, we shall briefly summarize the main 11 other countries. macroeconomic effects of the adjustment pro- a. Since the 3MF has responsibility for exchange rate policy, these figures underestimate the importance of exchange rate conditionality grams as reported by the World Bank (1988a). in the Bank's adjustment lending. This study attempted to assess the impact of ad- Source: World Bank (1988a), Table 42 justment programs on performance indicators,14 and it adopted a simple methodology for evalu- ating whether adjustment enhanced the perfor- justment lending by the Bank in Africa (World mance of these indicators. Two approaches were Bank 1988a, Table 4.1). adopted: the first compared the performance in- Second, while these data give an indication of dicators during the three years before the first the content of the loan agreements, they do not year of an adjustment program with the perform- necessarily reflect the ex post operation of policy ance during the following three years; the second instruments. Some attempt was made in the compared unweighted average values of the in- World Bank review to assess the extent to which dicators for all countries in receipt of adjustment these policy conditions were implemented. It lending (AL countries) with those of other coun- found significant variations in the ability of gov- tries in each country group (NAL countries). The emments to undertake the specified policy inter- latter method was to compare the change in the vention: while 70 percent of exchange rate policy indicators between the three-year periods before conditions were fully met in the 15 countries in- and after the implementation of the adjustment vestigated (four of which were in Sub-Saharan program with the change in the indicators expe- Africa), only 57 percent of agricultural policy rienced by countries not in receipt of adjustment conditions and 55 percent of trade policy condi- loans. Our interest obviously lies in the results tions were fully met. The report concludes that for Africa. "the policy areas where implementation has been The results for the Sub-Saharan Africa country most successful are those involving changes in group reported by the World Bank are given in prices such as exchange rates, interest rates, or Table 3.2 (overleaf). The numbers in the table agricultural and energy prices; those where po- indicate the number of countries in the AL group litical sensitivities are the least . . . and those that performed better than the NAL control. The where institutional changes ... are not required" sign indicates whether the direction of change in (World Bank 1988a, p. 90). There was some evi- the average value of an indicator was better (+) dence that the policy performance of African or worse (-) in comparison with the same indica- countries was not as effective as performance tor for the NAL group. These data show that the elsewhere. The differences in the proportions of performance of Sub-Saharan Africa AL countries all policy conditions fully met, however, (just is decidedly mixed when compared with the con- over 52 percent in African countries compared trol group. There is evidence in these results of a with 60 percent for the others) is unlikely to be general improvement in the two key imbalances significant in such a small sample. - the balance of payments current account and 21 Table 3.2 Relative performance indicators An orthodox framework for analyzing meso for 15 SSA countries effects Indicator Value Devaluation-expenditure reduction 1. GDP growth 7 (-) PRODUCr MARKET uFEcrs. This review of adjust- 2. Investment/GDP 6 ( ment policies in Africa and their effects would 3. Export growth 10 (+) suggest that our analytical framework must in- 4. Real exchange rate 11 (t) corporate both expenditure-switching policies 5. BoP C/A deficit/GDP 8 (+) (depreciating the real exchange rate and increas- 6. Budget balance/GDP 6 (+) ing the growth of agricultural production, ex- 7. Inflation 5 (-) 8. External debt/GDP 9 ( ports, and import substitutes) and expenditure- 9. Debt service/exports 7 ( reducing policies (as evidenced by the budget balance performance indicator). While there is Note: Number of AL countnes: 15; number of NAL counfr1es: 22. evidence inter studianes menictioned above thati Source: World Bank (1988a: Table 2.4a). evidence in the studies mentioned above tt institutional changes have been slower to imple- ment, so that our model may be a reasonable account of the effects that have been experienced, it must be acknowledged that this framework is the budget deficit. The former seems to have been only an approximation, intended to structure our brought about by expenditure switching policies, initial thinking. In many countries, institutional as evidenced by the favorable indicators of ex- reforms can have fundamental and dynamic ef- port growth and the real exchange rate.35 The ex- fects on the economic system - effects that can- ternal debt situation also improved compared not be handled in this simple comparative, static with the control group. The longer-term effects, framework. however, are not so favorable. Three key indica- To trace a policy combination of devaluation tors reveal an unfavorable comparison with NAL and fiscal/monetary contraction, we retain the countries - GDP growth, the investment-GDP apparatus described in the previous chapter. For ratio, and the rate of inflation. The first two of this policy package, we can assume that PJP.n these suggest that the longer-run prospects for remains unchanged, since in our simple model rapid growth have not been enhanced through (with no intermediate demands) this relative price the policy interventions that have been made. is unaffected by either a change in the nominal The poor inflation performance suggests that the exchange rate (e), or in P^. Figure 2.4 illustrates favorable real exchange rate indication may not the movement in relative prices as a policy com- continue into the future. bination of devaluation and fiscal/monetary These findings are similar to a number of other contraction is applied. Recall that following the recent studies - for example the World Bank destabilizing events traced in the previous sec- and UNDP (1989), Balassa (1988a), Sahn (1989), tion, the economy is in disequilibrium at D, on the World Bank (1988b), Thomas and Chhibber (1989), N'N' curve (that is, in internal equilibrium) but and ECA (1989).36 Taking the combined evidence off the LL locus - the money market being in of these studies, questions undoubtedly remain excess supply. This means that there is an excess over how responsive sectoral and aggregate out- demand for tradables - a balance of payments put in the region has been to the price signals of deficit. A judicious combination of fiscal/mone- structural adjustment. A slow supply response in tary contraction (shifting the NN curve to the left aggregate output suggests that a longer time to N*N* and the LL locus to L*L*) and devalu- perspective may be required to induce more rapid ation, taking the economy horizontally to E, will growth. We shall return to this important ques- restore both internal and external balance. tion. Moreover, institutional changes (which many The key indicator of this switching policy is the regard as critical to improving growth perform- depreciation of the real exchange rate, PJP,, which ance) require more time and are more difficult to induces a shift in production from nontradables achieve, even under adjustment lending instru- to tradables. A switching policy therefore affects ments (World Bank 1988a, p. 90). the product markets by changing the relative price regime in favor of tradables. This is illustrated in Figure 3.1, in which the production possibilities 22 of nontradables and the composite tradables are uct market liberalization, marketing reforms, and represented by the N*T* curve. In the initial the like) that affect both prices and quantities. disequilibrium situation (point D in Figure 2.4), Thus, the relative price changes brought about production is at x and expenditure is at y - there by the combined devaluation and fiscal contrac- being equilibrium in the nontradables market, tion will be modified by these policy-specific but an external deficit of xy. A policy of reducing changes. In addition, quantitative constraints may absorption (from OA to OB measured in be lifted as a result of adjustment, and these nontradables) and exchange rate devaluation somewhat profound changes in product markets (shifting the price line from AA' to BB')27 will re- can have important implications for households. store external and internal equilibrium, with both This account of the adjustment process under- production and expenditure shifting to z. Ac- scores two key elements of a switching strategy. cording to this interpretation, adjustment must First, it must be possible for policy interventions involve a depreciation in the real exchange rate, to change the underlying structure of relative which will in turn induce resource shifts into the prices - that is, change the real exchange rate. tradables sectors (exports and imports). Governments can only manipulate nominal in- This simple model predicts an across-the-board struments (such as the money supply and the increase in tradables prices - all tradables nominal rate of exchange). Whether the applica- (whether imports or exports) will benefit from tion of these instruments leads to the desired the devaluation to the same extent and will be in change in relative or real prices will depend on receipt of resource reallocations. In reality, how- accompanying macroeconomic policies and the ever, some tradables prices will be raised by more structural characteristics of the economy. If these than others and will benefit from greater resource lead to an increase in P., the real exchange rate inflows. First, if there is a parallel market for depreciation will be reduced, or even prevented foreign exchange, a devaluation will have com- altogether (Edwards 1988b, p. 29). These include: plex effects on exchange rates, because it will a continued fiscal/monetaryexpansion; real wage directly affect the official exchange rate and only resistance through, for example, collective bar- indirectly change the unofficial rate, so that the gaining or wage indexation; and the use of im- relation they bear to each other may well change. ported intermediate inputs. On the basis of This means that the price of tradables exchanged twenty-eight devaluation episodes, Edwards in the official market will change relative to those (1988b, p. 38) estimates that a 10 percent devalu- exchanged through parallel markets. If export- ation results in a 7 percent real exchange rate ables are more likely to be exchanged at official depreciation in the first year, falling off to 5 per- rates and importables at parallel rates, a devalu- cent by the third year, other things being equal. ation may change PJPm. This is the basis of Lele's The erosion of the devaluation is even greater if (1985) explanation of the increase in the price of an expansionary monetary policy is applied fol- food crops (exchanged through parallel mnarkets) lowing the devaluation. relative to export crops (marketed through offi- If P,/Pft is subject to some form of rigidity, ad- cial channels). justment cannot be achieved through switching. Second, if some importables are subject to With no change in relative prices, adjustment quantitative restrictions, devaluation will have would have to rely on expenditure contraction, first-round effects on their prices - it will simply along the income-expenditure ray (OE in Figure affect the trading margins of the importers. Third, 3.1, overleaf). To correct the external deficit, total since resource reallocations are responsive to expenditure would have to be cut to OC (meas- value added in the sector (and not simply the ured in nontradables), which would take the output price), tradables that use imported inter- economy to u. However, the restoration of exter- mediate inputs will not attract resources as much nal balance is achieved only by sacrificing inter- as sectors that do not, even though the percentage nal balance and creating factor unemployment, increase in the output price is the same. Thus, if since the economy is necessarily drawn within its an agricultural sector uses imported fertilizers production frontier. With price rigidity, therefore, and other inputs, the net resource allocative effect the nature of the adjustment process must change of the devaluation may be negligible. - involving greater short-run costs in terms of Finally, an adjustment package may entail other output loss. This "fix-price" case is considered in policy instruments (such as reduced tariffs, prod- more detail below. 23 Figure 3.1 slower than the orthodox model requires. The key to the medium-term price response in Africa tradabls lies in the implementation of nonprice reforms, such as in marketing, infrastructure, credit, and the like (Sahn 1989, pp. 72-76). A Expectations over the probable outcome of an B adjustment program will also be important. If the E program is credible in the view of the public, in the sense that the government is expected to N- \ y /persevere with its new policies, the shift in rela- A' tive prices will be viewed as permanent, and c ' < productive resources will be reallocated accord- ingly. But reallocating resources to tradables can be very costly to agents if the adjustment pro- I C' \ B' gram is canceled and the policy bias against 0 Tradable tradables is resumed. So, with uncertain expecta- tions about the program, agents will delay their The theory also highlights the key role played decisions about resource allocations as they gather by price responses in the adjustment process. information. New programs most often lack Even if governments can effect changes in the credibility when there is a history of past underlying structure of relative prices, the policy reversals. In Africa this has been all too achievement of internal and external balances common and governments, if they are comnmit- requires positive supply responses and resource ted to adjustment, must send out very clear signals flows into the tradables sector. Changing the that policy changes will be sustained. price signals should therefore be considered as a Insofar as devaluation conveys the message necessary but not sufficient condition for suc- that the government now intends to shift re- cessful structural adjustment. If there are weak sources into tradables, it enhances the credibility supply responses and impediments to resource of the adjustment program, and therefore raises reallocations, expenditure reduction (as opposed private sector confidence about investing in to production switching) would again have to tradables. Action at the microeconomic level to bear the main burden of the adjustment. While encourage greater price flexibility will also help the short-run price responsiveness of total agri- - in particular, inappropriate micro pricing cultural output (agriculture being the most im- policies that create price rigidities need to be portant tradable sector in most African countries) eliminated. In summnary, adjustment programs, has generally been found to be low (Bond 1983 which include devaluation and suitable and Binswanger 1989), there is evidence that ag- microeconomic reforms, can maximize the rate at ricultural production for export has been price which resources will move into tradables, and responsive. Cleaver (1988) has compared agricul- can therefore minimize resource unemployment tural growth rates in African countries that have costs. Thus, the way programs are designed has a implemented exchange rate depreciations and major bearing on the issue of social costs under other pricing policies with those that have not. adjustment. Although there was little difference in growth Before concluding this discussion of product rates in the 1970s, agricultural output growth in market effects, it is important to emphasize the the adjusting countries has increasingly out- critical role played by meso-level institutions in stripped that experienced among nonadjusters. transmitting the signals generated by policy to These data suggest that export production has the economic agents concerned. The model de- responded to adjustrnent policy, but the evidence fined above assumes that product markets are is too aggregative to be conclusive at this stage. A generally competitive and that an increase in the number of factors (risk aversion, poor market commodity price in the mnarket (whether it is the articulation, inelastic supply of factors of produc- domestic or the foreign market) is enjoyed by the tion, and the like) combine to suggest that supply producers (allowing for the transport and han- responsiveness in significant tracts of economic dling markups). But we know that this is not activity in Africa (especially in agriculture) is always the case. Whether the price signal ema- 24 nating from a policy change reaches its targeted be unemployed in this period while awaiting their destination depends on the nature of the markets reallocation to tradables. This transitional unem- concerned and, in particular, on the institutions ployment will gradually be reduced as tradable serving those markets. In the context of recent activities expand, because they are more labor- adjustment efforts, there is evidence that some intensive than nontradables. farmers are not receiving the full benefit of the Long-term effects, the real effects of the adjust- economic reforms because of the activities of ment process described, will depend on the rela- middlemen (perhaps we should call them tive factor intensities of the tradable and "mesomen") who fail to pass on the potential nontradable sectors. With full factor mobility be- price increase to the producers (see, for example, tween sectors, it is clear that production switch- Thomas 1989 and Thomas and Weidemann 1988). ing toward the tradables sector will redistribute When the nature of product markets is incomes toward factors used relatively intensively monopsonistic in this way, there can be no as- in the tradables sector. Since the tradables sector surance that policy reforms will achieve their is likely to be relatively labor-intensive (compared desired effects. It may be that the instances cited with nontradable) in most African countries, pro- are exceptions to the general rule, but even if this duction switching would, at existing factor prices, is the case they may well be important excep- lead to excess demand for labor. The increased tions. If the exceptions apply mainly to poorer demand for labor in the expanding tradables sec- groups of farmers, for example, it would have tor will exceed the supply of labor yielded by the profound implications for the social dimensions contracting nontradables sector. With labor in of adjustment. fixed supply, the real wage rate will rise in the long run (Knight 1976 and Lal 1984). LABOR MARKET EFfECrs. We now move to consider For what follows, it is helpful to illustrate the the effects of adjustment on the labor markets, since short-run and long-term labor market changes many poor households rely on the sale of labor diagrammatically. The horizontal axis of Figure servicesasthemainmeansof livelihood. To do this, 3.2 allocates total labor supply (O, - 0) to the two we distinguish between short-run and long-term sectors, measuring labor in tradables from the effects. In the short run, the switching effects of a left and in nontradables from the right. The ver- real exchange rate depreciation will induce a tical axis measures the wage and marginal pro- movement of labor into tradables and out of ductivity in the two sectors in terms of the nontradables. For this to happen, the real product nontraded good. The original demand-fof-labor wage (WIP, i= n,t) must decrease in tradables (to curves (the value of the respective marginal encourage increased employment) and increase product of labor curves - VMP), associated with in nontradables (to induce the release of labor).23 production at x (in Figure 3.1), give O0L, labor in The effecton the real consumption wage (which is tradables and O,Li in nontradables. With flexible the nominal wage deflated by the consumer price wages, the labor market clears at the wage WO. index - the latter being a weighted average of tradable and nontradable prices) will therefore be Figure 3.2 ambiguous and depend on the consumption bundle of workers. If wage earners consume wt w mainly nontradables, their real wages are likely to rise in the short run. While in theory the movement to point z in c Figure 3.1 entails a smooth adjustment around the production frontier so that total output re- mains constant, this is unlikely to occur in prac- w tice. Transitional unemployment may arise as the economy moves between these positions. \ /_\ Nontradable activities will generally contract wo , faster than tradable activities can expand, espe- cially if the latter require the rehabilitation of VMP VMPt VMPt, equipment and new investments. Hence, total _ _o output may fall in the short run, and factors will 'L L2 25 Assuming P. is fixed, the depreciation in the Figure 3.3 real exchange rate is achieved through devalu- ation - which raises P,. Given that wages and wt 'W marginal products are measured in terms of the VI" H VMP, nontraded good, the curve VMP, shifts outward to VMP,,, with VMPI remaining unchanged. La- W* bor market equilibrium is restored in the short run at the higher wage (Wi). The increase in the nominal wage (ab) is less than the increase in P, VMPt (ac), so that W)/P, falls and W,P~ rises. The short- run change in the real consumption wage, there- w\2 fore, is subject to the orthodox ambiguity, de- pending on workers' consumption propensities. wt> L L2 workers previously employed in the N sector now move into tradables production.2' The main weakness of our analysis of the I__ labor market effects of adjustment so far has been °t L2 L3 L °n the assumption that the market functions com- mal (T) sector wage to W,2, while L3LI workers opt petitively - wages are flexible and full employ- for search unemployment. The adjustment, ment guaranteed. As a representation of what therefore, leads to a rise in informal (tradables) happens in African labor markets, this is a seri- employment to O,L3; no change in formal ous limitation. The framework therefore needs to (nontradable) employment; a fall in unemploy- be brought nearer to reality by introducing im- ment to L3L,; a fall in the real consumption wage perfections in the labor market. This is done by in nontradables; ambiguity in the direction of assuming that the labor market is divided into change in the real consumption wage in tradables, formal and informal sectors, with wage rigidity depending on the consumption propensities of characterizing the former. To illustrate the impli- workers32 and a decrease in the real-wage gap cations of labor market rigidities, consider a situ- between the sectors. Whereas in the competitive ation where nontradables are produced in the case all workers experience the same short-run formal sector and tradables in the informal sec- real-wage change, with labor market imperfec- tor. This would seem to be a particularly appro- tions the effects depend on their sector of activity priate assumption in dealing with African coun- - in the case we are considering, workers in tries, where most wage data refer only to wages nontradables definitely lose as a result of a set in the formal sector, which is dominated by switching policy, while those in informal tradables public employment.30 Consider then the effects of may or may not suffer a cut in real wages in the a nominal minimum wage VW in nontradables, short run. restricting employment to O,L1 (Figure 3.3). The The model serves to underline the crucial need remainder of the laborforce takes up employ- to go beyond the simple competitive case, because ment in the T (informal) sector at the market- it yields qualitatively different predictions. This clearing wage or searches for formal N-sector account is also consistent with a number of ob- (higher-wage) employment. Employment in the servations recently made about how real wages informal tradables sector is determined where a in urban formal sectors (dominated by public rectangular hyperbola (labelled H and drawn sector wages) have fallen dramatically in many through x) intersects VMP, (at y).31 Thus, the in- African countries, while rural incomes (gener- formalsector wage settles at WwV, with employ- ated mainly through tradables) have benefited ment at 0,L2. Given this wage differential, L1L2 from favorable price trends, thus noticeably nar- workers decide to remain unemployed while en- rowing the urban-rural wage gap (see Jamal 1988 gaged in a job search (in the formal sector). and Sahn 1989, pp. 93-94). What are the mesoeconomic effects of adjust- To complete this discussion of labor market ment under these circumstances - how does it imperfections and adjustment, assume that ad- change this existing equilibrium? As before, an justment has been achieved solely through ex- expenditure-switching policy causes VMP, to shift penditure reduction, with no switching. Figure out (to VMP,,), but it leaves the position of VMP. 3.4 illustrates this case. Given the small country unchanged. The result is an increase in the infor- assumption, the reduction in expenditure will 26 not affect the production of tradables, since with Figure 3.4 WJP, unchanged, there will be no incentive for producers to change output. But the nontradables w, VMPI i sector will take the brunt of the contraction. Therefore, while the VMP, schedule is unaffected H H' by adjustment, VMPx shifts downward. With the wage fixed (at V<.) in nontradables, employment x x takes the brunt of the adjustment, falling to O0L3. Unemployment rises to L4L3, and employment in vMPt tradables rises to O L4. But note that the wage in the informnal tradables sector falls to W'2. Adjust- ment through fiscal contraction is thus seen to w", work through employment changes in the forrnal (nontradable) sector and wage changes in the wa t informal sector.33 Again, the prospects for work- -l ___ ers depend very much on the sector of occupa- L2 L L L3 0, tion - those in tradables this time take a greater real-wage cut than those in nontradables. They Which nontradable sectors contract will be gain, however, through increased employment largely determined by policy during this period. opportunities (in contrast to the employment re- The contraction in nontradables is induced by a duction in nontradables). cut in the government budget deficit (associated The analysis of Figure 3A assumes that a wage with monetary restraint). Some of the fall in cut in the informal sector is feasible. But suppose nontradables will therefore be a direct result of a that W,, is already at or near the subsistence floor, cut in government activities. Overall budget pri- and that further downward movements are sim- orities will determine which public activities are ply not feasible. No employment expansion in cut. The first to go are usually temporary em- tradables can occur, so the displaced workers ployees, and the hiring of new employees is gen- from nontradables (L1L3 in Figure 3.4) and the erally curtailed. Although the "output" of public unemployed who now abandon their job search services is reduced, the government may not, at in the nontradables sector will either find it diffi- this stage, decide to shake out its permanent work cult to find employment in tradables, or, if they force; it may prefer to leave this until later because do succeed, will cause some workers previously of implementation difficulties. Much of the con- employed there to lose their remunerative em- traction of nontradables could fall on urban ser- ployment. Either way, some workers will become vices, which, contrary to the assumptions under- destitute as a result of adjustment. If adjustment lying Figure 3.4, might be in the informal sector. relies heavily on expenditure reduction rather With the latter characterized by a flexible labor than switching, and if the free market wage in the market, the fall in demand will affect remunera- economy is already at or near subsistence, the tion rather than employment, and this is exacer- prospects for workers, and especially the poor bated by the entry of workers made redundant among the working population, are grim indeed. from other nontradable activities who seek in- This result points to a scenario that some observ- formal employment as a last resort. ers have suggested is typical of African countries in recent years. With real wages having fallen CREDrMAmKErEFECrs. InmostAfricancountries significantly during the 1980s, it is suggested that fiscal contraction is closely associated with mon- little room remains for a labor transfer process of etary contraction, so that adjustment is usually the sort suggested by the orthodox model. The associated with significant changes in the credit retreat into subsistence activities (even in urban market. The credit markets invariably consist of a areas where food cultivation is increasingly com- formal market, which is dominated by the orga- mon), together with the influence of extensive nized, modem banking system, and an informal, kinship networks, combine to suggest that there or "kerb," market. The former is directly subject to will be little labor movement to the rural (trad- the restraints that are imposed under monetary able) sector. Clearly, this is an important empiri- contraction, while the latter is affected only indi- cal matter (see ILO 198& for a review of some rectly. Typically, a credit squeeze will reduce the recent evidence). supply of credit in the organized banking systern, 27 so that many borrowers have to shift to the infor- at the meso level (for example, in the operation of mal market to obtain their credit requirements. marketing institutions), can only be addressed Because interest charges are fixed (and generally properly at the empirical (country) level. Our low) in the organized market, such borrowers face purpose here is to highlight their significance increased interest charges on their new debt. and to ensure that they are given proper consid- Interest rates in the kerb market, which are flex- eration. ible, will therefore rise as the credit contraction in In assessing the impact of adjustment programs the formal market pushes more borrowers into on infrastructure, care must again be taken to the kerb market. Thus the effect of monetary con- make the correct comparisons with the traction is to restrict the availability of credit in the pre-adjustment period of economic decline. organized market and to increase interest rates in Chapter 2 has shown that with the compression the kerb market. of imports and the decline in public resources In some programs the fixed interest regime of caused by poor economic performance, the effec- the formal market (referred to as a "repression" tive delivery of economic and social services has of the money market in the literature) is dis- almost always fallen sharply prior to the adop- mantled, so that interest charges are allowed to tion of a comprehensive donor-supported ad- settle at their market-clearing values. Thus, in justment program. If governments fail to adjust addition to any decrease in credit availability in comprehensively, or they pursue adjustment in- the formal market, borrowers may face increased sufficiently, the economy's ability to generate interest charges in the market as well. How this sufficient tax revenues to finance infrastructural affects the various borrowing units will clearly expenditures will remain weak. Without such depend on their credit dependence and on which revenues (and in the absence of external financial of the credit markets they rely on for their credit support), governments are forced to cut social needs.?4 and economic infrastructural budgets, whether they want to or not. Failure to adjust, or insuffi- iNFRAsTRucTuRAL iFFEcrs. One of the interesting cient adjustment, reduces the government's room features of Bond's (1983) study is the relatively for maneuver in its budget decisions. In such low long-run price responsiveness of total agri- situations maintaining infrastructure by increas- cultural output in Africa, which is in contrast to ing taxation is not a sustainable solution, since evidence from elsewhere in the developing world without sufficient adjustment the taxable eco- (see Binswanger 1989, Table 2). An explanation nomic base continues its decline. of why this is so must lie in what has been hap- If reductions are made in social budgets, special pening to the infrastructure of the region. In ad- attention must be paid to the preadjustment in- dition to markets, the meso economy involves cidence of such expenditures and the changes in the infrastructure, and to obtain a complete pic- that incidence under adjustment. There are serious ture of the real-economy effects of adjustment, it inequalities in access to public health and educa- is important to take into account how these in- tion services in many African countries. Reduc- frastructural elements have been affected. The tions in such social expenditures, if they occur evidence that is available suggests that agricul- without major changes in the incidence of the tural output is particularly sensitive to both eco- service, will disproportionately affect the size of nomic and social infrastructure (Binswanger 1989, benefit received by better-off households who Lele 1986, World Bank 1989d, Moock 1986, are the main consumers of such services. Never- Jarnison and Moock 1984). This may be particu- theless, while these households may suffer larger larly true of the physical infrastructure in Africa, cuts in such services than poorer households, the given the already large transport margins that latter may be more critically affected, because usually apply. even a smnall reduction could have a critical effect If the expenditure cuts applied as part of an on their health status and human capital. Fur- adjustment program adversely affect the provi- thermore, the application of reductions in health sion of economic and social infrastructural serv- budgets could be skewed toward cutting services ices, there may well be a weak or nonexistent mainly used by the poor. This is an empirical response to the relative price signals the same matter that can only be resolved through the program is communicating. These questions, careful examination of trends in social expendi- along with issues relating to institutional changes tures and their composition. Greater benefits to 28 the poor from public health and education serv- market. These relative price changes (increases ices can still be achieved despite reductions in in PJPm and P2JQ) will lead to predictable re- total expenditures. source reallocations toward exportables and out of both importables and nontradables. At the same Effects of liberalization time, these resource flows will affect the factor markets, only now there will be a tendency for Given the prevalence of import controls during the real wage to rise and the rental rate to fall, as the preadjustment phase, there is an obvious op- long as the factor intensity assumptions we made portunity for many African governments to re- earlier are maintained (Edwards 1988a). move these controls and liberalize product mar- kets. Trade reforms are currently under imple- LABOR mARKET EFFECrs. As with other structural mentation in many structural adjustment pro- adjustment policies, trade liberalization will have grams in Sub-Saharan Africa. As Table 3.1 indi- predictable effects on the labor market, and these cated, trade policy conditionality was present in can be demonstrated using a slightly modified over three-quarters of adjustment loans to Sub- version of Figure 3.2.36 To trace the labor market Saharan African countries (and import policy in effects of liberalization, we must distinguish ex- particular has been a feature of many programs portables from importables. The demands for la- - World Bank 1988a, pp. 54-56). The principal bor in exportables and importables are summed objective of these reforms is to adjust the domes- horizontally from the left in Figure 3.5, with VMPX tic relative prices of tradable goods into line with indicating labor demand in exportables and VMP, world relative prices. Thus, if governments had total labor demand in the tradables sector (note been relying on import controls, tariffs or export that demand for labor in importables is the hori- taxes and subsidies, domestic relative prices zontal distance between the two curves). As be- tended to deviate from world prices, causing a fore, VMP. indicates labor demand in policy-induced distortion in resource allocation. nontradables. Assume initially that the wage is Generally, import controls and tariffs create sys- fully flexible, but that PJ is inflexible downward tematic biases against the export sectors and (we shall review both assumptions presently). against unprotected import-competing sectors. The reduction of tariffs and/or controls will This is because the imposition of these controls reduce the price of importables, resulting in a leads to an appreciation in the real exchange rate.3 downward shift in VMP, (with VMPr constant). To correct for these biases, structural adjustment Assuming no change in PN (so that VMPx is un- frequently involves the dismantling of import changed), this will lead to a short-run decline in restrictions. This can take the form of replacing the wage (from WI to W2). The effect of this on the tariffs for import controls and reducing the level real consumption wage is again ambiguous, de- and spread of tariff rates. The net effect is to shift pending on workers' consumption propensities. the domestic relative price in favor of exportables The real wage in terms of exportables and and/or unprotected importables and to lead to resource reallocations accordingly. Figure 3.5 PRODUCr MARKEr EFFECrs. The mesoeconomic ef- wt w. fects of trade liberalization can be readily ana- lyzed using the Dornbusch-Collier model out- lined earlier. In this case, PJP, is raised through a discretionary policy that reduces P . This will re- sult in a decline in P:, assuming that nontradables VMP, VMP and importables are gross substitutes and that P, VMP \ is sufficiently flexible. Referring back to Figure 2.3, assume that the economy is in equilibrium at \ F (with the equilibrium loci being N**N** and wI L**L*). A trade liberalization would take the economy to a point such as D, thus creating ex- wz cess supplies of both money and nontradables. P _ would therefore fall to clear the nontradables Ot L L L L, O, L1 2z Ls L3 L4 29 nontradables will certainly decline, but it will Figure 3.6 rise in terms of importables (the fall in the wage being less than the decline in P,). Trade liberali- W, 'W zation will therefore cause labor transfers out of importables and into exportables (where em- H ployment rises by L1L2) and nontradables (em- ployment increasing by L3L4). The implications for income distribution and poverty are likely to be AW. favorable - it is more likely that workers in the VMP/ favored sector (exportables) will be low-paid and VMPu VMP/ unskilled compared with those in importables. v \ If it is assumed that P is flexible downward then we know that with nontradables and importables, w\ gross substitutes, P. will fall, thus shifting VMP " downward. The qualitative results just described w,2 (that is, labor transfers out of importables into exportables and nontradables and an ambiguous l change in the real consumption wage) hold even L3 L4 L5 L2 L° when Pn declines, as long as the three goods are gross substitutes in consumption and production and that the income effect never exceeds the VMP, to VMPt,, which induces a fall in the free substitution effect (Edwards 1988a, p. 173). market wage (to W,2). This causes an increase in But suppose that the wage is inflexible down- employment in exportables (to OtL4) and a de- ward and remains at W1 following the liberaliza- crease in importables employment (to L4LQ). With tion. This would create unemployment, equal to no change in nontradables employment (we as- L,L4, and there would be no labor transfers into sume for simplicity that P. is fixed), search un- exportables and nontradables. Obviously, over employment rises. While the real consumption time, as capital moves out of importables into the wage in the fixed wage (nontradables) sector is X and N sectors, the labor demand curves will certain to increase (W., P2 and P. are unchanged shift, thus restoring a long-run equilibrium and and Pm has fallen), the real consumption wage full employment. But until that happens, wage effect is ambiguous elsewhere. Note that liberal- inflexibility has meant that liberalization results ization leads to a conflict of interest among work- in labor unemployment. This result prompted ing people - those in the protected sector gain, Edwards (1988a, p. 178) to observe that "this pos- while other workers may lose (or certainly do not sible short-run unemployment effect might call gain as much). for a second-best argument in favor of a gradual reduction in tariffs," the pace being dictated by An alternative framework: adjustment with the speed of capital transfers across the sectors. unemployment There are also arguments that suggest that if the pace of liberalization is too slow, the capital The theoretical interpretation of adjustment used market may not pick up the signals (Mussa 1986). above views the process as a movement along This suggests that there is an optimal pace of the production frontier, with resource transfers liberalization that would minimize the short-run between sectors as the main equilibrating unemployment costs. mechanism. This assumes that there is sufficient Now consider the case of partial wage inflexi- wage and price flexibility to ensure that the bility, with the nontradables sector hiring labor economy remains at full employment and that in the formal (fix-wage) labor market. Assume a there are no "structural" rigidities that prevent fixed wage of VW in the sector, restricting em- the smooth flow of resources between the sec- ployment to O,L1 (Figure 3.6). The initial equilib- tors. Once we allow for the price and other rigidi- rium is at y, with OL3 workers in exportables, L3L2 ties, the orthodox model becomes less convinc- in importables, and L2L1 unemployed (while ing, since switching effects may no longer be searching for protected jobs in nontradables). The reliable. The main equilibrating mechanisms of decline in P, following import liberalization shifts the "structuralist" school are changes in aggre- 30 gate output and in income distribution (Taylor supply in the market) will be obliged to reduce 1983, 1988). It is not our intention here to review output and employment The effective demand the range of these models, but rather to focus on for labor will therefore be constrained by the the Keynesian model, which involves adjustments rationing of firms in the product market. The net in output. To avoid cluttering the mind, our effect of such rationing in labor and nontraded preference is to remain within the dependent product markets might be a continuing economy class of models, in which the funda- "disequilibrium" in which unemployment per- mental distinction between tradables and sists. Given the constraint faced by households in nontradables is retainedYDixit (1978), Cudding- selling labor services, their effective demands for ton (1980, 1981), Cuddington et al. (1984), and nontradables will be lowered. Faced with this Neary (1980) have applied the disequilibrium low level of effective demand, firms would be framework of Barro and Grossman (1971) to the obliged to reduce output. This level of output dependent economy case. would then lead to a constrained demand for In the short-run version of the basic model, labor, limiting household incomes and demands, three markets are analyzed-two product mar- and so on. The economy would finally settle at a kets (tradables and nontradables) and the labor point where the effective demands and supplies in market. Given the small country assumption, the two markets (labor and nontradables) are producers and consumers of tradables do not equal. But this temporary equilibrium is consistent face any quantity constraints in their product with the persistence of nonzero excess notional market because they can sell or buy as much as demands in any of the two markets. they wish in the world market at the given price. Assume that both Pn and W are set above their Although (under fixed exchange rates) the do- market-clearing levels, so that excess supply mestic price of tradables is fixed, no quantity pervades both markets, and Keynesian unem- rations are imposed in the tradables market. ployment is said to exist. What are the effects of Two main price inflexibilities are introduced adjustment under such rationed regimes? It into this framework - P, and the money wage should be remembered that with unemployment will be assumed to be fixed, so that market clear- in the economy, adjustment need not require a ing can no longer be guaranteed. Because reduction in aggregate absorption to reduce the nontradables are by definition only produced and trade deficit. In principle, this can be achieved consumed domestically, the fixed price (Ps) will through devaluation alone, since the slack in the lead to either buyers or sellers being rationed at economy creates room for tradable output to ex- the short end of the market. Moreover, rationing pand and to correct the external deficit. Before in one market will lead to further rationing in considering the meso effects of devaluation, other markets, so that the firms who are unable however, we shall trace briefly the effects of a to sell what they wish in the N-product market policy of expenditure reduction. will reduce their labor demand accordingly. In the labor market, therefore, demnand will be be- Fiscal policy low its notional level simply because of a ration in the product market. Similarly, if the wage is Since the levels of tradables output (Y,) and em- set too high, an excess supply of labor would ployment (L) are determnined by the firms' profit- ration households seeking to sell their labor serv- maximizing behavior, aggregate demand man- ices. They would only be able to sell what the agement policies will have no effect on the sector. firms are willing to hire. Thus, actual employ- This applies even if the government reduces ment would be determined by the demand for spending on tradables - firms will simply direct labor, and unemployment would persist. more of their output to foreign demand. The same The central proposition of these models is that is not true, however, for nontradables. Decreased the failure of any market to clear will spill over government expenditure on nontradables (as part into other markets. Thus, if there is excess supply of a fiscal and monetary contraction) will reduce in the labor market, households constrained in output in the sector according to the famniliar the amount of labor they wish to sell will reduce multiplier process, their demnands for commodities, thus affecting product markets. In the same way, firms con- dYJdG, = 1/(1-(dDJIdY)PJ,} > 0 strained in a product market (because of excess 31 where G is government spending, D the level of ployment model to trace the primary distributive demand, and the denominator equals one minus effects of devaluation no longer applies. With the marginal propensity to spend on nontraded output in both sectors increasing, and with un- goods. A general fiscal contraction - involving, employment falling, the net effect on incomes is for example, a decline in government spending quite different. on both tradables and nontradables - will have As W and P. are fixed (by assumption) and P, is unambiguously beneficial effects on the trade raised through the devaluation, the real con- balance. The reduced spending on tradables will sumptionm wage certainly declines. This induces reduce imports of tradables dollar-for-dollar. The the increase in employment in both T and N sec- decline in spending on nontradables will reduce tors. So in contrast to the orthodox equilibrium private sector demand for imports because of the model above, the real-wage effect is unambigu- decline in employment and incomes generated ously adverse in the presence of unemployment. by the nontradables sector. The net effect on labor's real-income share, there- The effects, then, of fiscal contraction under the fore, will depend on the relation between the cut assumptions of this fix-price, Keynesian two-sec- in the real consumption wage and the increase in tor model are as follows: the trade balance will employment. This, in turn, depends on the real- improve and incomes and employment in wage elasticity of demand for labor. Even if it tradables will remain unchanged; although the were shown that labor's share falls as a result of real consumption wage in nontradables remains devaluation (with an increase in employment in- constant, the level of employment falls. General sufficient to counteract the effect of the real-wage fiscal contraction will not typically lead to an decline), there can be no presumption that poverty across-the-board decline in incomes. It will leave has increased. some incomes largely unaffected but reduce the The beneficial employment effect predicted by incomes of others drastically. When it comes to this simple Keynesian model of adjustment arises the poverty effects of expenditure reduction, from the demonstration that a devaluation is ex- therefore, it is critical whether the poor are en- pansionary - it raises the level of aggregate gaged in sectors that are sensitive to reductions economic activity by raising demand for tradable in aggregate demand. and nontradable goods. Nevertheless, there are reasons to suppose that this may not always be Devaluation true, that a devaluation may turn out to be contractionary. If devaluation shifts income dis- Devaluation will lower the product wage in the tribution toward high savers, if imported inter- tradables sector and therefore increase Y, unam- mediate inputs figure prominently in production biguously. This has a directly beneficial effect on costs, and if the general price rise following a the trade imbalance (Y, - D,). But the increased devaluation reduces real money balances suffi- incomes generated (through increased employ- ciently, aggregate demand may decline. If this is ment and enhanced profits in the tradables sector) the case, devaluation offers very little for the will raise the levels of demand D, and D.. The in- working population - a decline in real con- crease in D, will have an adverse effect on the sumption wages and an increase in unemploy- trade imbalance, while the change in D. will ment. stimulate the production of nontradables. This is clearly an empirical matter, but there Two important differences from the basic model are specific reasons for anticipating expansion- therefore emerge in the case of Keynesian unem- ary effects in Africa. This is because of the adverse ployment. First, a devaluation will not necessar- output effects of the extreme shortage of foreign ily improve the balance of trade; the net effect exchange in the pre-devaluation period, during will depend on the relative magnitude of the which the exchange rate is misaligned. Purchas- different price and income effects. Second, the ers find themselves severely rationed in the for- increase in Y, is not at the expense of Y,, as in the eign exchange markets, which leads to rationed full employment model. Under Keynesian un- supplies of imported goods. If intermediate inputs employment, both Y, and Yn increase as a result are unavailable, there are adverse direct effects of a devaluation; this is made possible by a re- on output. But when imported consumer goods duction in the level of unemployment.- The neat become rationed (or simply unavailable), there theoretical reasoning employed in the full em- can be adverse indirect effects on output, because 32 producers are denied access to "incentive" goods The growth factor (Bevan et al. 1987). As devaluation removes this rationing in the foreign exchange market, it re- In this section we develop our story further by stores the needed supplies of intermediate and examining more closely aspects of recent adjust- "incentive" consumer goods, facilitating an in- ment experiences in Africa. We shall highlight crease in output. As Taylor (1988, p. 29) puts it, "a some important features that arise from the pre- scenario along these lines helps explain part of dominance of import compression in many Afri- the success of recent forex-intensive Bank/Fund can countries. programmes in Sub-Saharan Africa." An import-compressed economy is character- ized by poor output performance, high inflation, Other supply and institutional rigidities and extensive shortages. It must be emphasized that the preadjustment situation in the majority The Keynesian model we have just considered of African countries is almost never character- embraces two sources of rigidity - fixed Pn and ized by satisfactory growth in output or employ- W. The result is that policies have their effects ment. The overall picture is not one of fast- mainly through quantity adjustments. Once growing economies going into recession because relative prices change in the model (for example, of a balance of payments and financial crisis and a depreciation in the real exchange rate), output the application of stabilization measures. Donor- is assumed to respond smoothly. In some respects, supported programs with their associated finance therefore, the Keynesian model does not go far can raise total output, even in their early stages. enough in dealing with two distinctive charac- Aside from attracting increased bilateral and teristics of many African countries that influence multilateral aid, policy reforms will increase the this story - the slow supply response to relative confidence of commercial lenders. The injection of price signals and the institutional rigidities that foreign exchange from both concessional and can influence the mesoeconomic outcomes of commercial sources will raise capacity utilization policy intervention. where imports of intermediate inputs have pre- Rigidities that typically occur in African coun- viously been compressed. The distribution of this tries arise from three broad categories: extra foreign exchange will vary depending on * price rigidities, which are the result of pub- the sectoral priorities of the adjustment program, lic sector interventions in markets that lead to the and the recovery in capacity utilization will be emergence of parallel markets alongside official uneven across sectors. In most donor-supported markets; to deal with such structural characteris- programs agriculture and its supporting infra- tics, a fix-flex-price model (after Taylor 1983) structure, as well as factories producing agricul- would be appropriate tural inputs, are given priority. With the begin- * institutional rigidities, which go beyond the role that parastatal organizations play in the Figure 3.7 markets; these would include the prevalence of the household enterprise, in which production impoitab3' and consumption decisions are closely related (this is discussed in further detail below); as an illustration of this, the behavior of risk-averse smallholders might be quite different from "orthodox" enterprises if they retreat into subsis- tence production * physical rigidities, which constrain supply and the capacity of smallholders to respond to market signals; we have already mentioned the 1 importance of infrastructure in this context - if enterprises are not well served by economic in- frastructure (especially physical infrastructure), their capacity to raise output may be severely limited. XiX' XI Exp tk 33 ning of the recovery in capacity utilization, out- the production frontier (in Figure 3.1) toward put growth may occur in the initial stages of tradables. adjustment. The time taken to achieve the pre- Figure 3.8 introduces the growth effect. Again import-compression level of output will depend assume that preadjustment expenditure is at y on how far capacity utilization had previously while the production point is x on N*TP, thus sunk and the magnitude of the foreign exchange yielding a trade deficit of xy. The real exchange injection.9 rate PJP, is given by AA'. Our previous analysis In their early stages donor-supported programs showed that a program of demand deflation plus also begin to reform inappropriate policies that devaluation depreciates the real exchange rate to have caused the misallocation of resources - BB' giving a new production point z and the reducing the gap between domestic and world elimination of the trade deficit. The output of prices for export crops is one example. Figure 3.7 nontradables falls to N2 while that of tradables illustrates the case where domestic relative prices rises to T2. If at N*7* productive capacity is are set to discriminate against exportables. As- underutilized,.4 an injection of foreign exchange, sume that the world relative price is given by the plus an improvement in resource allocation price line I, while the imposition of import con- through the removal of distortions, would move trols (or tariffs) shifts the domestic price to I. the production frontier to N**T** in the short term Domestic output is at P2. so that exportables out- - say the first year. Policymakers could then put valued at world prices is OX2. Removing these include this supply expansion in their calcula- distortions shifts the economy to P,, and there is tions and would accordingly need only to cut an unambiguous gain in the output of exportables absorption to OC rather than OB in the case of no to OX, valued in world prices (Kanbur 1987c). output growth. Similarly, the size of the nominal This gain will contribute to the increase in total devaluation will be less. This adjustment pack- output in the early stages of the program. If this age would yield a real exchange rate depreciation growth process is achieved during the first years of CC' (which is less than the depreciation to BB' of the program, some of the social costs can be under no growth), and the new production point reduced. Our earlier discussion presented a rather would be at v. Now compare the preadjustment static picture of the adjustment process and did outputconfigurationatxwiththepost-adjustment not fully capture the dynamics of an economy configuration at v. Total output is higher at v recovering from import compression, because at- (since the economy is now on N**Tz*), and the tention was directed to the movement around output of both tradables and nontradables is higher, although because nontradables have in- creased by less than the rise in tradables, the share Figure 3.8 of nontradables in national output has fallen. The deficit has been eliminated and the structure of Non- A output shifted to tradables. We do not suggest tradables that these movements will be smooth, because c only some of the rise in tradable output can come B \ \ from the restoration of capacity utilization. The issue of transitional unemployment raised in the NW \ \ previous section still applies. It follows from this discussion that if growth \ \v \ occurs in the manner described - that is, "tradable- N* \ \ \ led growth" -then there are four important points for the social costs of adjustment: N1 -- * the impact of adjustment on households re- N2 - - - - - - - - - Wz)\ \ \ A ceiving incomes from nontradable activities will be less than in the no-growth situation because C \ \ c there need now be no absolute fall in the output B' of nontradables; but, overall, income distribution O T, T2 T' T Tradables will still shift toward tradable producers and away T, '- T" Tradabies from nontradable producers 34 * transitional unemployment will be less than exportables can only be removed once, so the in the no-growth situation, because sellers of la- initial increase in output that their removal cre- bor will find a better market with the economy ates is also a one-time gain (although it may take growing several years for the effect on output to be com- * because the real exchange rate depreciation pletely realized). is less under the growth scenario, the "knock-on" These reforms do, however, provide a policy effect into higher nominal prices will be smaller environment that is conducive to productive in- than in the no-growth situation - so households vestment, so in that regard their effects are last- hit by the rise in nominal prices due to the in- ing in the second phase of the adjustment process, crease in P, will suffer less of an effect planners will begin to engage in further policy * because the required cut in absorption is less reforms, particularly major trade and financial under the growth scenario, the government will liberalization. But the stock of policy reforms that have more leeway to protect social expenditures; can be undertaken will gradually diminish (pro- social services are nontradables and a degree of vided the adjustment program stays on course). expansion in their facilities may be possible since Consequently, while some growth - perhaps some growth in nontradable output is viable un- sizable - can be achieved in the early stages of der tradable-led growth. the adjustment program when starting out from It must be emphasized that these benefits only a base of capacity underutilization, sustained occur because the growth process engendered by growth over the medium to long term depends the adjustment package and donor support is on the magnitude of the new investments made based on tradables taking a larger share of (grow- and their rates of return. The evidence reviewed ing) output. In Chapter 2 we saw that the growth above, however, would throw some doubt on the process based on a rising share of nontradables growth and investment performance of many associated with macro destabilization was not African countries undertaking structural adjust- sustainable. Under the present scenario, tradable- ment. Why many of the countries reviewed in the based growth is sustainable because of the ac- World Bank studies that were mentioned exhib- companying policy shifts and finance. The growth ited such poor growth performance indicators is in output engendered by the foreign exchange beyond the scope of this paper. It must be pointed injection and the restoration of capacity utilization out, however, that most countries implementing is a one-off effect - full capacity utilization can structural adjustment programs in Africa have be recovered, but further large increases in output been faced with multiple shocks, which makes will not occur until new investments make their growth an increasingly difficult policy objective effects felt. Similarly, policy distortions against to attain. 35 4 The determinants of welfare Having established in principle the main the value of the goods that she produces for her mesoeconomic effects of adjustment, we must own consumption. In the next stage, goods and now turn to consider the micro economy - that services are purchased for consumption, and some is, the households at the center of the social di- income is saved for investment. mensions. Before developing a choice-theoretic These consumption and investment decisions, model of household behavior, we shall begin with in turn, feed back into her physical performance. a much broader perspective. Our purpose here is The first feedback (indicated by the broken line to establish the most fundamental relationships in Figure 4.1) is that between consumption and in the determination of individual and house- physical performance. She needs goods and ser- hold welfare. We proceed in stages, beginning vices for their "material characteristics" - for with a careful examination of the individual, and example the calories, proteins, and the like of moving on to consider household and then food; the warmth of clothing, and so forth. These intra-household and inter-household interactions and elements, in combination with the personal char- activities. acteristics of the individual (for example, her me- tabolism), determine her physical performance. Individual welfare That performance, or "functioning," (Sen 1987) has many dimensions. Most obviously the level The "well-being' or "welfare" of individuals in of nutrition is critical in determining physical African societies is the outcome of complex eco- performance, although there is much debate over nomic and social processes. In this section we the exact relationship between, for instance, examine the relationships among physical per- calorie intake and human energy (see Pacey and formance, labor input, income generation, and Payne 1985, chapter 3). The ability to purchase the consumption-investment decision. Our dis- health services is also crucial. The interactions cussion is based on Figure 4.1, which sets out the among consumption, nutrition, health, and work main links between these variables in the form of performance are likely to be critical, because there a flow diagram and describes a single individual may be significant productivity gains from im- who is able-bodied and working. We start with proving the well-being of the poor. Thus "pov- her physical performance, which determines the erty alleviation" may serve to further the aims of amount of work she is capable of undertaking in the structural adjustment program itself, and ex- any given period. Ignoring for the moment the penditures that raise the nutritional levels of sec- role of assets, we can see from Figure 4.1 that she tions of the population should be considered as obtains an income, which may be in the form of investments that will yield favorable productiv- wages or receipts from the sale of goods and ity outcomes in the future. For much of rural services. An income can also be "imputed" for Africa it is likely that the distinction between 36 Figure 4.1 The determination of an individual's welfare l J Consumption Income from: Own-production Physic~al _ Labor Marketed output performance ,/ Wage-employment / fb . : $ Savin~~~~~~~~~gs | _ _ _I_*_I I Assets: 'multipliers' Assets: 'adders" Human capital Financial assets l Physical capital Durable goods Legend: - - Feedbacks - Returns on assets "consumption" and "investment interventions" yield incomes but they do so in different ways. will be blurred. These relationships are certain to Human and physical capital increase the produc- be complex. For example, not all consumption tivity of the worker. In this way they multiply expenditures will improve nutrition, so that in- the return that would be generated by labor alone. creases in income and consumption will not nec- Returns are usually ascribed to human and real essarily improve nutrition. This will obviously producer assets (their marginal products) and depend on how consumption is allocated among valued independent of labor. But actually these alternative food and nonfood goods. returns cannot in most cases arise independently The second feedback - between savings and of labor, while labor can still generate a return the process of income generation - is shown in independent of access to many assets. We there- the bottom half of Figure 4.1. Savings are used to fore follow Lipton (1985, p. 1) in describing hu- acquire assets. Personal savings are among the man and real producer assets as "multipliers" of principal determinants of asset acquisition in most labor productivity. Financial assets such as sav- low-income countries. Formal credit markets are ings accounts, as well as stocks and bonds, also thin, and access is generally confined to higher- yield incomes in the form of interest and capital income - often urban - groups. Poorer people gains. Some durable goods, such as housing, must resort to informal credit markets, including yield rent incomes. Owner-occupied housing is a borrowing from their kin. Consequently, asset major source of unearned income in developing acquisition through borrowing is not shown in countries, and rents must be imputed (see Figure 4.1, in order to focus on the savings-in- Grootaert 1982a, p. 19). Lipton terms such addi- vestment decision. tions to the income generated by work as Assets can be classified into four kinds: human "adders." capital, physical capital (which can include Figure 4.1 shows the returns to these different physical infrastructure services), durable goods types of assets. Of course the distinction between (for example, a house), and financial assets. These income-adding and income-multiplying assets is 37 not watertight. Durable goods can also be said to is possible for families to be spread among sev- play some role in multiplying labor productivity eral households, either temporarily or perma- - for example, work efficiency will eventually nently. For example, a married woman while fall if someone is homeless. Owners of physical young may continue to live in her father's house- assets may choose to rent them to others rather hold, while her husband lives under a separate than use them directly. In such cases the owner roof. adds the rent to her income stream while the The household is an important social unit be- borrower uses the asset as a productive input. It cause within it many of the decisions concerning is therefore the relative importance that a par- individual members' activities and their con- ticular asset plays in either directly adding to sumption (and thus their welfare) are made and income or multiplying labor productivity that is its physical properties - that it is a collection of the determinant of our two-fold classification. individuals with an identifiable location - makes As we noted above, economic and social infra- it a useful sample unit in survey work. It must be structure are important elements of the emphasized, however, that households are em- meso economy; they transmit the effects of mac- bedded in wider social networks, their lineage roeconomic policies to individuals. group for example, whose actions partly deter- Although our exposition of Figure 4.1 has pro- mine their members' welfare. Given the impor- ceeded in linear fashion from the individual's tance of the household as a decisiomnaking unit, performance to consumption and savings, we can we need a conceptual framework to analyze its see that, with the addition of the feedbacks, a decisions over the allocation of resources. Two person's physical performance exists within a key issues are raised in the analysis of the house- system, and both determines and is determined hold. The first is the role of the household as both by the system. Obviously we have presented a the producing and consuming institutional unit. stylized picture of the individual's circumstances Whereas in much of orthodox economic theory to capture some of the most basic elements of her the firm is assumed to be the producing unit and economic life. But of course the individual's wel- the household the consuming unit, quite differ- fare is not just determined by her own actions ent institutional arrangements must be assumed and resources. Most people take part in social units for developing countries. This is especially the (or networks) in which decisions about produc- case in Africa given the predominance of agricul- tion and consumption are made, and which act tural activities in total employment and the lim- as conduits for the transfer of resources between ited share of formal employment in most coun- individuals. Such participation can enhance (or tries. diminish) the welfare outcomes for the individual. The second issue that has to be addressed con- It is to these interactions that we now turn. cerns how household decisions are made - are they reached collectively or does one individual The economy of the household or group dominate the process? A related issue is whether we can correctly speak of a "house- People are usually located in several overlapping hold welfare function," since there may be con- social networks at the same time. The nuclear flicts of interest within the household. In theo- and extended families are obviously two such retical work, individuals are aggregated into social units. The household is another. A num- households on the assumption that they possess ber of criteria can be used to define the house- identical preferences based on identical tastes hold. Those commonly employed include: (Deaton and Muellbauer 1980). Household deci- members have a common source of major in- sions are then analyzed in the same way as those come; they share a common source of food; and of a single individual. Why people should group they sleep under the same roof or within the themselves in a household is usually analyzed as same compound (Casley and Lury 1987 p. 163). a secondary problem, but it is generally assumed But the criteria used to identify households must that they make up a family. Sen (1983a) calls this be relevant to the local situation, since their size arrangement the "glued-together family."41 Al- and characteristics show wide variations by prin- ternatively a "despotic family" is one in which cipal occupation, locality, and country. The the head of the family takes all the decisions, so household may consist of a single family, but in that family behavior is simply a reflection of the Africa they commonly comprise several families, head's choice function. These are polar cases - kin, and even persons with no kin relationship. It in the former, members of the household are as- 38 sumed to share the same preferences; in the lat- in the nutrition of all household members. This ter, the preferences of the household head alone may be related to the respective positions of men are relevant. and women in households. Major problems exist, however, in using either Taking a disaggregated view of the household the concept of the "glued-together" or the "des- allows us to focus on intra-household transfers as potic" family.4A Preferences, particularly those important sources of individual welfare. Some that arise from age and sex differences, can differ individuals are effectively subsidized by "tax- widely among family members, so that they will ing" other household members. At the extreme allocate family resources in different ways. The ends of the life cycle - infancy and old age - eventual allocation of resources will differ, per- those transfers may represent a person's only haps substantially, from that under "glued-to- access to resources. Furthermore, in any one gether" or "despotic" families. These difficulties period some household members will be less apply with equal force to the unit of the house- productive through ill-health or pregnancy. But hold because large numbers of people can be not all transfers are made to safeguard the involved in decisions about its collective re- recipient's minimum needs, and some individu- sources.43 In such circumstances, assuming a als may command larger household "subsidies" single-household utility function is even less valid through their bargaining power. than making such an assumption for a single In Figure 4.2 (overleaf) we depict as a flow family unit. chart some of the components of the household To get around these difficulties, two approaches economy. This is an elaboration of the single- may be taken. In Becker's (1981) model of the person economy of Figure 4.1. The physical per- "super-trader family," members are assumed to formance of those able to work determines their maximize their individual utilities without regard labor effort, which, together with the returns on to social norms. They trade with each other at the household's assets, generates income (either implicit prices, and this determines the allocation in the form of wages or market sales; we impute a of resources and such important decisions as value for own-production). That income is effec- marriage. This model could be transferred to the tively taxed to provide transfers within the larger unit of the household. Nevertheless, while household, and the remainder is distributed for these trades do occur within families, and more the consumption of the workers and for savings. commonly in larger household units, we concur Such "household-taxation" may take different with Sen that to focus on such trades alone - and forms. For example, in households that are en- to assume that all the stringent conditions for tirely self-provisioning, the harvest may be di- market equilibrium are met despite the absence vided, with some of the working members get- of actual prices - is an unsatisfactory starting ting less food than others depending on their point for analyzing intra-family and intra- "tax rate." Alternatively wages and the income household behavior. from sales of produce may be pooled and then A second approach is to assume that family redistributed or used to buy consumption goods and household relations contain both coopera- that are shared, but not in proportion to the cash tive and conflicting elements, which can yield contributions of the household members. Not all many different arrangements, depending on the income may be pooled - an individual may re- bargaining strengths of the individuals involved tain some or all of the income she earns, but (Sen 1983a, p. 375; Schultz 1989, p. 10-17). This nevertheless receive transfers from within the approach does not deny that social norms play a household. largerole in determiningthe "space" within which A final and crucial dimension must be added such bargaining takes place. For example, women to our household framework. Time and other have a lower social status than men in many household resources must be allocated to the countries and this limits their bargaining power. bearing and rearing of children, to the provision Taking a bargaining perspective of the house- of health services, to the education of household hold allows us to focus directly on the inequali- members (for example, in cultivation), and to the ties that can be present within that unit and to daily tasks of household maintenance such as explain some important phenomena. For ex- cooking, cleaning, washing clothes, carrying wa- ample, in certain situations we may observe an ter, firewood collection, and house building and increase in household income (perhaps under repair." As with the production of other goods adjustment) that does not yield an improvement and services, the sustained production of these 39 Figure 4.2 The household economy I Net external Transfers transfers to within the the household household Consumption Income from: Own-production J Phsicl J La'bor 7* Marketed output performance Wage-employment LwJ / i 4 Savings rearid Assets: 'multipliers' Assets: 'adders' Family Human capital Financial assets l health Physical capital Durable goods ctare , I Legend: - - Feedbacks _ _. - Returns on assets items requires that the household's physical per- als are to varying degrees responsible for pro- formance be adequately maintained. Therefore, duction and income and expenditure decisions. in Figure 4.2 physical performance determines At present, such a model does not exist, although childrearing and the provision of household ser- there is a growing awareness of the need for an vices, and these feed back into physical perfor- initiative in this direction (Fapohunda 1988). The mance so that we effectively have a "closed loop" conceptual presentation that follows therefore in that part of the flow diagram. The burden of takes the glued-together (or despotic) household these home tasks usually falls disproportionately as its starting point. on females, with often deleterious consequences Our previous discussion highlighted the dual for their welfare. We return to the "gender-divi- role of the African household as both a unit of sion of labor" in the next section. consumption and a unit of production. Not all households fulfill this dual role in Africa, but a Analyzing household behavior majority of them do, and this creates special problems for predicting the consequences of Faced with a multiplicity of household decisions, policies. Within the majority of African house- we need a suitable framework for their analysis if holds there is an interdependence between con- policy effects are to be correctly understood. sumption and production decisions, so that deci- Within such a framework, different models of sions on output also directly affect consumption household decisionmaking can be analyzed con- and labor supply, and vice versa. Changes in the sistently, and cause and effect can be assessed in parameters determining one aspect of household a quantitative manner. The question therefore behavior also affect others. For example, a change arises of what "choice-theoretic" framework is in the market wage will not only affect a appropriate in the African context. Ideally, what household's labor supply, but may also influence is strictly required is a model of heterogenous its labor demand (because the household is a nonpooling household units in which individu- producing unit). Moreover, the level of home 40 production will be affected (because the retums Figure 43 on such an activity relative to market work will CoodY change), together with the level and pattern of household consumption. A similar chain of in- terrelated effects could be specified for other policy shifts such as alterations in consumer and ------------- b producer prices. All these effects must be identi- Mu _2 y fied if a comprehensive assessment is to be made -Y ..; of the welfare impact of policy changes. These different facets of the typical household C are well recognized by the new class of models of agricultural households.'5 But the arguments ° / apply with equal force to other household types, for example those in the urban informal sector. Much of the work in this area has been undertaken in the Asian context, a region that differs from o L4 Li 14 L Hcwd Africa in several important respects, including l I. L. *N(L relative factor scarcities and tenurial arrange pnMdku wok How3k ments. But the studies of Africa undertaken thus far have already yielded important insights: for household can either produce or purchase a com- example, Low's (1986) analysis of food security modity - labeled Y on the vertical axis of Figure in Southern Africa, Smnith and Strauss' (1986) work 4.3. The horizontal axis rneasures household time. on nutrition in Sierra Leone, and Braverman and The equilibrium of the household in production is Hammer's (1986) analysis of pricing policies in given at point a, where W/P (which is the real Senegal, to name a few. With these concerns in wage) is tangent to the household production mind we proceed to outline a simple household function (QY)." The household produces Y1 units model. The analysis is then extended to consider of output using L1 units of its time. The the distribution of welfare within the household. household's equilibrium in consumption is given We should repeat the sentiment expressed above at point b, where W/P is tangent to the household about the purpose of simplified or model repre- indifference curve. We see that the household is sentations of the real world: these should be an willing to devote L2-L1 units of its time to market aid to our thinking on the issues, and not an work in order to obtain Y2-Y, units of good Y intellectual straightjacket. through market purchases. The household has a maximum of L. units of time at its disposal, and A basic household model so the remaining time, L,-LZ is free for other ac- tivities. This segment could be labeled 'leisure," The model presented here is a simple version of but since the household has to produce nonpur- the Barnum and Squire (1979) and Singh, Squire, chased goods and services (Z goods such as and Strauss (1986a, 1986b) class of models. The childcare, food preparation, and so forth), we basic model assumes that household utility is label this segment "home work," although it will maximized subject to a production function, a inevitably contain some pure leisure time as well." time constraint, and an income constraint. The In this simple model the household engages in time of household members can be allocated to three sets of decisions concerning: the household production of goods (for sale or * allocation of its members' time among the their own consumption), to home work (con- production of Y, market work, and home work cemed with household maintenance and repro- (which produces Z goods) duction), to (labor) market work, and to leisure. * allocation of its nonlabor factors of production These models therefore take a "full income" among these activities (Becker 1965) approach. * allocation of its consumption between market Consider the simplest case in which the house- purchases of Y and home production of Y. hold has a single utility function (equivalent to One important feature of this model is the de- the "despotic" or "glued-together" household, cision by households to buy or sell labor services. discussed earlier). Further assume that the In the situation described, the household is as- 41 sumed to sell its labor services to finance con- Figure 4.4 sumption in excess of its own production. Alter- natively, the household may be consuming at a GCoodYi point such as c (indicating a stronger preference for leisure), in which case it must be buying-in labor services. In this case, the production point IY remains at a, but the household is seen to hire in labor services (L3-L,) in producing its output, thus Y - . a generating for itself more available time for lei- sure and home work (L3-L.). It is immediately apparent that all these decisions will be revised if wages or prices change in the markets facing the household, an obvious key point for analyzing the welfare impact of adjustment. Analytical models in the African context L aj Household time (L) One major determinant of model structures ap- propriate to Africa is the characteristics of the valuation of the household's labor time is given markets in which households engage. In the by the market wage. A household that can nei- model presented above it is implicitly assumed ther sell nor hire labor will still value its labor that households engage in "complete markets" time, but this will be done subjectively, and this for goods and labor services, so that household valuation will accordingly vary across house- decisions are made with reference to a set of holds. This situation is depicted in Figure 4.4. exogenous prices.4 Whether this condition is met Again we have a household production function or not has important implications for modeling (OY), and household income equals its output the functioning of the household economy, since because there is no opportunity for market work. the structures of such models are dependent on The household can still sell its product or con- the assumptions made conceming the markets sume it entirely. The equilibrium point for this that households face. The basic model described household is at point a, where the marginal rate in Figure 4.3 depicts the main features of a class of substitution of leisure for income (the "subjec- of models that has been developed for analyzing tive wage") is equal to the marginal product of households that face complete goods and factor labor (Ellis 1988, p. 110). Since preferences, and markets. These models are recursive in character. thus the shape and position of indifference curves, First the household sets its level of output by the differ across households, the marginal product of maximization procedure that we described above. labor will vary.49 To do this it needs information on the price of Similarly, the presence of a complete goods output, the wage rate, and technological relation- market allows the amount of output produced to ships (described by its production function) be- be determined independently of the amount con- tween inputs and outputs. In the structure of sumed since the household can always buy or these models, the household's production deci- sell the good at the market price. When neither sion is made separately from its consumption and factor markets nor goods markets exist, the labor supply decisions (Singh, Squire, and Strauss household must make all its producer, consumer, 1986b, p. 7). Consumption and labor supply de- and labor decisions simultaneously - it can only cisions, however, are dependent on production consume what it produces and use its own labor decisions because the latter determine household (Singh, Squire, Strauss 1986b, p. 6). Consequently, profits, which are a component of income and in models describing such households, the pro- therefore affect consumption and labor supply. duction solution cannot be made independently The model is recursive in character because the of the consumption solution (production is production equations are solved first and the re- nonseparable): the model must be solved simulta- sulting solutions are fed into the household's neously rather than recursively. consumption and labor supply equations. Separability also breaks down when markets When a complete labor market exists, the are absent for some important goods that may be 42 produced within the household. For example, if assessed in making judgements about the effect any of the commodities consumed or used as of changing market wages on the household. It inputs into household health production are ab- would also be necessary to respecify any migra- sent, then to attain the desired level of health tion function that contains origin and destination production the household cannot rely on buying- income earning opportunities for such house- in the required amount of health inputs, and must holds, since their perception of these opportunities divert labor resources away from home produc- will be affected by their lack of complete market tion activities to produce the nonmarketed com- access. The returns to information gathering for modities (Pitt and Rosenzweig 1986, p. 158). The such households are likely to be significantly production of marketed commodities thus be- higher than for those in constant communication comes dependent on consumption (of health with the market on a day-to-day basis. goods). This is an important point if household welfare is to be analyzed in all its dimensions. The distribution of welfare within households In short, it is obvious that great care must be taken in the assumptions made about market The model presented in the previous section as- completeness if analysis is to be appropriate to sumes that a single household utility function the African context. Aside from incomplete mar- exists - in other words, the multiple-person kets for labor services and goods (including health household is analyzed in the same manner as a goods), the absence of markets to cover risks (in- single-person household. Our previous discus- surance) and imperfect capital and land markets sion, however, emphasized the importance of must be noted as a general feature of Africa. different preferences among household members These issues apply with special force to the situ- and differences in their circumstances caused by ations of small farmers, a key policy target group. both social and economic factors. This, as we Much market imperfection is related to low lev- noted, applied with particular force to the gender els of infrastructural development, particularly division of household labor, especially in Africa. transfer and communications, which aside from We therefore need to extend our household model their vital role of distributing goods and services to provide some insight into this. bring buyers and sellers together and establish the information flows that are the basis for all The gender division of labor markets. Many of the services that provide the essential, but often overlooked, context for proper Of special interest in the African context is the market functioning are "public goods," provided gender division of labor, which is frequently ob- by government rather than private markets served in rural farming communities and has themselves (Stiglitz 1988, p. 98). The depletion of been the subject of much recent research. In infrastructure, which is a common feature of many general, women tend to be relatively specialized countries, has retarded the market participation in food crops rather than cash crops.50 This divi- of households. Participation has also fallen prior sion is particularly prominent in the forest areas to current adjustment efforts because inappropri- of West and Central Africa: men provide 80 ate policies have reduced the benefits of market percent of the labor for commercial tree crops, participation and increased its costs. but on average only 9 percent of the labor for the Thus the context in which analysis is to be root food crops (Guyer 1986, p. 396). This spe- undertaken is changing, and analytical develop- cialization reflects that while women do some ments must be sensitive to this. In the first place, work in cash crop fields, their work on food crops, of course, the model underpinning any empirical whether in jointly or separately worked fields, work can no longer be recursive, so that produc- takes up most of their time. tion and consumption decisions must be simulta- In Figure 4.5 (overleaf), we drop the assump- neously determined. The household may not tion that the household has a single utility func- have any access at all to labor markets, in which tion in order to consider a two-person household, case earnings functions of the kind usually esti- consisting of a man and a woman, each with a mated in household models become irrelevant. specific utility function. In this model each per- If households have partial access to a market (for son is responsible for producing a different good.51 example, if casual work is offered only sporadi- The left-hand vertical axis measures the quantity cally), some probability of such work needs to be of good Y, which is produced by the man only, 43 while the quantity of good X, which is produced males (Birdsall and McGreevey 1983, p. 5), so it by the woman, is measured on the right-hand would be common for the man to take a high vertical axis (the output of both goods is meas- proportion of L.. - LM2 as leisure, with the ured in the same unit). This model can be taken woman taking most of L. - LW2 time for home to describe a farm household, but it could also work; thus, in total the woman would work longer apply to urban informal producers. The horizon- hours than the man.; tal axis measures the time input of the man and Figure 4.5 shows just one possible outcome the woman. We define a separate production among many for the division of labor by gender function for both individuals (OX for the woman in the household. The time dispositions of the and OY for the man). two sexes, their participation in market employ- Given their relative productivities and the real ment relative to self-employment, and their lev- wage that each can obtain in the labor market, the els of output are determined by the relative prices equilibrium of both persons in production and they face, their preferences, and their respective consumption are shown. The production equi- production functions. As in the basic household librium for the man is given by point A, while he model, adjustment can change all these outcomes consumes at point B. This gives him a time allo- by altering relative prices, production functions, cation of 0 - LM, for the home production of Y, and preferences. The net effects of adjustment LM2 - LM, for market work, and L. - LM2 for now become more complicated, for the prefer- home work and leisure. The woman's produc- ences, production functions, and applicable sets tion and consumption equlibiums are given A'and of relative prices facing men and women can B' respectively. Her time allocation is 0 - LWI for change by different amounts and in different di- the home production of good X, LW2 - LW1 for rections. Thus, when men and women face dif- market work, and L. - LW2 for homework and ferent sets of relative prices in goods and factor leisure. As shown, the woman would be respon- markets, adjustment can affect the gender divi- sible for more home work or could have more sion of labor within the household by changing leisure than the man. Time-use studies in Africa the structure of those relative prices. Similarly, (as well as for other developing regions) show uneven changes in the production functions of much longer working hours for females than for men and women can arise from policy revisions. It is thus possible that the welfare outcomes of Figure 4.5 adjustment for men and women may vary by IM degree, and sometimes in direction. Intra- Clo Y household transfers - which are not represented B GoodX in the model of Figure 4.5 - may change in a Y2_ D y compensating fashion, thus influencing the net outcome. A In the above model the gender division of labor Yl -X2 B_x2 is determined by the comparative advantages of (w/p)M the respective sexes; for example, males undertake / P)w>/; ^ " Xi more market work than females because their wage rate is higher (Gronau 1973). Social obli- gations and customs may predominate over in- dividual preferences, however, especially for women, so that outcomes may not correspond to 1/ I | , ' marginal utility principles in some (or many) cases (Ellis 1988, p. 181). This can affect all the seg- ments of a woman's time disposition. For ex- _~1 LW1 LW -+ ample, women may be precluded from engaging O LM1 LWI LW2 LM2 Lrnxx Hoin certain household production activities by so- Mans'timel X cial customs backed up by community sanctions. Home production Market Hogne work Other "barriers to entry' may be more indirect, ofY woerk orleaure for example when women lack access to a factor w*ffisanSe111 of production important in a particular product Hameprduction Market Honework line. These circumstances, rather than their of X work orl Isure comparative advantages, may dictate the respec- 44 tive specialization of men and women in the Mean household income as an indicator of welfare production of goods Y and X in the above ex- ample. Similar constraints can apply to women To what extent does knowledge of a household's selling factor services, especially labor. mean income give us a reliable indicator of its Women can be constrained in either home members' welfare? Consider again the example production or market work by the time involved of a two-person household in which the man in reproduction and household maintenance. farms an export crop and the woman separately Thus, in Figure 4.5 the women's time involved in farms a food crop. Further assume that the re- home work may be constrained to amount L- turn on labor is higher in export crop production LW, which is greater than amount L - LW2 the than in food crop production. The male earns a period determined by her preferences and the higher income than the female. Say he earns $120 applicable set of relative prices. Social custom while the female earns $80. Total household in- may determine that males will undertake only a come is therefore $200, while the mean house- limited amount of home work (and of a particu- hold income is $100. Now assume that out of his lar type), so that male time has a low substituta- earnings the male transfers $10 to the female, and bility for female time in home work. retains the rest. Post-transfer, the incomes of the Socially determined constraints on the time al- male and the female are $110 and $90 respectively. location of women have two important effects. The variance of the within-household income First, these constraints induce allocative ineffi- distribution is lower after the transfer than before. ciency within the system. Thus, insofar as the The degree to which mean household income labor resources of the household are not allo- provides an indicator of an individual member's cated in accordance with its members' respective income therefore depends on how far their pre- comparative advantages, output, and thus transfer incomes differ and the size of the net household income, is lower than it would other- transfers between them. If all income is pooled, wise be. Such inefficiencies may be an important the mean household income is a perfect indicator source of female poverty, as well as a contributor of its members' incomes. With no pooling, the to the overall poverty of a household. Second, gap between the household's mean income and the low substitutability of male and female labor that of the individual member rises as the variance time in specific activities reduces the ability of of the within-household income distribution in- women to reallocate their time in accordance with creases. In summary, when the pre-transfer in- changes in market and nonmarket opportunities. come gap between the male and female is large This has important effects on the welfare out- and pooling is small, then mean household in- comes of adjustment, again both for women indi- come is a poor indicator of the female's income vidually and for the household unit. (and thus of her overall welfare if we measure We have listed the respective production func- this by income levels).-' tions, preferences, and applicable relative prices of male and female household members as subject Implications for inequality and poverty measurement to change under adjustment. We must now add a fourth item to this list: the social customs There are basically two broad problems that arise themselves, which determine (and are determined from a research design that takes the household, by) the social and economic status of men and rather than the individual, as its basic unit of women. These, too, can change under adjustment, analysis and data collection. First, it may pre- since policy revisions, by altering the parameters clude (or at least make difficult) a detailed inves- under which household members control re- tigation into welfare effects of specific groups sources and allocate their time, alter the balance within the household (such as females or chil- of household bargaining power. Programs of dren). This is particularly the case when data on poverty alleviation implemented under adjust- individual-level expenditure or consumption are ment will also have their effects on this balance. required. Second, the calculations of estimates of As the balance of household bargaining changes, inequality and poverty (either for the whole so too may the wider social norms it underpins. population or subgroups) will be biased, since Consequently we must not paint too static a pic- they ignore intra-household income variations. ture of the respective constraints affecting men Haddad and Kanbur (1989) have assessed the and women; adjustment can cause complex shifts direction and size of the biases arising from the in the structure of such bargaining. use of mean household income rather than in- 45 come at the individual level. They conclude that in household composition can be represented in by ignoring within-household income variance, terms of "outlay equivalents," defined as: the use of mean household income is certain to underestimate the true variance of individual in- the additional total expenditure that would comes, and this underestimate can be quantita- generate the same change in expenditure tively significant (of the order of 30 percent). The on the good in question as does the pres- effect on measures of poverty are ambiguous, ence of an additional person of each demo- however-using mean household income might graphic type [Deaton 1987, p. 21. over- or underestimate poverty, depending on the underlying distributions involved and the Certain goods are usually consumed only by poverty index selected. For example, taking the children, while for other goods it can be pre- simple headcount measure of poverty, a poverty sumed that their consumption is mainly by adults index calculated using mean household income (for instance, alcohol and items of adult cloth- will understate the true (individually based) ing). Additional children will not generate a poverty level if the poverty line is less than mean direct demand for "adult goods," but outlays on individual income (which of course is equal to such goods will have to be reduced to release the average of all household mean incomes). resources for goods consumed by children. However, Haddad and Kanbur also demon- Therefore, the outlay equivalents for additional strate that it is unlikely that the patterns (or children on adult goods should be negative. This orderings) of both inequality and poverty meas- result can be used to test for age and gender ures will be affected by ignoring intra-household biases in the household's allocation of expendi- inequality. They conclude that the collection of tures - a phenomenon of key policy interest. individual data is important if levels of inequality For example, the reduction in expenditures on and poverty are required. But if the object is to adult goods will be larger following the birth of a identify only the patterns of inequality and pov- male child than a female child if discrimination erty across socioeconomic groups, there would against girls exists (Deaton 1987, p. 3). Similarly, seem to be little point in incurring the heavy costs the ratios of adults to adult goods can indicate of individual inquiries. whether there are differences by age and sex in the allocation of goods within the household." Household expenditures The equivalence scale technique thus uses cross- sectional data to build a picture of intra-house- The level and pattern of household expenditures hold expenditure inequality at any one time. If is a crucial dimension of living standards. A the technique is applied to data sets from succes- household's expenditure on items of consump- sive years, an understanding can be gained of tion and intermediate goods for its economic ac- how intra-household welfare is changing as ad- tivities can be subject to large changes during justment proceeds. It can be used to pick up adjustment. The level of consumption provides a changes in intra-household welfare that are being more direct measure of household welfare than generated by adjustment policies that have vary- income (Deaton and Case 1987, p. 1). Expendi- ing effects on different household members. For ture includes outlays for commodities and ser- example, if a particular policy is benefiting males, vices, both purchased items and those produced and they are not increasing their transfers to other for own-consumption. Even if income data can household members proportionately, this can be be constructed for individual household mem- identified by the equivalence scale method. The bers, some household members have little or no advantage of the equivalence scale technique is income of their own because of their position in that it uses data on total household expenditures. the life cycle, and thus consumption levels are To apply equivalence scales, expenditures do not more relevant for assessing their welfare. need to be broken down by household member. One method for analyzing intra-household is- Expenditure data can accordingly be provided sues is the equivalence scale technique, which by the household member with the most knowl- uses data on total household expenditures to edge of household expenditures. In addition to compare expenditure patterns across households gathering household expenditure data, it is de- with different demographic and gender charac- sirable to undertake more direct measures of in- teristics (Deaton and Muellbauer 1980, p. 191). dividual welfare, such as weight-for-age and Effects on household expenditures of differences height-for-age, particularly for children. Such 46 anthropometric measures describe the "output' the observed welfare changes. These causes may of the household's activities. In the Living Stan- lie on the consumption side of the household's dards Measurement Survey (LSMS) for the Cote activities, the production side, or within the d'Ivoire, for example, the heights and weights of intra-household transfer mechanisms. Conse- all household members are recorded (Grootaert quently, although consumption and anthro- 1987, p. 138). pometric data alone may identify a problem - Why is the household used as the sampling for example, a deterioration in child nutrition - unit in cases where policymakers are concerned they are insufficient for analyzing its causes and with the welfare of individuals? For example, for prescribing solutions. data on child nutrition could be collected by Having anthropometric or expenditure data drawing a random sample from all the children alone allows us to respond with special feeding in a given community, or using schools or health programs and other such measures. But to de- clinics to locate individuals of interest. Much sign policy interventions that reduce the root valuable data are currently collected on child causes of malnutrition requires an understand- welfare from sample units outside the house- ing of the economic activities of the child's social hold: for instance, the National Nutrition Sur- unit. Thus, the kind of data policymakers require veillance System in Botswana reports (on a depends on the type of poverty alleviation strat- monthly basis) the nutritional status of all chil- egies they intend to pursue. If the direction of dren under five attending health clinics (Stewart attack is mainly to be through secondary income 1987, p. 266). These data are useful for policy transfers, data that signal a problem on the con- because they are timely and they are collected sumption side of the household are sufficient. within an organizational structure that rapidly But attacking poverty through people's primary feeds information on deteriorations in nutritional incomes - generated by employment and self- status to executing agencies. Such systems are employment- requires information on both the extremely valuable in times of rapid economic production and consumption sides of household change in providing an early warning system. activity. Whether the focus of poverty alleviation For many policy purposes we need to place infor- should be on people's primary or secondary in- mation on the individual's welfare in the context comes will depend on the situation at hand. Ac- of his or her economic environment. Nutritional cordingly, we need a data set that integrates all surveillance exercises can provide some impor- the main dimensions of household activity and tant information on this environment: for ex- that allows analysis of within-household welfare ample, the national surveillance exercise in issues when necessary. Botswana disaggregates data on child nutrition on a regional basis, and further by villages, cattle Interactions between households posts, and settlements (Stewart 1987, p. 266). This information, along with data on food supplies Although many important decisions are made at and agricultural conditions in the areas concerned, the household level, households are in turn em- allow executingagencies to monitor, analyze, and bedded within wider social networks whose intervene. Transfers, for example, can be mobi- functioning has important effects on them (Guyer lized in times of drought and more general dis- and Peters 1987, p. 206). The most important of tress. these is the lineage group, comprising a duster of Such exercises, however, do not permit the households bound together by ties of marriage or analysis of child malnutrition in relation to blood, and under the control of an elder (Swindell household characteristics. Policymakers need to 1985, p. 33). Control of vital productive assets, know not only the level of child nutrition within such as land, is frequently vested in kinship a community, but also its distribution across groups, with the seniors determining the "use households of differing characteristics. This need rights" of its members.-" While such social to relate individual characteristics with those of groupings are most important in rural communi- the household arises from the requirements of ties, large numbers of urban dwellers maintain policy design. Since consumption data and an- their links with them. Such networks mobilize thropometric measures describe the end results transfers to households in times of distress of the processes occurring within the household through taxing other network members. This (of work and giving and receiving transfers), such reduces the variability of household income, a data by themselves tell us little about the causes of major concem for people living close to subsis- 47 tence levels and for whom insurance and other primary "social security" in most countries, it is inter-temporal capital markets are unavailable important to verify changes in their importance (Grootaert 1987, p. 32 and Binswanger and McIn- and the pattern of their distribution during peri- tire 1987, p. 82). These characteristics have im- ods of adjustment. Adjustment, through chang- portanceforanalyzingboth the welfare impact of ing the incomes of donors and recipients, may adjustment and the policy interventions that can generate important shifts in the structure of be made for poverty alleviation, and we shall inter-household transfers. For instance, poor frequently refer to their operation. families may find their social support weakened While households undergo periodic changes in or strengthened by changes in the incomes of their size and composition (for example, follow- their donors. ing marriage or the death of the household head), Aside from transfers and the sharing of factors they may also divide or combine in response to of production, households also interact with each "shocks": illness of family members, displace- other through the market - more specifically ment, drought, the creation of new market op- through the product, labor, and credit markets. portunities, and so forth. The wider social net- Thus some of the contracts within the household's works may also suffer displacement, and the so- village may be explicit - for example, over the cial structure in which households function needs provision of labor services, or sharecropping ar- to be viewed as a dynamic process. Various rangements. Credit markets will also exist within forms of inter-household cooperation can be villages. The relative role of market and noted. Within rural communities households nonmarket mechanisms in the allocation of goods frequently cooperate with each other over work and services across communities will vary by arrangements - for example pooling their labor region and by country. As development pro- at harvest time, sharing childrearing and food ceeds, market transactions will increasingly take preparation. Households make "implicit" con- over from other ways of allocating resources. As tracts with each other over the sharing of work farm productivity rises, more output will become loads, either on a regular basis or to provide available for sale, and greater opportunities will assistance to each other in times of personal dis- arise for selling produce as private and public tress. Such contracts may be underpinned by ties investments are made in marketing. The exten- of kinship, although this is not always the case. sion of public goods such as transport facilities Implicit contracts reinforcing reciprocal rights and will increase the access of remote regions to obligations can exist over a large number of ac- markets. In turn, the greater availability of tivities. By providing safety nets they reduce risk modern consumer goods will increase the incen- and make households less risk-averse than if they tives to sell produce and to move from barter to a operated on their own (Peters 1986, p. 136). greater use of money. Finally, urbanization The importance of affirming and strengthening weakens the extended community, and house- ties between households will lead to the periodic holds resort more frequently to market transac- distribution of gifts of cash, food, or manufac- tions with each other. tured goods. For example, farmers sometimes The increased role of market transactions in the distribute "harvest gifts" to important allies distribution of goods and services will lead to within the village (as described, for example, by further specialization in the division of labor as McMillan 1987, p. 306, in Burkina Faso). Such greater market opportunities allow households gifts often form important components of the to concentrate on their comparative advantages. "exchange-entitlements" of households and indi- Some will find that they no longer need to pro- viduals, and household survey questionnaires duce certain goods themselves because these can need to be designed to identify these. It is impor- be obtained more favorably through the market. tant to correlate such transfers with household Activities that were undertaken to minimize the characteristic, in order to analyze their role. Sev- risk of shortfalls in income may be cut back if eral hypotheses concerning the function of these households perceive that the market offers less gifts suggest themselves: transfers may take income risk, or provides a better safeguard than higher than average shares in the incomes of poor own-production. Changes in the size and charac- households or, alternatively, they may be dispro- teristics of product markets will, in turn, lead to portionately concentrated among wealthier changes in both the size and characteristics of households. Since such transfers provide the factor markets. The development of product 48 markets has been one of the factors behind a poverty in one society (at a given time) may not reduction in the importance of communal be the same for another society (or the same soci- mechanisms of allocating resources within Afri- ety at a different time). In restating the "abso- can societies. lute" case, Sen (1983b) distinguishes between the In sum, households will interact with other "capabilities" that incomes confer and the goods institutions, not only with other households. De- and services needed to produce them. Poverty, pending on the circumstances of the country and according to this view, is the absolute absence of region concerned, they may enter markets in certain critical capabilities -induding avoiding which firms are buying and selling - both prod- shame, community participation, and self-respect. uct and factor markets. In the case of the credit But the bundle of goods required to provide these market, they may have direct dealings with the capabilities varies from place to place and from organized banking system or other financial in- time to time, and it is in this respect that poverty termediaries. Of course, they may be taxed by is relative. If the absolute aspect of poverty is local or central government, or receive services ignored, as Sen (1983b, p. 156) puts it, "poverty provided by the state (such as agricultural exten- cannot - simply cannot-be eliminated and an sion services, marketing services, health and anti-poverty program can never really be quite education services, and the like). Thus, we must successful." distinguish the market and nonmarket interac- The debate about whether absolute or relative tions between the household and other house- concepts are appropriate is simply about the ap- holds, firms, and the state. We emphasize these propriate choice of poverty line, and this is certain otherwise obvious points because adjustment, to be a subject that would need further clarifica- together with the macroeconomic disequilibrium tion for each country. Biases in the measurement that precedes it, causes changes in the structure of of poverty (for example, the neglect of within- markets, the relative roles of market and nonmarket household income variance discussed above) tend mechanisms in allocating resources, and the de- to become more problematic when the poverty gree of market integration of households. Changes line is considered sacrosanct, since this implies in the "market parameters" under which house- that the level of poverty has some real policy holds operate can have profound effects on their significance. If the poverty line selected is some- welfare. what arbitrary (for example, in selecting a line that considers 30 percent of the population as The meaning and measurement of poverty poor), the biases often become less serious, since they rarely affect the pattern or ranking of pov- Any assessment of the social dimensions of struc- erty across groups in the society. tural adjustment must deal with the effects of But this is only one, albeit fairly crucial, dimen- policy reform on poor households. For this, sion to any measure of poverty. Apart from the measures of poverty must be devised. These selection of the poverty line itself, the degree of measures cannot hope to capture the full com- poverty will depend on three basic factors: plexity of poverty, but they are unavoidable if * the incidence of poverty, as measured by the policymakers are to understand how their inter- numbers in the total population living below the ventions affect the degree of poverty in the com- poverty line munity. Our intention here is not to review all the * the intensity of poverty, reflected in the extent available measures, but to focus only on one class to which the incomes of the poor lie below the of indicators that has proved useful in the analy- poverty line sis of poverty and adjustment interactions. Two * the degree of inequality among the poor, in broad issues arise in the measurement of pov- that transferring income from the poorest to the erty: the establishment of the poverty line and the better-off poor should raise measured "poverty." choice of a single index to measure poverty. These Any index or measure of poverty should ideally are considered in turn. reflect all three of these dimensions. Moreover, The literature on poverty has been concerned for our purposes, we need an index of poverty with the respective merits of absolute and rela- that can be used to assess the effects of adjustment. tive concepts of poverty (Sen 1983b reviews much Since adjustment frequently entails changing the of it). Recent work has questioned the usefulness sectoral composition of output - from nontraded of absolute poverty, because what constitutes to traded goods, from import-competing to ex- 49 porting sectors, and favoring agriculture - our Therefore, with a = 0, the index is simply the poverty index must be decomposable across sec- head-count ratio. Alternatively, with a = 1, the tors (Kanbur 1987a). We propose that the typical poverty index becomes presentations of poverty (usually by region in a q country) should be supplemented with measures PVQ s = (1/n) . [(Y, - Y)/Y,] = HI by socioeconomic group, with the choice of groups i =l partly determined by the relation each bears to where the markets and other elements of the q meso economy. In other words, if the structural 1= (llq) E (Y,, - Yp/Y, adjustment programs feature changes in the price i =1 of a particular food (for example, maize), it would is the "income gap" ratio. I is simply the average be important to estimate measures of poverty for of the poverty gaps expressed as a fraction of the maize producers and consumers separately from poverty line. PVl 1, or Hl, therefore takes into other groups. The measure must therefore be account how poor the poor are and reflects both subgroup decomposable. the incidence of poverty (as reflected in H) and its A useful index that meets this requirement is intensity (as given by 1). The PV/s, measure, suggested by Foster, Greer, and Thorbecke (1984). however, is insensitive to income distribution Their class of poverty index takes the following among the poor. Transferring income from the form: poorest unit to a richer (but still poor) unit will q leave PV.., unchanged (both H and I will be un- PVa= (1/n) X [(Y - Y)/Y ]l affected). For this to be reflected in the index, i=1 greater weight has to be given to the poorest where Yp denotes the poverty line, n the total income earning units. This can be achieved in population, and q the number of income earners this class of poverty indexes by assuming values below the poverty line. Essentially the index of a in excess of unity. With a > 1, a transfer of takes the poverty gap of each poor person as a one dollar from the poorest units to other (better fraction of the povertyline, (Yp- Y)/Y, raises it to off) poor units will increase the poverty index. the power a, sums over poor units, and divides In short, the PV1 class of poverty indexes sug- by the total population. Not only does this index gested by Foster, Greer, and Thorbecke (1984) take into account the incidence and the intensity permits the user to specify a, and thereby select of poverty, it is also sensitive to the degree of an index that reflects his or her aversion to pov- inequality among the poor. This is governed by erty. the value of a, which determines how sensitive This class of poverty indexes is flexible also in the index is to transfers among the poor. For a > that it is subgroup decomposable. The "overall" 1, transfers from the poorest to better-off poor index of poverty can be shown to comprise the groups will increase the measure of poverty. summation of poverty indexes among all the This class of poverty measures is flexible in subgroups in the population. If the study popu- two important respects. First, a is a policy pa- lation consists of m groups or sectors, then, rameter that can be varied to reflect correctly m poverty "aversion". If a = 0, it can be readily PV1 = X. PV. I a shown that PV simply becomes, j = PV, 0=q/n=H wherePV isthepovertyindexofgroupjandx, the population weight of group (j = 1, . . . ,m), Yx; where H is the head-count ratio, that is, the pro- = 1. This decomposition property is useful in portion of total income-receiving units below the analyzing poverty changes because it generates poverty line. Note that if a = 0, it simply means both overall indexes in each country and breaks that the measure is entirely indifferent to how these indexes down for each socioeconomic group poor each poor unit is - it does not matter how and sector under consideration (see Kanbur 1988 far below the poverty line each poor person is. for a recent analysis along these lines). 50 5 The effects of destabilization and adjustment on households Thus far, our analysis has traced how destabiliza- nontradables, while tradable prices will remain tion and adjustment lead to significant constant, under the assumption that their domes- mesoeconomic changes in both markets and in- tic prices are given exogenously by world market frastructure. We have also explored the main conditions (and that there is no change in the features of the household economy, including terms of trade). the factors that have a bearing on the welfare of We begin with household-based production individuals within the household. We come now activities since these have been identified as one to bring these two elements together by tracing of the most important sources of household in- how the mesoeconomic changes we have ob- come in Africa. The value of a household's out- served are likely to affect the household and the put is given by the price of the good or service welfare of its members. Again we should reiter- multiplied by the quantity produced. Prices are ate that the purpose here is to outline a way of exogenous to households, but they do have some structuring thinking on these issues, rather than control over the level and composition of their pretending to represent household effects in their output. In the very short run, when households entirety throughout the region. cannot adjust their allocation of factors among activities, they will be confined to producing the Household incomes under macroeconomic same goods and services that they did before disequilibrium nontradable prices began to rise. Thus, in the very short run, with the composition of output To trace the effects of this policy scenario on fixed, the effect of the monetary expansion on the household incomes, assume a one-off rise in the value of household output will be determined money supply because of a rise in the budget entirely by the shift in relative prices toward deficit. Ignore for the moment the issue of nontradables. The nominal incomes of households whether this policy is pursued in successive years, producing nontradables will accordingly increase. and assume that the level of national output is We can label this the impact effect. Noniinal in- fixed. We employ a number of simplifying as- comes could remain constant in the tradables sumptions in order to illustrate the most impor- sector if this sector does not use nontradables as tant distributional effects. At the end of this intermediate goods (and given that tradable prices section we discuss these assumptions further, but remain constant). Otherwise they will fall. show that while relaxing some assumptions in- One outcome of special interest is that, holding creases the complexity of the processes, the con- all other parameters constant, the distribution of clusions of the simpler model are not fundamen- income will shift in favor of urban households tally altered. Recall that excess monetary expan- during a period of destabilization, because they sion will lead to an increase in the price of engage more in the production of nontradables 51 than do rural households, whose production ac- remain unchanged. This, together with the in- tivities are concentrated on tradable agricultural crease in the price of the goods they sell, will products. Urban households that produce items unambiguously raise their real incomes. protected through quotas will gain the most, since Nontradable households will be hit by the in- the scale of their protection rises as the real ex- crease in the prices of their purchased consumer change rate appreciates. Given that it will be tradables; this, together with the unchanged prices urban households with the largest factor endow- for the goods they sell, will unambiguously re- ments that receive the largest gains, we can expect duce their real incomes. The households affected a rise in the level of income inequality in the the most severely in this sector will be those who urban sector. This, combined with the location of purchase large amounts of intermediate most poor households in the rural sector, will nontradable goods and have large shares of mean that a rise in the overall level of income nontradables in their consumption baskets. inequality is a likely outcome. In summary, we will observe a shift in real As time elapses households will attempt to shift income shares toward households in nontrad- resources into the production of nontradables as able sectors. With terms of trade shocks also a result of the destabilization, and possibly causing macro disequilibrium, relative prices were importables if the terms of trade have declined. also shown to move in favor of importables. The Within the household-based production unit, la- relative price configuration of the move from A bor is the easiest resource to reallocate - the to D in Figure 2.2 would affect the various house- household's capital resources are invested in holds, the manner depending on the proportions equipment that may not be so easily transferable of their sales and purchases of importables, to nontradable activities. The period when only exportables, and nontradables. Some households some factors can be reallocated will be labeled may find themselves not only switching produc- the short run, while the long run begins when it is tion to nontradables, but also to protected possible to reallocate all factors. importables. What will happen to poverty? While Incomes from household-based production ac- the rise in real incomes in the nontradables sector tivities will be affected not only by the prices at will be unevenly spread depending on the own- which households sell, but also the prices at which ership of factors and so forth, low-income house- they buy. First, consider purchases of intermedi- holds will benefit, and this may be sufficient to ate goods. Recall that in theory, if importable bring some of them above a defined poverty line. goods are pure tradables, their domestic prices How many of them cross this threshold will de- will remain unaffected (assuming a fixed nomi- pend on the distribution of income within the nal exchange rate) by the government's mone- sector and the magnitude of the real income in- tary expansion, since the market for importables crease among poor households. immediately clears at given world prices through At the same time, poor households engaged in further imports in response to extra domestic the production of tradables (and especially demand. Thus, producers of nontradables find exportables) will find their real incomes declin- that the value added (and hence their implicit ing, with a consequent increase in poverty. The profits) of their goods and services is increasing, net effect on poverty in the country will therefore because their output prices are rising but the depend on the relative strengths of these oppos- costs of their imported intermediate inputs re- ing effects and on the ability of households to main constant. The effects within the tradable switch from tradable into nontradable (or pro- household will depend on whether they use tected importable) activities (Kanbur 1987c, nontradables as intermediate goods. If they do Demery and Addison 1987a). One hypothesis is not, their value added - and therefore their that nontradable goods, because they consist nominal income - will be unchanged. House- mainly of urban services and domestic manufac- holds that use nontradables as intermediate goods tures, take a higher share in the consumption will see their value added and nominal income baskets of wealthier households than poorer fall. households, whose consumption basket is more Households will also consume tradable and heavily weighted to (tradable) foods. So with nontradable goods in different proportions. If tradable prices constant, and nontradable prices households producing nontradables consume rising, the cost of living of better-off groups is only their own nontradables, and thus purchase more affected than the cost of living of the poor. only tradables, their cost of living basket will This may offset some of the negative impact on 52 the poor resulting from their high participation assumption and assume that the government ei- rate in tradables. ther maintains or raises its budget deficit in suc- What of households selling factor services? cessive years, and thus continues to add excess Under the impact effect with no factor mobility money balances to the system through financing between the tradable and nontradable sectors, its deficit from bank borrowing, then the level of factor sellers in the nontraded sector will gain output may increase in a Keynesian manner higher money incomes because the demand for through the pull of rising aggregate demand. The all factors will rise. If labor is underemployed in demand for both tradables and nontradables will the sector, most of the initial benefits to labor will grow, which will induce a growth in the output accrue in the form of higher employment rather of both types of goods. Thus, we may observe a than higher wages. Over the longer term as fac- pattern of economic growth in which tradables tors become mobile, they will flow from tradable take a declining share of total output, but their to nontradable sectors. Recall that public and absolute production level rises. Households pro- private service employment is classified as ducing tradables may see their incomes increas- nontradable, while formal manufacturing - and ing if this growth process is achieved, but not by to a degree some informal manufactures - are as much as households in nontradables. afforded sufficient protection to make them Nevertheless, such demand-induced growth is nontradable goods. In such circumstances the unlikely to be sustainable for a long period of direction of labor flow will be toward public and time in most African countries because of their private service employment together with supply constraints. While these supply con- nontradable formal and informal manufacturing, straints can be reduced by appropriate invest- and away from tradable agriculture. ment strategies, public investments in Africa have The final outcome for sellers of factor services generally yielded low returns during periods in will depend, as we have shown, on the structure which macro and micro policies have been inap- of the labor market. In the simplest case, the short- propriate (World Bank 1986a, p. 1). Thus, a rapid run effect was determined by their consumption reduction of supply constraints is unlikely during propensities and in the longer run by the relative the policy scenario under discussion, especially factor intensities of these activities. Ranking sec- given the disincentive effect of the real exchange tors by factor intensity is a difficult business in rate appreciation on agriculture. With domestic developing countries. We know that most supply increasingly constrained, and demand smallholder agriculture is labor-intensive in Africa continuing to expand, the inflation rate will and uses relatively little capital. Informal manu- eventually accelerate, and may generate its own facturing is also labor-intensive, as are the public momentum as price- and wage-fixers adjust their and private service sectors. The formal manufac- forecasts upward. turing sector, however, is generally character- The acceleration in nominal inflation will hit ized as relatively capital-intensive because of the households with a high dependence on market promotion of cheap capital imports through purchases to meet their needs. Poor urban overvalued exchange rates and explicit subsidies households, while they may have benefited from to capital users. So the final outcome for factor- the swing to nontradables, are likely to be the sellers will depend on the shares of these activities most substantial losers from high inflation rates. in the nontradables sector. If nontradables Rural households dependent on wage employ- amounted to only labor-intensive informal ac- ment may also lose, although many are still paid tivities, the long-run effect on the wage of the predominantly in food in Africa, which will re- shift to nontradables would be small. But if duce the effect on them. Subsistence farmers will nontradables are dominated by capital-intensive be relatively protected in good years, but may be manufactures then the wage would fall in the very vulnerable if drought drives them into a long run. If sections of the labor market are sub- food market that is undergoing rapid inflation. ject to some form of imperfection, however, the Inflation will also redistribute income among real-world outcome would be different, depend- savers. Households that hold their savings in ing on which sector was affected. cash will be hit the hardest. Deposit rates are set The analysis of this chapter has so far been by governments in many African countries and conducted under the assumption that while the have not generally kept up with the rate of infla- composition of total output can change, the level tion. Households dependent on such income of output itself remains fixed. If we relax this will lose out unless they are able to transfer their 53 savings into foreign currencies. Loss of confi- the consumption behavior of the households in- dence in the domestic currency as a store of value volved. Since we know that W/P, will decrease will encourage capital flight, both legal and illegal. and W/P. will increase (assuming full employ- Wealthier households have more access to inter- ment), households that consume mainly national stores of value and will adjust the share nontradables may find their real incomes rising of domestic financial assets in their portfolios. in the short run, while those consuming mainly Households holding their savings in commodi- tradables will face a cut in real incomes. ties - for example, cattle - may actually gain as Added to these real wage effects, households the prices of their savings assets rise. may face deteriorations in employment prospects, especially during the transition period. House- A framework for considering the household holds experiencing unemployment among their effects of adjustment members will obviously respond and seek job placements in expanding sectors. This may re- Chapter 3 established the effects of adjustment quire some additional skill acquisition, or it may policies on the meso economy. To draw the involve geographical migration. The time such threads of our argument together, we now sum- households take to perceive the need for these marize these effects and describe briefly how they responses and their capacity to make the adjust- will influence households. The main meso effects ments are critical in determining their share in of the adjustment policies reviewed here are: these transition costs. It may be that for certain • an increase in P1/Px in product markets households, some advantage maybe gained from • possible changes in relative prices within the direct micro interventions (such as retraining or tradables category - for example, formal and relocation grants) by the government. This type informal tradables of assistance will reduce the transition costs, re- - resource reallocations from nontradables into duce the welfare loss of such households, and at tradables the same time enhance the capacity of the econ- * a short-run increase in WIPE and decrease in omy to achieve its structural adjustment objec- W/P, tives. Changes in the "social wage"- benefits short-run real consumption wage changes transferred to laboring households by the state - that are dependent on consumption propensities should also be taken into account. Adjustment and the structure of the labor market may well reduce the provision of these benefits - long-run changes in the real wage that depend through cuts in food subsidies, other transfers, on factor intensities education and health expenditures, and so on. i short-run or transitional increases in labor The incidence of such cuts falling on the various unemployment; also changes in search unem- households is essentially an empirical issue. ployment Many poor households in Sub-Saharan Africa * increases in domestic interest rates and cuts possess productive assets other than their labor: in credit availability they include those engaged in the urban informal * cuts in government expenditure, with impli- sector and smallholder groups producing agri- cations for economic and social infrastructure. cultural commodities. Changes in the product Evidently the effects of these changes on the markets affect these households not only as con- various socioeconomic groups in any country sumers (or purchasers in these markets) but also undertaking structural adjustment will be quite as producers (or suppliers to the markets). Simi- complex. They will depend on the choice of larly, changes in the labor market may affect these policy instruments, the nature of the markets in- households in their capacity as hirers of labor volved and of the institutional setting of the services and not only as sellers. Because of this, meso economy, and the characteristics of the the mesoeconomic effects on such households households in each group. Consider households are rather more complex. Despite this complex- that possess only their labor - that is, the house- ity, it is possible to set out a range of possibilities holds that comprise the urban poor and the rural using the simple microeconomic framework dis- landless. For such households the labor market cussed in the previous chapter - namely, the and the social infrastructure will hold the key to recent literature on agricultural household mod- how they are affected by adjustment. Under an els (Singh, Squire, and Strauss 1986a). expenditure-switching adjustment strategy, the We shall take for illustrative purposes the re- short-run effect on the real wage will depend on cursive version of these models, which is valid 54 for cases where households are price takers in Figure 5.1 product and factor markets. With the recursive model, it is possible to analyze the effects of these GwdY i mesoeconomic changes in stages, beginning with the household's output decision and moving on B to evaluate the consumption effects. In order to trace the effects on the household of the changes induced by adjustment using this class of mod- els, it is helpful to proceed in stages, utilizing the A simple framework described in Chapter 4. We begin by tracing how households are af- fected by the relative price changes induced by an adjustment program. In terms of our simple c A household model, this is depicted by a change in c the relative price, W/P. If the household pro- duces and consumes tradables, this ratio is likely to fall during adjustment, whereas producers of nontradables will experience a rise. Assume ini- tially that the household produces and consumes - only one commodity - a tradable food good. 0 Using the framework that was presented in Fig- ure 4.3 (and repeated in Figure 5.1 for conven- consumes another. In modifying the analytical ience) we can trace how its production and con- structure used in Chapter 4, we assume that the sumption decisions are affected by the relative household produces one commodity (say the price change. tradable commodity, T). It consumes both T and Assume that adjustment reduces W/P, rotat- a nontradable (N), with market prices of P, and PX ing the price line to (C'/B'). The first effect is to respectively. We shall assume also that the con- increase the level of production of the tradable sumption of these commodities alone yields util- good from A to A'. This occurs independently of ity to the household, which means that no utility consumption decisions and whether or not hired is gained from leisure (or housework as defined labor is used. But how the household's utility is in our earlier discussion). Figure 5.2 (overleaf) affected by this relative price change will depend depicts the initial equilibrium of the household on whether it is selling or buying labor services. prior to the adjustment program. The production It clearly falls for households that sell labor ser- decision is derived in quadrant I, which traces vices, with the consumption point being drawn household labor time horizontally against total in from B to B', and this will involve a cut in both household output (of 7) vertically. The initial the consumption of food and leisure and home production point is at A. care. It is only households that buy-in labor ser- Because, under our assumptions, maximizing vices that will gain, their consumption point ex- utility is equivalent to maximizing income (leisure panding from C to C'. Thus, although all house- yielding no utility), the household will devote all holds increase output as a result of adjustment, its available time to work, either in own-produc- only those buying labor services are likely to tion or in market work. To derive consumption, gain.`' we must identify the consumption possibility set This analysis is useful in showing how con- facing the household. In maximizing the avail- sumption and leisure and home care can be ad- ability of T, the household would devote OL' to versely affected for some producing households, producing T and all the proceeds of its market even though they may be producing the com- work to buying T. Since each unit of labor de- modity favored by adjustment. The main draw- voted to market work yields a real return of W/PI back of the analysis is the assumption of only one in terms of the tradable, LT* is the maximum commodity consumed, since relative product consumption point for tradables.5' To maximize price changes are generally crucial elements of its consumption of nontradables, the household an adjustment program. We therefore proceed to will be required to devote all the proceeds of its the next stage and analyze how a household may market work to purchasing N, and at the same be affected if it produces one commodity and time sell all its output of tradables in exchange 55 Figure 5.2 nontradables, and their price has fallen (relative, that is, to the nominal wage). This reduction in Tradables' W/P. obviously benefits the household as a con- r \ IV sumer and more than compensates for the in- crease in P,. Even if W/Pn were to be held con- c stant for discrete changes in W/P,, however, many producers who were previously tradables-deficit A / \ households become net surplus households as a A - x result of the production increase brought about by the increase in P,. For some of these house- W// | \holds, a rise in P, will be beneficial, even though W/Pt / § L \ they were net consumers prior to the price changes Man hofrs ' ' , N- 0 brought about by adjustment. Manhoas 0 L' /;D 'N' Non- Similar, although opposite reasoning may be e / \ , applied to households that produce nontradables / \ | and purchase tradables in the marketplace. For / \ ' such households production would decline, and / \ | net sellers of N would experience welfare losses. / \ ' The effects on net purchasers would depend on their consumption bundles. To summarize, the effect of changes in product and labor markets on ---W/P0-------- -------- p production will be to unambiguously increase . the output of tradables and reduce that of Man houm nontradables. It will also cause predictable real- locations of labor time, with tradable households for nontradables. This gives a maximum con- reducing supplies and increasing demands for sumption of LN*. Household equilibrium is then labor, and nontradable households increasing given at the point of tangency of the indifference their supplies (or reducing their demands) for curve and the budget constraint (N*T* in quad- labor. The net effect on household welfare will rant IV). Consumption points to the northwest of also depend on the consumption preferences of X on the budget line indicate that the household the household groups. consumes T in excess of its production, so that it This simple framework abstracts from a num- is a net purchaser of the commodity it produces. ber of real world complications that are clearly Similarly, the household must be a net seller of T important. In the first place, households may if its consumption point is to the southeast of X. produce both tradables and nontradables, so that The household in the case of Figure 5.2 is assumed changes in the relative price structure will induce to be a net purchaser of T, consuming at C. them to switch resources in response. A key de- Now consider a decrease in W/P, and an in- terminant of the net welfare effect would be the crease in W/Pn brought about through adjust- extent to which such production switching is ment. This will increase the production of possible in both the short and long term. Simi- tradables from T, to T2 (Figure 5.3). These rela- larly, they may use intermediate commodities in the tive price changes will also change the budget productive process, and the price changes of these line in quadrant IV, from N*T* to N**T**. The commodities will have to be taken into account consumption (and welfare) effects will depend in computing the net effect of adjustment policies on the original consumption point. For house- on the rates of return that can be obtained. Farm- holds originally consuming along the segment ers relying on tradable inputs (such as fertilizers) ZT*, the relative price change will involve a wel- may find farm profits declining during adjust- fare loss, while households along ZN* will expe- ment, even though the output price is rising. Ac- rience a welfare gain. Note that it is possible for tivities that are intensive in the use of nontraded some households to be net purchasers of the good inputs (such as irrigation) will receive stronger in question, yet gain from the increase in the positive price inducements. relative price of T (that is, households originally One of the main characteristics of poorer rural consuming along the segment XZ). The reason households in Africa is risk aversion, a factor that for this is that the households also consume is ignored in this framework. Faced with a seri- 56 ous threat to both livelihood and survival, some amount of credit they need at the existing rate, or risk-averse farmers may choose to return to sub- they will find interest costs increasing. The effect sistence, and tum away from market opportuni- of these changes in the credit market are more ties. If this is the case, the relative price shifts that likely to be indirect in the African context - they are signaled in the mnarket may not evoke the will have adverse effects on farm productivity output response predicted by this basic orthodox (on the ability of households to purchase inputs framework. One of the most important reasons in a timely manner), farm incomes, and thereby that rural households are risk-averse lies in their on household welfare. imperfect access to credit. Because of this, they Finally, such household enterprises will be af- are unable to finance periods of poor output and fected by changes in the economic and social infra- income through borrowing against future income structure. Both of these will influence farm pro- streams. This restriction on inter-temporal sub- ductivity and profitability. A reduction in stitution makes them all the more vulnerable to physical infrastructure services and other farm unanticipated changes in their economic circum- support services is certain to impose constraints stances. Credit policy is therefore viewed as an on farm profitability and the ability of farmers to essential complement to price policy, especially respond to the relative price opportunities of- in the context of rural Africa. fered by structural adjustment. These "condi- The framework also abstracts from issues of tioning" effects have been largely ignored in the inter-temporal choice that households face. adjustment literature, and yet they are likely to Households will be affected by changes in the be very important for many groups of African credit markets that we have observed result from smallholders. In many African countries the monetary contraction. Again, they will face ei- physical infrastructure serving some rural areas ther increased quantitative constraints in the has seriously deteriorated, so thataccess to needed market, so that they simply cannot obtain the input and output markets has become signifi- cantly more difficult. The fact that prices in ur- ban or near-urban markets have improved may Figure 5.3 be of little relevance to farmers in remoter areas. Similarly, services that enhance human capital Tradables (such as health and education) may have indirect r, effects on farm productivitys in addition to any direct effects on household welfare. These social IV changes will also influence household welfare directly (and not only through raising economic -- - T2 productivity) and it is to these direct effects that we now turn. A x Other aspects of household welfare effects Thus far our concern has been to establish how (Wlpi/ L adjustment influences the economic welfare of 4O L / - N N"Non- the household. The theory suggests that the best hours /45 \ radables single indicator of this is aggregate household expenditure per capita or per adult equivalent. But because we are primarily interested in the welfare of the individual, this must only be con- sidered as a proxy. It is a proxy in two senses. ' / \ \ , First, it assumes that measuring household-level indicators will give us a reasonable measure of the welfare of its individual members; second, it 1. I> \ \ , assumes that the economic circumstances of the , / \WJPi\ I household is a useful indicator of its welfare. We )t---I1 ; 0) equals the generally accepted current definitions, the impli- excess of income received by domestic residents cations of any variations in them could materi- over their expenditure, ally affect any set of estimates. 68 (4) BOP = Gross national income - domestic tions of all three categories of these policies en- expenditure capsulated in macroeconomic stabilization pro- grams. = (Y + NFP - NTR) - (Cs + Cp + + Measurement problems = (X + NFP) - (M + NTR) Some standard, though largely unresolved con- and from (3), ceptual problems relating to the measurement of economic activity are especially relevant in the (5) BOP = (Sp - Ip) + (T - Cs- Id African context. It may therefore be pertinent to draw attention to these problems at this juncture (6) = ANFA insofar as they have a direct bearing on our mea- surement of the social dimensions of adjustment. where (S - Ip) represents the private surplus, (T - For example, a major conceptual problem relates C - I) the government surplus, and ANFA the to defining the appropriate boundary of produc- change in net foreign assets. Equivalently, a bal- tion, which, in turn, means defining the point at ance of payments deficit implies a decrease in net which nonmarket activities should count toward foreign assets. Thus, for instance, if net savings the measure of output or income (Havrylyshyn in the private sector are zero, a public sector fiscal 1977; Hill, 1977; Kusnic and da Vanzo 1980, and deficit automatically implies a deficit in the bal- Pyatt 1987). This is important because no impu- ance of payments account, and hence a decrease tation is currently made for certain activities, such in the holding of NFA. The decrease in NFA as gathering firewood, on the grounds that they would be realized by a decrease in foreign ex- are "nonproductive." In more developed econo- change reserves held by the central bank by an mies, however, "equivalent" products would be increase in government liabilities held abroad, or included because they are traded in a market (for by compensating changes in both elements so example, fuel). There are many quite difficult that there is a net decrease overall. conceptual issues that come into sharp focus as The balance of payments identity helps us to soon as we begin to integrate macro and micro focus on the alternative policy measures that a data outside the confines of the household government can pursue to correct external im- economy and within the scope of macroeconomic balances (Helmers 1988). Three policy measures conventions (Ruggles and Ruggles 1986). could be designed to address BOP deficits as One significant practical problem concerns the expressed through (4), (5), and (6). The first are estimation of private consumption (C9. This is termed expenditure-reducing policies, and these the largest single aggregate on the expenditure consist of monetary and fiscal policies. Examples side of the national accounts, and yet it is usually include the reduction of government expendi- estimated as a residual. Even if a household tures (or raising taxes), which would lead to an expenditure survey exists, it is not common prac- increase in the government surplus, or the pursu- tice to use it to estimate Cp. The SNA includes a ance of policies to curtail investment or increase recommendation that the national accounts be saving in the private sector. Under ceteris paribus estimated through the commodity balance ap- assumptions either of these would lead to a re- proach and, in principle at least, this would have duction of the BOP deficit [(see (5)]. The second the virtue of helping to reconcile inconsistencies class of measures are termed expenditure-switch- among estimates obtained from different sources. ing, and these operate directly on trade flows The basis of the 1968 SNA is a proposal that each [(see (4)]. Thus, for example, trade policies in- country compile a set of input-output tables. In volving a change in the regimes of quantitative this way supplies of commodities can be matched restrictions, tariffs on imports, export subsidies, with dispositions, and the inter-sectoral (inter- and changes in the real exchange rate would all industry) transactions representing intermediate lead to a switching of expenditures among do- consumption can be netted out to leave the ag- mestic goods, exports and imports. The third gregates consistently estimated. The problem is category could be termed financial policies, and that African statistical offices rightly view the these would arise when a current account deficit SNA system as being too demanding and funda- is financed by reducing net foreign assets as indi- mentally unworkable in its present form. As a cated in expression (6). We often find combina- result there is rarely any attempt to incorporate 69 any data sets that might lead to reconciliation Markets problems. This is basically why it is usually preferred that personal consumption expenditure There are many ways of characterizing the mar- be estimated as a residual. It is a worrying out- kets in an economic system. In the first instance come nonetheless, because if the national accounts it is useful to make broad distinctions among estimates of personal consumption at the product, factor, and financial markets, and possi- macro level are relied on to provide the basic bly asset markets as well, although the latter could information on individuals' living standards, the be viewed as a set of markets that overlap the margins of error are likely to be enormous. So markets for goods (physical assets) and financial there must be some early improvement in na- assets. Markets can be of several types. They can tional accounting procedures, at least in regard be either more or less formal, and official or par- to personal consumption as it is currently de- allel. Macroeconomic adjustment policies directly fined. or indirectly affect all these markets by altering market conditions through changes in relative Information requirements at the meso level prices and the quantities traded. Most households trade in at least one of these markets and, indeed, The meso level has already been established as many African households trade in all of them in crucial in determining how changes in policy at their various capacities as consumers and pro- the macro level ultimately affect individuals and ducers of products and as suppliers and users of households at the micro level. This intermediate factor services. Hence, if market conditions level has been defined to include two key elements change, so will the well-being of individuals, and - markets and infrastructure - that serve as con- it will all happen in a complex and not easily duits of policy between the macro and micro predictable way. It is for this reason that we need level. The recognition of the meso level has to examine in a little more detail the range of prompted the need to examine two stages in our possible effects of adjustment policies on the analysis: the "macro-meso" stage whereby macro conditions prevailing in each of these markets. policies affect market conditions and social and economic infrastructure, and the "meso-micro" PRODucr MARKETS. A basic distinction was made stage, in which changes in the meso economy earlier among three categories of products: ex- affect households. The establishment of the role portables, importables, and nontradables, where of the meso economy and the two-stage analyti- tradables canbeconsidered as comprising export- cal approach implies the need to examine very ables and importables. We saw that nontradables carefully the intervening variables in the process. are a class of goods and services produced and In particular, it is necessary to establish the infor- distributed only within the country. This can be mation requirements and the range of possible because of high transport costs or commercial empirical approaches in order to carry out an policy obstacles. Tradables, on the other hand, are effective two-stage analysis. It should be recog- goods that cross frontiers and, in theory, their nized that some of the important changes that prices are determined by world market condi- take place in the meso setting may be qualitative tions. This distinction proved useful in establish- rather than quantitative, which are not usually ing some basic results in the context of a highly observed very well empirically. Nevertheless, stylized, three-sector model on how adjustment and in spite of these caveats, there is a significant policies are likely to affect market conditions, and number of quite tractable analytical and empiri- hence the well-being of households. In practice cal approaches that involve information that ei- thedistinctionbetweentradablesandnontradables ther already exists or can be readily identified as is more arbitrary and, as regards data require- a priority requirement in a survey or information ments, it usually means that quite detailed com- gathering exercise. It is the purpose of this sec- modity disaggregation is necessary. tion to set out these requirements in a systematic Starting from the position of an extemal imbal- fashion. To facilitate this, our initial discussion ance (which can be expressed in terms of an ex- will be organized around each of the two con- cess demand for tradables), we can consider first duits in turn to establish the role they play and the effects on the product market brought about their possible implications for empirical analysis. by an expenditure-reducing package consisting of fiscal and monetary contraction. The reduction in domestic demand, designed to bring the exter- 70 nal account into balance, will result in an excess tions where parallel markets exist alongside the supply of nontraded products. The question now official markets for certain commodities, includ- arises of how this excess supply can be elimi- ing the market for foreign exchange. Many of nated. The theory is that if the price of these are illegal and involve smuggling or illicit nontradables is perfectly flexible, they will move dealing in goods obtained under quota at a price downward to clear the nontradables market. In below the true scarcity cost. Empirical informa- other words, the price of tradables relative to tion is therefore not easily obtained. One conse- nontradables (the real exchange rate) will rise, quence of parallel markets, however, is that eco- thus changing the relative profitabilities of the nomic rents will accrue to some members of soci- two sectors. There will be a tendency for re- ety and not to others. Hence, if market liberaliza- sources to shift out of nontradables and toward tion reduces the activities in parallel markets, tradables. marked distributional and social consequences In practice, product markets will respond to will result. Furthermore, the shift from parallel to these policies in a variety of ways and it will be official markets could exaggerate our measure of crucial to find out how they might operate in output change and of the amounts traded, and order to assess the social impact. To begin with it there is some evidence to suggest that this could must be said that the assumption about flexible be substantial. prices has limited application to Sub-Saharan Af- The effects of the relative price and output rican countries. This is partly because of adverse changes in the product market referred to above expectations in the sense that producers of impinge on households in a variety of ways. In nontradables may be reluctant to lower their terms of the three-sector distinction, some basic prices when demand falls if they believe that results are immediately clear. For example, as policy reversal is possible. But it is also a result of consumers, the beneficiaries will be households the inherently oligopolistic nature of much of the that consume relatively more nontradables than manufacturing system and to the fact that many tradables. But many households are also produc- African countries start from a position where ers who will be affected by the shifts in relative stringent controls and strong government inter- producer prices, the implication being that vention are in force. All these factors underlying tradables producers (for example, small-scale ag- price rigidities tend to slow down the process of ricultural activities) will benefit most from the adjustment. Expenditure-switching policies such increase in the relative price of tradables. In as devaluation or tariff reform are designed to restricting our attention to the product market directly influence the change in relative prices in implications, the net effect on household well- favor of tradables, and hence speed up adjust- being might be ambiguous and need to be as- ment. Devaluation has a tendency to be infla- sessed empirically in each case. tionary, however, and tariff reform will also ad- just the relative prices within the tradables cat- FAcroR MARKETS. Many households rely on the egory between exportables and importables, so it sale of labor services to derive wages for their is by no means insignificant to know just how livelihood. But in Sub-Saharan Africa there are this realignment of prices comes about in assess- also many households, especially those engaged ing the overall social impact. in the urban informal sector and smallholder It should be noted that in practice the distinc- groups producing agricultural commodities, that tion we have drawn between tradables and hire as well as sell labor services. Therefore, nontradables is not clear-cut, and this can com- households will feel the effects of adjustment plicate empirical analysis when we move beyond through changes in the factor markets as well as in the simplified two-sector model. In some in- the product markets, which will add further com- stances sectors such as construction, housing, and plexity to an overall assessment of the effects on government services can be clearly designated as particular household groups. A few features of nontradables and others, such as the main export the factor markets need to be examined in the commoditie, as tradables. In general, however, context of empirical analysis and information re- tradability is a matter of degree rather than some quirements. absolute division. One of the main objectives of adjustment is to It is important to introduce one more feature of switch resources from nontradables into tradables the real world at this point. Many African coun- production. In assessing the implications for the tries begin their adjustment programs from posi- factor markets, however, we must distinguish 71 between the long and short-run consequences. very little information about the informal sector, In the short run, the transition is unlikely to be and yet these kinds of outcomes could have sig- smooth and unemployment will probably rise as nificant bearing on how adjustment affects the the economy adjusts. There are two principal poor in society. reasons for this. First, non-tradable activities will In carrying out a detailed analysis of the labor generally contract faster than tradable activities and other factor markets during adjustment it is can expand, especially if the latter require the important to recognize the existence of several rehabilitation of equipment and new investment. quite distinct markets. We have just referred to Hence, factors will be unemployed for a time the feature of labor market segmentation in which while awaiting their reallocation to the tradables the market for a particular kind of labor is di- sector. Second, the expenditure-reducing pack- vided into the formal and informal components, age usually involves curtailing government ser- and where the structure of wages does not mirror vices and hence a direct retrenchment of govern- the distribution of skills and human capital. It ment employment. In the long run, however, will be useful to distinguish labor markets in real wages will adjust and move toward a new other ways too: by employment status (self-em- equilibrium, but whether these are higher or lower ployed, employer, employee); by level of educa- will depend upon the relative factor intensities of tion attained; by gender; and by region, espe- the tradable and non-tradable sectors. cially if labor is geographically immobile. It might An important dimension in the labor market be necessary to identify different forms of capital concerns the existence of a traditional informal (distinguishing housing from other capital, for sector alongside the organized formal sector. The example), especially because of the relative im- informal sector is sometimes closely aligned with mobility of capital among production sectors in the nontradables sector because many people the short or medium term. Only by observing work as artisans and petty traders. A substantial the relative changes in the factor markets, and part of the production of tradable goods in agri- the remuneration of labor in particular, can we culture is also informal, so the distinction is not begin to see how the primary income generation nearly so clear cut. What is fairly clear is that the of households is affected by adjustment. informal sector is significantly affected during the adjustment process. For example, as a result FiNANCIAL MARKEIS. Although households will of deflationary measures there will be employ- be principally affected by changes in the product ment contraction in the formal nontradables sec- and factor markets, they may also be affected by tor (for example, a cut in government employ- changes in the credit markets resulting from mon- ment) because employers have little alternative etary contraction. They will either face increased in the organized labor market but to "shake-out" quantitative restrictions in the market, so that they excess labor. The redundant formal sector work- simply cannot obtain the amount of credit they ers will now tend to move into the informal sec- need at the existing rate, or they will find interest tor. This will include a move into urban services costs increasing. Poor households have very lim- even though the demand for these services may ited direct access to financial markets anyway, so also have contracted. A major characteristic of the effects on them will depend largely on the the informal sector, however, is that it is a very extent to which the policy effects in the formal flexible labor market, so that a fall in demand market spill over into informal markets. Again we will tend to affect remuneration rather than em- see that the informal and parallel markets play an ployment. Thus the labor markets in the formal important role. In the pre-adjustmentphase, credit and informal sectors may reflect very different may be available in parallel markets but at a features. There will be wage resistance in the premium rate. Liberalization might bring with it organized sector, while the informal sector will somenarrowingof the differentialbetweeninterest act as the "residual pool" of labor during the rates, although tight monetary controls will gen- adjustment phase (Scobie 1989), with employ- erally keep rates high. Note also that the rationing ment rising and average income declining. It is of foreign exchange in the pre-adjustment era has also suggested that labor may flow more easily often created official and parallel markets in for- between formal sectors (nontradables to tradables) eign currency with multiple exchange rates that than between the formal and informal sectors havepersistedwellintopostadjustment. Of course, within tradables or nontradables. We often have it is not necessarily the case that changes in all 72 these markets will have a major direct impact on farmers in remote areas. It is clear that informa- the poorest households. Information in needed to tion is required on how these economic infra- monitor the changes and their influence. structural effects are likely to affect smnallholder productivity and income. A further point is that Infrastructure many of the effects may not be immediately ap- parent if the overall reductions have taken the It has already been asserted that economic and form of maintaining current expenditures at the social infrastructure forms a crucial part of the expense of withdrawing capital projects. meso economy in Sub-Saharan Africa. By "infra- Reductions in social infrastructure will affect structure" we usually refer to publicly provided households both directly and indirectly. A cut in physical capital and government services. The the provision of teachers, for example, will di- subdivision into economic and social components rectly affect the incomes of households whose comprise the following: members are teachers, but the cuts would also * Economnic infrastructure: major elements of affect those households who receive fewer (or physical infrastructure such as roads, irrigation, inferior) educational services. The indirect ef- and transport facilities, and current expenditures fects of human capital enhancing services (such on a range of support services such as agricul- as health and education) may be felt through a tural marketing and extension services lowering of labor productivity. The effects of * Social infrastructure: capital expenditures on reductions in social infrastructure, though possi- schools and hospitals and current expenditures bly more pervasive than economic infrastructure, on health, education, and nutrition. may disproportionately affect some groups. Ru- Bearing in mind the nature of the expenditure- ral areas may fare worse than the influential reducing elements of adjustment policies and the middle- and upper-income urban classes in re- limited scope for increasing tax revenues with- gard to both health and education services. Once out endangering the very objectives of policy, the again it will be a central aspect of the SDA pro- brunt of adjustment will tend to fall on reduction gram to seek systematically the information in government expenditure, and especially on needed to monitor the differential effects of re- infrastructure provision. Clearly, changes in the ductions in social and economic infrastructure - provision of economic and social infrastructure for example, whether there is any decline in health will have a direct and major impact on some facilities or in books available in education, and households, but only a limited and indirect im- so forth. Such data are not routinely collected pact on many others. In assessing the social di- and must therefore be sought as part of an over- mensions of adjustment it is therefore imperative all survey initiative at both the community and that these impacts be assessed in a methodical the household level. and consistent way. A direct effect of this overall reduction in eco- Data and information nomic infrastructure will be felt by households operating as producers. For example, a reduc- Information for meso-level analysis must be tion in physical infrastructure and other farm sought from several key sources. In all Sub- support services is certain to impose constraints Saharan African countries this will involve the on farm profitability and hence on the ability of introduction of new surveys to supplement farmers to respond to the improvement in agri- available data sources. Unlike the major macro- cultural prices brought about by structural ad- level institutions (central and regional govern- justment. These kinds of effects have been largely ment, central bank, customs and excise depart- ignored in the adjustment literature, and yet they ments) or the nicro-level institutions of house- are likely to be important for many groups of holds with which we are ultimately concerned, African smallholders. Furthermore, in many Af- the meso level is not observable in an institu- rican countries the physical infrastructure serving tional sense. We do not observe markets as many rural areas has seriously deteriorated, so physical entities and therefore we cannot di- that access to the necessary markets has become rectly derive information about how they func- significantly more difficult. In these circum- tion. At the same time, our interest in economic stances, improvement in prices in urban or and social infrastructure is not simply with re- periurban markets may be of little relevance to ductions in expenditures per se, but more in terms 73 of the flow of services derived from this expen- they are often inappropriate or inapplicable. diture. Hence, information has to be sought from As noted earlier, some sectoral production sta- individuals and households across all groups in tistics are routinely generated as part of the com- society to ascertain the overall impact of changes pilation of the national accounts. Individual line in infrastructual provision. ministries, such as agriculture, usually play a key At the national level, meso-economic data are role in this process. In addition to the require- available chiefly on product prices, wages and ments for estimation of the current value of out- other labor force data, and sectoral production. put, it is usual to find estimates of real or constant Information on prices has been accorded a rela- price output (or possibly just indexes) as well as tively high priority by national statistical offices, series of physical outputs for some selected com- especially following the widespread participa- modities. Various methods exist to obtain these tion by African governments in the International estimates, but there is a heavy reliance on mea- Comparisons Project (ICP). Although the prime suring physical changes with a single indicator of purpose in collecting this information has been to a key crop or product which is then applied to a compile purchasing power parities, there is usu- base year value. ally some parallel attempt to compute a consumer There are two principal survey initiatives that price index (CPI) to provide the basic measure of will provide critical data and informational re- a country's rate of inflation. The main problem quirements for analysis. The Household Surveys, here, however, is not so much the price data but carried out as a sequence, or some combination the determination of the weights of the index. In of a Social Dimensions Integrated Survey (IS) or a the past this has been one of the principal reasons Social Dimensions Priority Survey (PS) will not only that statistical offices have conducted household provide an invaluable source of information at expenditure surveys, and it clearly requires that the micro level, but will also provide some key expenditures be obtained at a substantial level of data about the functioning of markets and how commodity disaggregation. Producer price se- adjustment has affected the availability and qual- ries and import and export price statistics are ity of economic and social services to households. generally far less prevalent, except for a few ma- Some information on market behavior will be jor products such as those representing the prin- obtained from households concerning their roles cipal export commodities. Similarly, information as both producers and consumers of products. In on wage rates is virtually non existent, with a few addition, the household surveys will provide notable exceptions such as public sector occupa- some information about the operation of labor tions. markets. The IS is designed to provide informa- Data on employment and earnings can be derived tion on the household and its members across a from several possible sources. One major source wide range of topics in order to improve our is sometimes a purpose-built labor force survey. understanding of what determines the welfare of These are usually carried out fairly infrequently. the household. Thus, for example, the survey Yet when they are undertaken they clearly pro- should not only tell us that levels of education vide the very best information on labor force are low among household members, but it will participation, occupational structure, earnings, provide the basis for understanding why they and general labor force characteristics available. are low (for example, because of high opportu- A population census also provides some baseline nity costs resulting from returns to child labor or information on employment, as do some produc- low household income). The PS is less ambitious tion censuses and surveys. It must be noted that in what it sets out to achieve. Its principle objec- the information obtained from production sur- tive is to obtain, as rapidly as possible, informa- veys is usually meagre and it is rarely possible to tion on a critical set of welfare indicators. Be- utilize it in any subsequent analysis. Likewise, cause the range of variables obtained is more although some partial information on earnings limited than with the IS, the PS will only trace can sometimes be gleaned from income tax and what is happening to the indicators - it will be social security returns, these sources suffer heavily limited in its capacity to explain them. In both from a problem of bias and non coverage. Like cases, however, the household surveys will yield output, however, the measurement of employ- relatively little information about prices or wages. ment is fraught with difficulties brought about Another survey initiative, the Community Survey, by adapting concepts borrowed rather too freely is designed to generate information on the move- from the developed world to economies where ment in prices at a community or regional level, 74 and, just as important, information on how ad- band lives under a separate roof. justment has affected the supply of social ser- The basic United Nations' definition of a vices and economic infrastructure. Although the household, used in many surveys, is a "group of effect of changes in provision will be felt at the people who live and eat together." While this is a household level, information on health, educa- clear distinction between households and fami- tion, and other public services is sometimes best lies, there is a problem regarding individuals who monitored at a community level where there could board (eat with the household) and lodge (live be differential impacts of adjustment of the poor with the household). The general United Na- and the nonpoor within a community. tions guidelines suggest that the boarder who does not lodge should be included, but the lodger Information requirements at the micro level who does not board should be excluded; thus those who board and lodge, such as domestic We now examnine the particular requirements at servants, are regarded as household members. the rnicro level and its relationship with the so- In many cases this could establish a household cial dimensions of adjustment. Four issues are unit that is too heterogeneous in its composition. discussed below. The first concerns the nature of The degree of intra-household inequality could the household as a unit of study. Second, we be too great and could cloud many of the impor- consider the question of grouping households tant social dimensions. So a second criteria can and identifying target groups. Third, we con- be added: that all persons living and eating to- sider the range of data constructs currently avail- gether should acknowledge the authority of a able for indicating levels of welfare. Finally, we single head of household, regardless of whether briefly discuss the specific data requirements for the latter is living with the other household micro-level analysis. members or living away. Hence the decision as to whether servants and lodgers are considered The household unit to be household members or are a separate household depends on whether or not they ac- Individuals are usually members of several over- cept the authority of a household head. In the lapping social networks at the same time. The African context the situation of polygamous nuclear and extended families are obviously two households can present problems, depending on social units; the household is another. A house- whether each wife is treated as a separate house- hold can be a one-person or multi-person unit. A hold or as a part of one large household. In the number of criteria can be used to define the former case it is necessary to apply an arbitrary household. Those commonly utilized include: rule, such as linking the household head to the members who have a common source of major senior wife to avoid double counts. income, live under the same roof or within the The household is an important social and eco- same compound, and have a common provision nomic unit because within it many of the deci- for other essentials of living; (Casley and Lury sions concerning individual members' activities 1987; United Nations DTCD 1989a). While the and their consumption (and hence their welfare) criteria used to identify households must be cho- are made, and its physical properties - that it is sen to suit the local situation, a basic distinction a collection of individuals with an identifiable should be drawn between the household and the location- make it a useful sample unit in survey family unit. The latter would involve imposing work. Nevertheless, the household may not al- an additional criterion of kinship, where mem- ways coincide with a single economic unit. In bers are related by blood, marriage, or adoption. this regard we can distinguish two kinds of eco- The size and characteristics of households can nomic units: consuming and producing. As a show wide variations by locality and country. consuming unit, living and eating together might They could consist of a single family, but house- not imply the pooling of all incomes and sharing holds in Africa commonly consist of several all expenditures. Even if it does, and the house- families, whether these are of the same kin or hold constitutes a family unit, two possible con- persons with no kinship ties. It is possible for cepts of household decisionmaking based on the families to be spread among households, either "glued-together' or "despotic" family unit can temporarily or permanently. For example, a mar- lead to widely different results on the intra- ried woman may continue to live in her father's household allocation of resources and, hence, of household when she is young, while the hus- individual well-being. The problems stemming 75 from the household's role as a producing unit lar dimensions - such as the role of women can be even more problematic. Small-scale agri- during an era of adjustment - that otherwise cultural holdings of nonfarm enterprises may not might not emerge in a household-level analysis. match the household unit. The agricultural hold- ing is usually defined as the land and livestock Classifying households and identifying target groups that is managed as a single unit by a holder, who may not be the head of the household. For ex- The classification of households by socioeconomic ample, in many countries the wife of the house- group constitutes an essential element in data hold head is allotted a certain area of land on analysis at the micro level. Such classifications which she grows a limited number of crops tradi- are important if policymakers want to know how tionally cultivated by women. Also, nonfarm shifts in macroeconomic and sector policies actu- enterprises may be operated by more than one ally change the income composition of various household, which creates a problem both in iden- social groups, and thus determine how their well- tifying and attributing the incomes generated. In being is affected. In theory, any number of crite- view of this nonalignment between the house- ria might be used to establish the classification of hold and economic units, in some surveys in Af- socioeconomic groups, but the basic requirements rica alternative institutional units have been used, of a useful criterion is that it should be unam- such as individuals living in a hut in the village biguous and have a clear policy focus. Apart or cattle post (Botswana) and a homestead from the obvious importance of policy targeting, (Swaziland). Also, special arrangements may household classifications ought to be chosen so have to be made to definine appropriate units for that they are relatively homogeneous. Hence, nomadic societies (United Nations 1989a). Not- variations (for example, of incomes or expendi- withstanding these complications and exceptions, tures) within groups are smaller than variations the household unit is to be regarded as the basic between groups. In this way we have greater preferred institutional and sampling unit, al- scope for analyzing behavioral responses to policy though its precise definition has to remain a changes. Finally, since households are most often country-specific choice in the light of local multi-individual units, a classification should circumstances. ideally be chosen that is applicable to all indi- A further decision about household composi- viduals in a given household or, alternatively, to tion has to be made - whether to adopt the defacto the household as a whole; otherwise, the funda- or de jure approach. The de facto definition per- mental notion of the household being a "unit" is tains to those household members present at the lost. time of the survey and is most commonly used Criteria for enumerations over short periods. The de jure definition relies on a concept of normal residence A broad set of criteria has been used for classify- and is usually preferred in surveys conducted ing households, including economic (wealth, in- over a longer period of time. come or expenditure); sociological, location, and In spite of the mnany good and convincing rea- characteristics of household head.62 sons for choosing the household rather than the individual or any other social grouping as the WALTHm,iNcoME,oRExPwrDmIRs. Althoughthese appropriate unit, there are still many problems in are included in a combined category of "eco- deriving inferences about the individual's well- nomic" criteria, each one has a different merit. being from household-level responses. Prefer- Wealth is important at several levels. In rural ences of household members could differ widely, areas access to land is a critical consideration. The as could the basis for allocating resources among landless or near-landless households can be af- them. For example, in certain situations we may fected quite differently from the smallholder by observe an increase in household income (per- development policy. The rich, measured in terms haps under adjustment) that does not generate of accumulated wealth in the form of physical or an improvement in the nutrition of all household financial assets, are clearly a distinct group; but members. An assessment of the social dimen- even among the relatively poor, those with mod- sions of adjustment that is primarily carried out est assets might be distinguished from those who at the household level could therefore mask a have none. Wealth, as with income or expendi- number of consequences at the individual level. ture, has the advantage of being a household-level For this reason it is important to monitor particu- criterion. Nevertheless, the use of income, expen- 76 diture, or, to some extent, wealth as a classification * pastoralists criterion suffers from a major drawback - ac- * landless (or near landless) agricultural cording to any of these criteria, a household's workers relativeposition,andhenceitsclassification,nmight * non-agricultural workers. change over time or as a result of policy inter- Urban sector vention. For instance, the mobility of households * government employees (skilled) between income deciles makes total income a * government employees (unskilled) poor classification criterion for targeting policy * private formal sector employees toward particular households. The wealth crite- * private sinformal ector employees, self-em rion is much more effective because households ployed are relatively less mobile between wealth groups * inactive or unemployed. in the short or medium run. The purpose of defining such classifications is to facilitate the design of policy interventions, SOCIOLOGICAL. These criteria include a range of which frequently affect the socioeconomic groups factors such as race, religion, or language and in different ways. Policymakers may define target assume a particular importance in African socie- groups of special concern, and the selection of ties where market fragmentations or even ethnic socioeconomic categories obviously has to reflect discrimination might be a common characteristic these target groups. It is important to distin- of the majority of the poor. guish, however, between the concepts of socio- economic group and target group. In the first LOCATIONAL. Location is usually justified on the place, socioeconomic groups are defined to cover grounds that policy often has a locational ele- the whole population; morever they are mutu- ment. Rural households need to be distinguished ally exclusive. In other words, each person (or from urban households but, beyond this, our fo- household) in the population belongs to only one cus on the meso economy has indicated a strong socioeconomic group. On the other hand, people spatial dimension in the way policy is transmitted may belong to more than one target group, and through markets and infrastructure. Thus, it of course many will not belong to a target group might be important to use an even finer locational at all. Target groups are neither exhaustive nor division for classifying household groups and to mutually exclusive. A target group may be a capture the regional effects directly. subset of a broader socioeconomic group - for example, maize smallholders may be considered CHARACrRSTICS OF THE HOUSEHOLD HEAD. The a target group and part of the broader socioeconomic characteristics of the household "smallholders" socioeconomic group. Alterna- head (for example, occupation or employment tively, a target group may cut across socioeco- status) are often used as criteria for classifying nomic groups, as in the case of female-headed households. In doing so one is implicitly assum- households as a target group. ing that the behavior and level of well-being of While it is important to maintain this distinc- all individuals in the household can be deter- tion between socioeconomic and target groups, mined or adequately represented by the status of the choice of socioeconomic categories must re- the head. Nevertheless, the economic status of flect the groups that are targeted for policy and the household might be determined by the char- program purposes. Hence, socioeconomic groups, acteristics of a main earner who could be a differ- representing the basic taxonomy for data classifi- ent individual from the household head. So this cation, should be carefully selected to allow the criterion has to be used with much care in its identification of target groups for policy and pro- practical application. gram intervention, as well as representing In the case of Sub-Saharan Africa, an indication groupings of households that are relatively ho- of some broad categories of socioeconomic groups mogeneous in levels and sources of income, ex- that might figure in a number of taxonomies is as penditure patterns, and economic behavior more follows: generally. If the classifications are too finely di- Rural sector vided, however, then the analyst could run into * export-oriented medium- and large-scale empty-cell or small sample problems. At the farmers same time, some nontargeted household groups * export-oriented smallholders might be aggregated into fairly broad categories. * food/subsistence farmers Nevertheless, it is not always obvious prior to 77 formal modeling or analysis just how important We have seen that some evidence can be gleaned the behavior of nontargeted groups may be at the micro level to derive core macro indicators, through their indirect effects. This suggests that but very little of real substance can be established great care must be exercised in selecting the ap- about the social dimensions of adjustment unless propriate household classifications. The ques- reference can be made to the effects of adjust- tion of target groups is a major issue in its own ment on particular groups in society. Analysis of right and requires further consideration. micro-level data and information will therefore Concern for the social dimensions requires a constitute a major part of the empirical effort. clear picture of particular socioeconomic groups' The present section will consist of a brief over- vulnerability to adjustment programs and the view, relating especially to empirical indicators impact of adjustment on the existing pattems of of welfare, the measurement of poverty (includ- poverty found in a given country. In turn, there ing the derivation of poverty profiles), and an is a direct consequence of this implication in the indication of how the effects of adjustment might information requirements and analysis at the be accounted for at the micro level. This section micro level. will be confined to the question of measurement and the information requirements that it implies, Data sources rather than analysis per se. A first objective in performing empirical analy- Household- and community-level surveys pro- sis at the micro level is to be able to identify the vide the main data and information for house- poor and their main economic activities. This hold-level analysis, especially analysis to help should lead us directly to the identification of the identify the incidence and intensity of poverty target group for policy purposes. Having already and the socioeconomic characteristics of the poor. established the household as the unit of analysis, The SDA is a multi subject survey covering in- a measure of welfare is required to identify the come and expenditure, education, health, em- poor. Many of the theoretical arguments relating ployment, migration, housing, agriculture, to the determinants of welfare at the individual nonfarm enterprises, transfers and savings, and household level have already been discussed in anthropometry. The preliminary sections of the Chapter 4 and need not be repeated here. But questionnaire also generate some basic informa- there are a number of implications concerning tion about household membership, including the empirical measures of the standard of living number of individuals, gender, age, relationship that have been proposed that require further con- to household head, and the like. This informa- sideration in the light of survey and other data tion provides the key source data, but the esti- requirements. It is manifestly clear that no single mates and the indicators required for analysis "scalar" measure will suffice as a unifying meas- must then have to be constructed. For example, ure of household welfare. Each measure has its household income estimates, which constitute relative merits and drawbacks. The wide scope some of the key statistics for poverty analysis, of the multisubject household survey will there- have to be built up and derived from the source fore permit the computation of many alternative data provided. The same is true regarding ex- measures, including those based on income, ex- penditures, although they are generally less penditure, and nutritional status. problematic in concept and usually more reliably reported than incomes. The IS provides a wide INcoME. Income-based measures appear, at first range of additional information to facilitate a sight, an attractive way of assessing relative number of alternative classifications based on the household living standards. The level of house- characteristics of the household as a whole, the hold income represents the limit to which house- household head, or of other individual house- hold members can acquire goods and services hold members. without running down their assets or increasing indebtedness. We have noted in the earlier dis- Indicators of welfare cussions in this volume that households could receive income from a variety of sources and that At the micro level of analysis there are many we may distinguish two broad classes: primary analytical and empirical issues to be tackled, and income and secondary income. 'Primary" in- therefore the data and information derived from come is the term usually referred to as factor the multisubject household surveys are crucial. income, whereas "secondary" income consists of 78 transfers from government or from other house- income and an expenditure at a value in excess of holds and institutions. Let us distinguish the its apparent price (Meerman 1979) if income is to income categories more precisely: be considered as an appropriate measure of the Primary income standard of living. This is difficult to achieve in * income from employment (wages, salaries, practice, in part not least because of the multi- and the like) tude of factors underlying the effect of quantity * net income from agricultural holdings restrictions as well as price distortions, which * net income from nonfarm enterprises would affect the imputation procedure. * rent from dwellings A well-known practical deficiency of income- * consumption of home production. based measures of household welfare is the prob- Secondary income lem of underrecording income in household sur- * transfers from other households veys. There are a number of points here. * transfers from other enterprises and from Underrecording of incomes can be of two kinds: abroad either incomes are deliberately underreported by * transfers from government. the respondents or, the survey may fail to pick up Income is not received by all households from certain components. A special effort has been each and every source. Indeed, it is precisely made in the prototype IS questionnaire under the because households receive different combina- SDA program to minimize the latter by seeking a tions of incomes that our understanding about response even when the household and the pro- the effects of adjustment on different social groups duction unit do not coincide. But no matter how is complicated. Income from employment (em- complex the question paths, it is difficult to over- ployee compensation) is payable to individuals come the former problem. Second, the in the household, and there may be more than underrecording problem appears to affect certain one employee, each of whom could have different income components more than others. For in- skills and belong to different labor markets. Net stance, net income from household enterprises income from household-based activities (farm and (and, hence, the informal sector) appears to be nonfarm) generates returns to both self-employed particularly unreliable, as does the reporting of and domestic productive assets. Rent from secondary incomes of all kinds (Grootaert 1986). dwellings is included, and this covers both actual and imputed income from owner-occupied ExPENDYrURE. Expenditure-based measures of dwellings. The latter is important, especially in the standard of living are considered both more the case of cross-sectional analysis, otherwise reliable and more satisfactory than income-based rental income (and expenditure) could distort the measures from several standpoints. First, it can patterns across households. The final element of be argued that expenditures follow movements primary income is also an important item - the in permanent incomes, so that households save estimated (equivalent market) value of home- or dissave out of transitory fluctuations in their produced goods and services, which augments actual income. Second, expenditures are much both the incomes and expenditures of certain closer to the notion of "consumption" of goods households. This principally affects farm house- and services (and hence the satisfaction of hu- holds. man wants) than incomes. Third, expenditures Secondary incomes consist of transfer incomes, tend to suffer less from underrecording prob- which are shown above as three separate compo- lems and people may tend to have a better recol- nents, depending on their source. It is usually lection of their expenditures than of their incomes, difficult to distinguish such income in the re- especially if they are asked precise questions about sponses from a household survey. From the ana- particular expenditure categories. Nonetheless, lyst's point of view, it would be useful to be able there are difficulties and drawbacks with expen- to estimate government transfers in cash and in diture measurements. Some expenditures are at kind because there could be a direct effect on the household level, while many others are at the their magnitude as a result of structural adjust- individual level and must be aggregated. Also, ment. There is a further conceptual issue con- there is a problem with expenditures comparable cerning the extent to which one should make with that in the measurement of income concern- imputations of certain government transfers to ing the imputation of certain items. households. Free or subsidized health care ought, The main categories of expenditures can be in principle at least, to be imputed as both an itemized as follows: 79 Household expenditure anthropometric and clinical measurements of in- * expenditure on food items dividuals in a household, or perhaps of some * expenditure on nonfood items subset, such as children. "Input" measures pri- * consumption of home production marily relate to food intake. * rent for housing services * transfers to households, government, and Data for micro analysis other institutions, domestic or abroad. To identify the type of information to be collected The item "rent" is identified separately to ac- through household surveys and the related mea- count for the possibility that if rent is imputed a surement issues, attention must be paid to the an income it should also appear as an expendi- category of respondents from whom it may be ture for the household. A simdlar remark applies realistically collected. It is also useful to distin- to the imputation of the consumption of home ish between the of information that can production, which particularly applies to subsis- be obetwen a type ofainf(either can In som isac s,nlsshv be obtained on a single occasion (either because it tence farmers. pn . . aertains to the current situation or because it can preferred to base measures of standards of living on on copnn°fepniue o x be obtained realistically through retrospective on one componeendieture s - for ex- questioning relying on the respondents memory) structsu ate p ropor of therhouh and the type of information that by its very na- budget spent on food. ture requires repeated visits to the same house- hold or individual. Most suitable reference peri- In the case of both income and expenditure ods will also vary among different items. Such measures there is a problem about how to account v fo th aibehueodsz. Soenoml factors can be helpful in appropriately organiz- izationris nhecessaryan variabl e ods pos ies . ar ing the information to be obtained in the survey -.zation is necessary and various possiblities are qusinnies and dem-ngthaprri suggested, but they usually amount to a choice of que stnrve(s)randudee ate survey structure. whether to measure expenditure per capita, per y adult,or pe adlt Th difrec Several categories may be distinguished in the equivalent . context of the household surveys concerning the between the latter two normalizations is that the "pe eqialn adlt is base upn"qia type of units to which the information to be col- idper equivalent adult" IS based upon "equiva- lence scales," which require econometric estima lected pertains and the type of respondents from tion and are not guaranteed to give usable results which it may be collected: ,a . ..and Muellbauer 1980), while the "Per * Simple information that canbe easily obtained a.tn , a . on a relatively extensive scale by observation or a adult"sis aemore orbhitdra approc exanelies one brief interview with any member; this can in- choosing weights for children (for example, one duelsigohueod,enmrtgter child equals one-half an adult unit). With these clude lhisting of households, enumerating ther normalizations~~~~~~~~ wecngnrt.eea lea basic characteristics for stratification and sarn- normalizations we can generate several alterna- tive measures based on incomes and expendi- pling, as well as for identifying target groups of ture, such as total expenditure per capita, total special interest in the survey household income, proportion of the household * General information on the household such budget spent on food, and the like, as housing conditions, amenities, possession of durables, and the like - also obtainable from HEALTH AND NUTRITIONAL STATUS. There is sub- any adult member but usually collected only for stantial literature on the possible effects of ad- the households selected for the sample justment on the health and nutritional status of * More specialized and complex information individuals (Behrman 1988; Cornia, Jolly, and pertaining to the household as the unit, such as Stewart, 1987). In principle, measures of living consumption and expenditures, typically requir- standards based on health and nutritional status ing a lengthy interview with specified member(s) are attractive because they directly reflect basic of the household judged to be the most well in- human needs. But it is complex and difficult to formed on the topic analyze the effect of adjustment in these dimen- * Information on production, inputs, income, sions either from a theoretical or from an empiri- and the like pertaining to each household enter- cal standpoint (Behrman 1988; Scobie 1989). prise (agricultural and nonagricultural) as the Concerning measurement issues, there are essen- unit, again requiring interviews with individuals tially two approaches: measures of outputs and identified to be the best informed of inputs. "Output" measures cover various * Identification of household memnbers and their 80 basic demographic and other characteristics, volve a combination of more formal modeling usually obtainable as a part of the general house- techniques to investigate macro-meso linkages, hold interview with any adult member through which counterfactual experiments would * Detailed personal information on each indi- be feasible, with survey-based meso-micro vidual member, such as on employment and in- analysis of household welfare. come, usually only obtainable directly from the This is the underlying philosophy in what fol- person concerned lows. Our first preoccupation is with establish- * Information pertaining to other types of units ing methods of analysis to gain an understand- outside the household, such as the community, ing of macro-meso linkages. This is then comple- local markets, other organizations, and institu- mented with a discussion of the different types of tions - usually obtained from specially selected meso-micro analysis that are likely to prove use- 'key" respondents. ful for policy purposes. Toward macro-meso-micro analysis Macroeconomic modeling The collection and collation of relevant and timely The purpose of this section is to provide a brief data are critical prerequisites of a policy initiative review of some of the modeling approaches that to enhance the social dimensions of adjustment. could be used to examine the macroeconomic But they are really only the beginning. The major consequences of structural adjustment and, more challenge such an initiative faces is in under- specifically, to explore the effects on particular standing what it is that determines the mi- socioeconomic groups. For a more detailed dis- croeconomic (or social) outcomes that are ob- cussion and assessment see N'cho-Oguie (1989). served. In particular, some understanding has to As indicated earlier, an ultimate aim is to work be gained of how structural adjustment policies, toward an integrated macro-meso-micro model- both those that have been applied historically ing and analytical framework. At this stage, and any alternative sets, have affected house- however, the state of the art is such that it is not holds. This is a challenge of significant propor- possible to propose a definitive model or even a tions, since the analysis that is applied to the data class of models capable of tracing the effects from must be ultimately capable of performing the macro level right down to the micro level of "counterfactual" experiments by tracing what individual households. Moreover, even at the might have occurred had an alternative set of macro level it is unlikely that one simple model policies been applied and tracing these effects specification would be suitable for all purposes. through to households in sufficient detail to be of Usually macro models are designed to focus on some practical use to policymakers. particular macroeconomic features and to sim- The first of these capabilities implies some form plify the rest. It is therefore essential to develop of structured and quantitative appraisal based on and maintain an information system capable of a model (or models) of the economic and social servicing a range of possible models. Our aim system. Since most models that satisfy the first of here is simply to note some of the existing model- these capabilities are limited in the degree of ing options and their applicability to social di- disaggregation that they can meaningfully handle, mensions issues. they need to be supplemented with complemen- Before embarking on this review a more imme- tary data analysis that can meet the second re- diate question needs to be considered: "Why quirement. It may be that empirical household construct a macroeconomic model at all when models can assist in tracing the household effects the ultimate concern is the micro level?" The in sufficient detail for policy purposes, but these answer to this is a standard one: the nature of the are not the only approaches available. A macro- instruments of an adjustment package (for ex- meso modeling initiative can be complemented ample, manipulating the interest and exchange by a careful meso-micro analysis of household- rates) are manifestly macroeconomic, even if our level data directly without any formal household concern is to examine the effects on particular modeling. In this way, macro-meso linkages can household groups. Furthermore, besides the be examined and at the same time the implications countries themselves, international organizations for different households can be worked out in including the World Bank and its donors are also sufficient detail for policy purposes. In other interested parties in assessing the broad economic words, a useful research methodology would in- and social effects of these policies. 81 Aggregate models come-expenditure loops are combined with the standard inter-industry multipliers to generate EcoNoMErRIc. It is sometimes suggested that each consistent income and output effects of any pos- country should maintain a fairly aggregative ited exogenous changes in final demand. For econometrically based macroeconomic model for example, suppose we want to know the distribu- short-term forecasting purposes. Without enter- tional consequences (that is, on incomes of differ- ing into a debate about the efficacy of building ent household groups) of a reduction in govern- such models for policy analysis in Sub-Saharan ment expenditure of 10 percent and/or an in- African countries, it should be noted that models crease in export demand of 15 percent. These based on time-series estimates of key parameters models work by attempting to trace the conse- are likely to be unreliable. The policy shifts are so quences on incomes by assuming that any change dramatic that many parameters will not remain in demands for products is translated into a pro- constant, and even if time-series data span the portional change in outlays, inputs, and outputs period of policy switches there are unlikely to be in accordance with the patterns that existed in sufficient observations to satisfactorily estimate the base year structure of the SAM. the structural breaks. The simple conclusion is The principal appeal of fixed-price multiplier that aggregate macroeconometric models are models is that they are simple to understand and probably unsuitable for SDA purposes. are similar to widely used input-output models. They are highly intuitive in the way the repercus- WORLD BANK RMSM. This family of World Bank sions are traced around the system. All of this "Revised Minimum Standard Models" are rudi- makes them especially appealing to nontechnical mentary country macroeconomic projection tools policy analysts. Against this, however, there are built around the macroeconomic balances and significant drawbacks, especially to their ability detailed country debt data. They are designed to - or inability - to track the effects of adjustment project a country's external resource needs based policies. There are three main points here. First, on alternative GDP and export growth scenarios, while multiplier models are quite robust under together with the resulting debt servicing pros- conditions where there are no capacity and skill pects. Their principal virtue is that they are appli- constraints and where income transfers conform cable across countries and therefore are "mini- to constant patterns, such assumptions are not mum standards" in the modeling sense. Never- necessarily valid for economies undergoing ma- theless, they are not flexible in prices, they have no jor structural adjustment. Second, they are price household sectors, and they are simply not de- insensitive, which means they are not at all ca- signed to simulate structural adjustment policies. pable of addressing the key question of the pos- Currently there are attempts within the World sible consequences of shifts in the relative prices Bank to develop a standard macro model beyond between, for example, tradables and nontradables. RMSM that introduces prices and accommodates Third, multiplier models rely on our ability to features of the open economy and yet retains a specify changes in real final demands, which is a medium-term perspective. While this may make doubtful exercise in its own right. the model more suitable for analyzing adjust- ment, it would still be deficient in capturing the Macro-meso modeling distributional dimension and is therefore likely to remain inappropriate for our purposes. The existence of meso-level activity, particularly the roles played by markets and infrastructure, SAM-based fixed price models creates a special set of challenges for our quanti- tative analysis. The fundamental question to be The basic structure of a SAM, whereby one can addressed is how to determine empirically the trace the sequence of flows around the economic effects of macroeconomic policy on individual system, has led to the evolution of a class of market conditions and economic and social infra- models that rely on fixed price, fixed coefficient structure. As in the case of our discussion at the assumptions. Thorbecke (1985) has described macro level, the issue is much broader than just such models as "first generation" SAM-based seeking to establish a particular modeling ap- models. In essence they are simple extensions of proach. The ultimate aim within the empirical the input-output model whereby Keynesian in- context should be to set up an information base 82 that is capable of accommodating a wide range of commodity prices. But CGE models are suffi- analytical and modeling options in recognition of ciently flexible to allow for rigidities in some the considerable variation in the circumstances markets, as well as for alternative closure rules in of each country and in the type of analysis that the system as a whole. needs to be undertaken. CGE models are more suited to medium- or It would be ambitious to expect too much from long-term analysis and do not purport to parody a formal model of meso-economic variables. We the actual movement from one equilibrium to have already noted the sectoral shifts that take another. This is why they are usually viewed as place during a period of structural reform and being comparatively static in nature. There have the probable emergence of private sector activi- been some attempts, even in the earliest models, ties in small-scale agriculture, urban services, and to make them more dynamic by embedding the transport. Such dynamics will affect market con- CGE component within a two-stage framework, ditions but are very difficult to capture in an where the second stage generates the "between explicit modeling sense. Nevertheless, there are equilibrium" shifts. This can mnake the models two quite fruitful modeling approaches that can more useful for considering short-run distribu- be referred to in this context, each of which may tional consequences, and more applicable to the play a constituent role in analyzing the impact of analysis of adjustment policy issues (Michel and adjustment. Models can be useful in performing Noel 1984, Dervis, de Melo, and Robinson 1982). counterfactual experiments in a comparative static SAM-based CGE models have a number of at- setting, but it should be emphasized that, given tractive features for analyzing the effects of ad- the inherent complexity of the meso level, there justment at the macro level, and this appeal ex- is much to be gained from just observing out- tends to their ability to capture some meso- and comes, computing deviations from trends in the micro-level features as well. By their very nature intervening variables, and carrying out a direct CGE models focus directly on the market-clear- analysis of the results. ing behavior of economic agents in the system SAM-based computable general equilib.ium (CGE) (households, firms, and the like) and on the out- models comes in the product and factor markets in par- ticular. They are therefore directly concerned Computable general equilibrium models can best with the operation of markets, one main ingredi- be viewed as a further stage in the evolution of ent of the meso economy, and they therefore po- the class of models based on the structure of a tentially provide an important set of signals about SAM, although their initial development predates the effects of adjustment at the meso level. Of much of the work on SAMs (Adelman and course it immediately follows that the richness of Robinson 1978). In their most elementary form a CGE model in this regard depends critically they may be characterized as single, within-pe- upon our ability to define the important markets riod, comparative static models in which there is and quantify the mechanisms - "important" in an explicit representation of the markets for factors the sense that the outcomes will tell us what we and commodities so that supplies and demands need to know about the effects on household adjust to external shocks through changes in rela- groups at the micro level. For this, the design of tive prices. Within this broad characterization the underlying SAM and our ability to derive the there is a wide range of variants, including differ- requisite information to quantify it becomes a ent specifications of behavioral relationships, crucial factor in the process. In principle, any closure rules for the markets, and degrees of dis- number of different product and factor markets aggregation of the factor, commodity, and can be specified within a CGE framework. In household categories (Robon, 1989). The behav- practice, however, the limiting factors are the ioral equations describe various agent behaviors, availability of data and computing capacity and such as factor supplies by institutions (especially capability. It is almost essential that models be of labor services by households), consumers' ex- manageable enough to be solved on a PC since penditures, factor demands, and commodity mainframe capacity is severely limited in most supplies. Most important, these relationships are African countries. Models do not have to be usually price sensitive, so that factor markets, for large to be useful for meso-level analysis. It is example, can be cleared by adjustments in factor much more important to focus on the variables prices and commodity markets by changes in and markets that are most likely to affect the 83 poorest household groups. This means identify- numerical estimates, one can choose functional ing the commodities which these households forms and specify them in the cells of the matrix produce as well as consume and the labor mar- and let the software solve the model (Drud, Grais, kets in which household members trade their and Pyatt 1986). Moreover, if it is really desirable services. If this is not achieved, the CGE model to generate a fixed price multiplier model, this will not capture the meso level as we have de- could be obtained as a special case in which all fined it. substitution elasticities are set to zero. Hence the It has been noted that CGE models could cap- scope for CGE model specification within a SAM ture some, though not necessarily all, the required framework is now very wide indeed. meso-level features to track the movement of the A recent and quite promising macroeconomic important intervening variables. Let us briefly simulation model, combining some features of indicate some deficiencies that will require spe- computable general equilibrium models, has been cial attention, and possibly some further research. designed within the OECD Development Centre First, changes in infrastructure can be captured with a view toward its general applicability to a only in an indirect way. Many of these changes relatively wide range of countries (Bourguignon, are exogenously determined and are indepen- Branson, and de Melo 1989). This would be dent of the sorts of market forces a CGE model is achieved by changing the institutional character- designed to capture. Second, the whole question istics that describe commodity markets (for ex- of the response of the informal sector to adjust- ample, supply and demand elasticities, price for- ment policies is difficult to address in the context mation and quantity dearing), financial markets of a stylized model of this kind, and still less the (such as, credit rationing, foreign exchange con- nonmonetized and nonmarket economies that trols), and labor markets ( extent of wage flex- probably respond to entirely different sets of sig- ibility, for example). Most important, it purports nals. Third, markets may behave very differently to link the short-run impacts of macroeconomic at the regional or community levels than they policies that affect the distribution of income appear to do in aggregate, and our modeling through inflation, interest rate, and other asset capability does not stretch to cover spatial vari- price changes with the medium-run impacts of ables and geographical markets. Nevertheless, structural adjustment policies arising through and in spite of these limitations, CGE models are relative commodity and factor price changes. a promising approach for linking the macro and As a general class of economywide model, CGEs meso level of analysis. certainly seem to offer promise, not least because The data requirements for building a CGE model of their ability to incorporate some macro-meso are almost entirely fulfilled by the information characteristics and their reliance on micro-level available in a SAM, and this is now seen as the optimizing behavior at the household and firm best possible framework to calibrate CGE mod- levels. Nevertheless, while some countries may els. It can be noted that CES (constant elasticity already be at the stage in the development of a of substitution) functions are widely (though not statistical base to be able to implement models of exclusively) used to specify the behavioral com- this type, others are not, and for them it may ponents. They have the virtue of being a flexible have to remain a longer-term objective. CGE functional form that allows patterns of responses models should thus be viewed as just one of a to be sensitive to changes in relative prices. range of modeling options that can be explored Generally the substitution elasticities are obtained on a country-by-country basis to develop an on the basis of previous experience rather than analytical capability for monitoring the social di- direct estimation. Other than this, the main sta- mensions of adjustment. tistical requirement is to obtain the most appro- priate classification of accounts and correspond- Multimarket models ing estimates of transactions in a SAM so that the base year shares can be computed directly. CGEs A number of detailed models have been con- are often criticized as being too complex for the structed to analyze the impact of policies in a user to maintain an intuitive grasp, but with particular sector or group of sectors. Perhaps the modern software capability even the model itself most interesting and useful of these is the agri- can now be specified within the strongly intuitive culture sector - or multimarket - model framework of a SAM. Hence, in parallel with the (Braverman and Hammer 1988). This model is 84 built around the characterization of market equi- tion assumed in the model; second, modeling librium for a set of interrelated commodities and usuallyrequirestheanalysttoassumethatwithin- is based on a system of behavioral equations for group income and expenditure variance is con- both the demand and supply sides. A critical stant (or that it bears some constant relation to the aspect is the specification of the way markets group mean). For most policy issues concemed operate, and in this regard the models are flexible with households, therefore, we must go beyond enough to allow for either price or quantity ad- formal economy-wide modeling. justment as required by the analyst. There are The objective of the exercise is to trace how also substitution possibilities in both supplies and macro policy and sectoral policy interventions demands that make these models especially suit- influence household welfare through their effects able for application to a very detailed set of com- on markets and infrastructure. Assuming that modity markets. As already indicated these there exists a reliable household data set, the first models are currently restricted to the agricultural priority is to establish which dimensions of house- sector and have been designed to simulate the hold welfare are to be the subjects of meso-micro effects of agricultural price, tax, and subsidy analysis. We have already highlighted (both ear- policy reform on supplies, demands, incomes, lier in this chapter and in Chapter 4) the complex and fiscal and external resource gaps. Conse- network of factors that govern household and quently, effects on the rest of the system are ex- individual welfare, so that too narrow an ap- cluded. proach (concentrating, for example, only on in- Although these are strictly sector models, they comes) is not warranted. Meso-micro analysis, do have certain meso-level features because they therefore, must feature a variety of variables to focus on market responses in key product mar- capture properly some of the complexities. kets. Their main strengths are that they allow for The basic indicator of welfare emerging from relative price changes and show some distribu- our earlierdiscussionis income (ortotal household tional effects of price and trade policies, and that expenditure) per capita. This should be used in these effects can be traced at a more disaggre- the first instance to identify which households are gated level than would be permitted by a CGE. poor. The preparation of a poverty profile based There has also been a suggestion that they might on household income and expenditure orderings be extended beyond the agricultural sector to should be the starting point of a country-based include urban services. But against their obvious analysis program into the social dimensions of analytical and practical appeal, these models do adjustment. Similarly, income- and expenditure- have certain limitations. To begin with, they have based indicators might be computed for other a weak macroeconomic link and are not well groups of policy concern, such as smallholders or suited to the three-level "macro-meso-micro" pastoralists. analytical framework. Second, because of its par- But analyses based on these indicators must be tial equilibrium nature, unlike the CGE model complemented with more in-depth studies cover- there is no guarantee that the multimarket model ing other aspects of welfare. Analyses of the will provide a consistent closure of the commodity human capital implications of adjustment merit and factor markets. priority, with high returns expected from studies of the implications of adjustment for the health Meso-micro analysis and education of household members. Similarly, an analysis of the nutrition effects of adjustment The main strength of formal economywide mod- policy should have priority, especially in Africa. eling techniques lies in their capacity to perform Other topics that also deserve explicit attention in counterfactual experiments and to take into ac- the context of the social dimensions of adjustment count how household behavioral responses can include employment, earnings, and the role of havehigher-ordereffects on markets. These mod- women. Each of these topicsreflectsan outcome at els, however, are usually too aggregative to be of household and individual levels that can be said practical use in addressing micro-level policy is- to reflect welfare. sues. There are two broad limitations to note:first, A simple schematic presentation of the main the number of socioeconomic groups that an elements of meso-macro-micro linkages is given economy-wide model can meaningfully distin- in Figure 6.1 (overleaf). The purpose of the exer- guish is limited by the overall level of disaggrega- cise is to trace how macroeconomic interventions 85 Figure 6.1 Schematic presentation of macro-meso-micro analysis Adjustinent Devaluation |Trade policy Fiscal/monetary policy Market reforms |Other institutional l l l l | ~~~~~~ ~ ~~~~~~~~~~~reformsl F-m-kets ~~~~~~~~~~~~Instrastructure Labor Product Credit Sodal Econo3ic Household Household Household income assets expectations |Dernand factors Supply factors| influence various welfare outcomes at the house- understand these effects, a distinction has to be hold level, outcomes suchashealthandeducation made between demand and supply factors that status, poverty, women's status,and so on. Macro affect outcomes at the household and individual and sectoral adjustment policies are listed in the level. For example, changes in markets and social upper boxes, and these are shown to have effects infrastructure will influence the effective supply on both markets and infrastructure - the critical of education services that are available to house- elements of the meso economy. Both market and holds (labor market changes may influence the infrastructural changes affect households in a availability of teachers, prices of books will influ- number of ways, critically changing their asset ence the quality of education services, investment holdings, incomes, and expectations. In the case of in schools and even roads can affect the access of the analysis of poverty or smallholders, this effect households to education services, and so forth). on household income is the main part of the story, Similarly, changes in household income and em- since it is concerned primarily with assessing how ployment opportunities will influence the demand adjustment policies have affected the incomes of by the household for education services. The the poor (and of smallholders). combined influence of these demand and supply For the other elements (health, education, nutri- effects will lead to observed outcomes (for ex- tion, women's status, and employment), a further ample, of the educational status of household stage in the logic is required, because our concern members). While our illustrations of these link- is not so much with the change in income itself, as ages have referred to education outcomes, the with how this will affect other relevant outcome same reasoning can be applied to other dimen- variables (such as nutrition, health or education sions of household welfare, such as health and status of household members, and the like). To nutrition. 86 7 Strategic policy issues Adjustment raises new possibilities, but also new Social interventions and economic distortions problems, in securing an improvement in the liv- ing standards of the poor. For poverty to be re- The key policy problem duced, the circumstances of the poor and their interactions with the wider social and economic The key problem currently confronting African system must be fully understood. Earlier chap- governments is how to assist poor and vulner- ters have shown that the determinants of welfare able groups without distorting economic mecha- in Africa are complex and varied. If policymak- nisms. If such distortions are severe, not only ers do not have sufficient information or analysis will economic recovery and growth be under- to understand the processes through which target mined, but ultimately the attainment of social groups are affected by different types of policy objectives as well. In the final analysis the re- intervention, it will be extremely difficult for them duction of poverty cannot be secured by trading to take effective action. For this reason, the appli- off the restoration of economic stability. At the cation at the country level of the principles pre- same time, prosperity will never be secured if sented here should be seen as part of a wider Africa's most important resource - its human exercise that aims to develop national capabili- capital - is not developed. Allowing such capi- ties for information collection and policy analysis tal to depreciate is not only undesirable in itself, on the social dimensions of adjustment. but also makes little economic sense. Malnutrition The policy framework offered here will evolve and sickness reduce productivity, low productiv- as more is learned from country experiences about ity limits earnings, and poor earnings limit in- the design and implementation of poverty-sen- vestment and economic diversification. Nowhere sitive adjustment. And since it is only a frame- is this more apparent than in the loss of national work, much of the setting of priorities can only income resulting from the stunted growth and be done at the country level. Nevertheless, this loss of schooling of African children over the last and the next chapter aim to explore the "policy decade (UNICEF 1985a, 1985b, and 1989). The space" of opportunities and constraints within goal is therefore adjustment with poverty reduc- which governments and donors operate. This is tion, not adjustment or poverty reduction. done through outlining the major issues sur- The objectives of adjustment have widened rounding poverty-sensitive macroeconomic and during the 1980s. In addition to the restoration of sectoral policy and identifying the room for ma- macro balance, programs have sought to improve neuver available in redesigning adjustment mea- efficiency and increase economic growth.13 To sures to enhance their sensitivity to the social these macro and meso objectives, poverty reduc- dimension. tion (and environmental protection) must be 87 added. In part, this broadening of objectives Figure 7.1 reflects the persistence of the adjustment process over a longer period than was initially antici- Gaine pated (World Bank 198&, p. 1). In the early 1980s many observers considered the macro balance objective as sufficient, implying that some loss of growth would need to be accepted, but only for a G short time. Continued turbulence in the world economy, however, including further terms of T trade deteriorations, has delayed the attainment of adjustment goals. And the political difficulties of policy reforms have made some governments waver in implementing their adjustment pro- grams. Adjustment has therefore evolved into a much more complicated process. It now entails L Eor longer-term financing facilities, an intensive sec- toral focus, and therefore a wider range of policy options. A second reason for the broadening of the ob- harming the welfare of other groups. Most policy jectives of adjustment is the now common view interventions are not of this type, however, and among governments and donors alike that ad- this applies particularly to policy changes being justment cannot ignore the needs of the poor. implemented under structural adjustment. The Although it provides many benefits to the poor, preadjustment situation may be at a point such as some of its adverse effects are by now well known. X, which is suboptimal (since one individual can In the early days of policy reforms, opportunities gain without others losing). But past experience were missed to protect the poor from their ad- suggests that adjustment is likely to take the verse effects and to encourage their longer-term economy to a position such as Z, which, although participation in the recovery process. With ad- on the efficiency frontier, cannot be said to be justment programs taking a longer-term focus, necessarily preferred to X (since some groups are there is now more scope to take up these oppor- worse off as a result of the policy change). tunities. One solution to this problem is the Kaldor- Hicks compensation test, according to which a Gainers, losers, and social welfare change is recommended if the gainers can in principle compensate the losers and still remain Implementing major policy adjustments inevita- better off compared with the situation prior to bly involves making painful choices. Most the policy change. In terms of Figure 7.1, this changes involve both gainers and losers. The means that having attained a position such as Z, it policy problem is how to decide whether a change would be theoretically possible through lump- is for the better or the worse if only some groups sum transfers to attain Y. This criterion, how- gain as a result, while others lose. Orthodox ever, requires only that the compensation should welfare economics is based on the view that a be possible in principle, not that it actually be change is desirable if it improves the welfare of paid. If the losers from a particular set of policies some without reducing that of others. This is are already poor, it can hardly be maintained that illustrated in Figure 7.1. The welfare levels (mea- a movement to Z is preferred to X, and that social sured conceptually as utility, but approximated welfare at Z is higher than X, simply because it is by, for example, expenditure) of two groups possible in principle to compensate the losers. (called gainers and losers) are measured on the This is especially true if the decline in welfare of two axes. Assume that the maximum levels of the losers is associated with a serious deteriora- welfare (or expenditure) that the economy can tion in their human condition. This would be sustain is given by the curve GL. If the economy inconsistent with any notion of human dignity is initially at X, it is clear that a policy that takes it (Nath 1988). This possibility is a very real one for to Y will be unambiguously beneficial because it African countries, where many groups exist at or improves the well-being of the gainers without near subsistence. In the context of structural ad- 88 justment programs aimed at poverty reduction, nomic efficiency and raising growth rates (World therefore, the application of the Kaldor-Hicks cri- Bank 1987b, p. 111). Returns to nontradables terion would require at the very least that com- must fall relative to tradables, but if nontradables pensation be paid if the losers from the policy are producers are fully compensated for their losses poor. the resource transfers that adjustment requires Many decisions to provide assistance under will not occur. For this reason, targeted transfers adjustment, however, rest on more broad-based must be carefully packaged with measures that criteria than those of welfare economics. If the encourage and enhance economic activities that losers are already destitute, so that the prospect are consistent with the adjustment effort. of their survival to gain the long-run benefits The emphasis on sustainable growth in struc- promised by adjustment is in doubt, assistance tural adjustment programs changes the complex- can be justified on the grounds of need. Using this ion of this policy dilemma. The problem is no criterion, there is much less of a case for assisting longer how to take the economy from Z to Y, but the losers if they come from the better-off sec- rather how to expand the utility frontier (GL), so tions of the community (unless they lose so that the welfare of both groups can be raised heavily that they become poor). At the same without lowering that of others. If a position time, if the better-off group owes its relative com- such as V can be achieved with growth, both fort to the previous policy distortions, there are groups will gain. Our review of policies (in no grounds for providing them with assistance. Chapter 8) suggests that there is scope for achiev- Nevertheless, governments often feel compelled ing not just protection of the poor under adjust- to help nonpoor losers because of their political ment, but also a significant improvement in their power (which could undermine the whole ad- situation by encouraging their participation in justment effort). Hence, assistance is provided growth. Hence, the objective is not just to allevi- on the grounds of political expediency. ate poverty, but to reduce it by contributing to Once the decision has been made to assist a the process of the expansion of livelihoods of the particular group, the issue of the type of assis- poor through their becoming more important tance to be given becomes crucial, together with participants in the economic growth that struc- the effect of this on the efficiency of the economy tural adjustment can generate. and on its growth path. It is this process of how to take the economy from the postadjustment Implications of the conceptual framework position attained at Z in Figure 7.1 to Y that is the principal concern. One of the issues that will be Our main concern in this section is to assess the raised in considering the alternatives governments room for maneuver in designing adjustment face is encountered when it is simply impossible policies to take on board social dimensions. In to take the economy to Y, because having at- doing this each adjustment policy instrument is tained Z, the only methods of compensation in- considered in turn. But before we embark on this volve some distortions in product or factor mar- review, two preliminaries are in order. First, it is kets. This means that the move may have to be important to treat the policy framework as a logi- from Z to a position such as T, which is subopti- cal extension of the conceptual framework re- mal. One of the most important challenges, viewed earlier, so we must establish the main therefore, is to assess how far African govern- implications of the conceptual framework for the ments can effect the welfare improvements that policy approach outlined here. Second, the char- are considered socially desirable without intro- acteristics of the target groups must be clarified. ducing serious distortions. The most important feature of the conceptual Some critical problems are raised when com- framework is its distinction among the three levels pensation is carried out in practice. First, the of economic activity - the macro, meso, and direct and indirect losses brought about by a pol- micro economies. Just as this was regarded as an icy change are many, and it is costly even to essential distinction in understanding how any identify them precisely (Rottenberg 1986). Im- given set of adjustment policies might affect proving the data base will help, especially at the household welfare, so it is equally relevant - if micro level. Second, full compensation can un- not more so - in considering how policy inter- dermine the changes in the incentive structure ventions can be designed to enhance favorable that adjustment must signal in improving eco- welfare effects. In making an assessment of how 89 a particular adjustment policy instrument affects effects, which can easily occur in programs that households and how policies might be targeted emphasize reforms in the structure of incentives. more effectively, the level at which an interven- The conceptual framework also underscores tion is to take place has to be clarified. the importance of a sound microeconomic data For example, consider the effects of a devalu- base for policy design. How households are af- ation on a particular poverty group. According fected by adjustment policies (and their related to the orthodox theory outlined in the earlier mesoeconomic changes) will depend on their chapters, this will favor producers of tradables production and consumption patterns. At the and consumers of nontradables. Suppose for very least, therefore, policymakers should gain some reason that the devaluation has made a some understanding of the sources of income of particular group worse off. What should be done the target households and their broad consump- about it? Does this mean that the devaluation tion patterns. It is also important to know how should be moderated, or that disadvantaged households will respond to policy changes, since households should be compensated in some way? this will determine the longer-run effects of poli- The answers to these questions depend on the cies. The conceptual framework highlights the reasons the group is made worse off. If the distinction among the different time horizons in households are net producers of tradables, they judging how adjustment affects households. It is may not have benefited from the devaluation if perfectly possible that households may lose in the product prices they receive did not rise in line the short run, but gain significantly over the longer with the devaluation. It may be, for instance, that term. This has obvious implications for policy product markets are essentially monopsonistic, design. Longer-run effects are often difficult to so that the devaluation only raised the trading predict, however, in part because they depend on margins of middlemen, and not the producer household behavior and responses. Similarly, the prices received by the target households con- phasing of policy changes can influence the net cerned. In this case, policy interventions are re- effect on households. In the above example, the quired at the meso level - improving the func- improvement in the operation of domestic prod- tioning of the product markets. Alternatively, uct markets (or in the physical infrastructure) the households may not be able to respond to the should ideally precede a major devaluation. This more favorable market opportunities because of will both raise the incomes of some (target) groups a deterioration in the physical infrastructure that of producers and enhance the output-switching isolates them from the main market centers. effects of the policy. Again, policy interventions at the meso level are Finally, the conceptual framework emphasizes called for, aiming to improve roads and commu- the advantages of defining policy-relevant socio- nications. If markets and infrastructure are func- economic groups. Because the effects of adjust- tioning well, and the households produce ment on households depend critically on their nontradables, the focus for policy intervention socioeconomic profiles, there are advantages for must be at the micro level, encouraging a change policy analysis in defining groups of households in the production patterns of the households con- that share certain key characteristics and are cerned to gain the advantages offered by the shift therefore affected similarly by adjustment poli- in market incentives. cies. The use of socioeconomic groups also en- The most important principle for policy design hances the analysis from a policy perspective: emerging from the conceptual framework, there- the socioeconomic criteria used in aggregating fore, is the critical importance of understanding households makes them readily identifiable for the process of the transmission of the effects of policy purposes. For example, it is much easier adjustment to households. On the basis of this to define a set of policies to assist maize-growing understanding, policy interventions can be ap- smallholders in a particular region than house- plied at the appropriate level - macro, meso, or holds in a particular percentile of the income micro. Most corrective interventions are likely at distribution. Clearly, policymakers will need to the meso level,64 involving changes in markets, take the concept of the socioeconomic group fur- marketing institutions, economic infrastructure, ther by defining target groups. It is to this that and social infrastructure. The conceptual frame- we now turn. work cautions against a neglect of infrastructural 90 Poverty and vulnerability * the new poor, who were above the poverty line prior to the recent shocks and adjustment African governments are faced with a wide range measures, but who have now fallen into poverty of competing social needs, and setting social pri- as a result; some of these may only experience orities is an inescapable part of the policy pro- transitory poverty, while others may experience cess. But because adjustment entails fundamen- deprivation over a longer period, becoming tal changes in policy, its effects are spread widely chronically poor through society. It therefore presents new prob- * other vulnerable groups, who remain above the lems (and new opportunities) to governments in poverty line but who have been severely affected matching scarce resources to social needs. If the by adjustment and therefore merit policy consid- objective is to address the social dimensions of eration. structural adjustment, which groups in society While some of the chronic poor have not been should have priority? vulnerable to recent policy reforms, they should In deciding on the broad domains for policy still be our concern. These people have been concern, two criteria should be uppermost: pov- largely bypassed by policies in the past and are erty and vulnerability. A household is poor if its relatively unaffected by recent policy changes. income (or total expenditure) falls short of the Thus while they have been unaffected by some standard that society sets, such as a poverty line, adverse consequences of policy, they have missed sometimes defined as the bottom 30 percent of a out on opportunities as well. The process of country's population ranked by per capita policy revision now underway provides an op- household income (Kanbur 1987b). At the same portunity to address their problems anew and to time, a household is vulnerable if it is particu- bring them into the process of growth led by larly open to adverse external events or shocks adjustment. At the same time, it is important to and cannot make the necessary adjustments to point out that assistance might also be given to protect its standard of living. While it is true that nonpoor groups because of their particular vul- ultrapoor households are certain to be vulnerable nerability to recession and policy adjustment. because of their poverty, these are two quite dis- tinct dimensions of need. Some households may Defining target groups for policy purposes be poor and not vulnerable, either because they are not affected by external events (for example, The chronic poor, the new poor, and other vul- in the case of subsistence farmers), or because nerable groups provide the domain for concern they can readily cope with the changes (for ex- about the social dimensions of adjustment. But ample, production- and consumption-switching for operational purposes, we need a more precise in the light of relative price movements). Others definition of the groups to be helped. To do this may be vulnerable but not poor, a case illustrated we must first recognize that people differ widely by retrenched public sector workers. in their socioeconomic circumstances. Policies Taking these two key criteria, the focus of policy therefore affect them in different ways. From a concern comprises three categories:65 policy perspective, socioeconomic criteria pro- * the chronic poor, whose malnutrition, illness vide a useful taxonomy for dividing the popula- and illiteracy limit their productivity and em- tion into groups of people with common charac- ployment and prevent them from accumulating teristics. As we showed earlier, such socioeco- sufficient resources to free themselves from pov- nomic groups can be used to guide data collec- erty. Their poverty is deep-rooted and existed tion, collation, and analysis and to provide the before the recent deterioration in national econo- basis for focusing on target groups for policy mies. Among these may be listed the ultra poor, intervention. The relation between socioeconomic or the destitute. They are generally at risk from groups and target groups will be country-spe- drought. Moreover, the economic shocks of the * cific, but: 1980s and the resulting policy reforms have fur- a an entire socioeconomnic group may consti- ther increased the poverty of some members of tute a target group because, for example, it com- the group. Others have been relatively unaffected, prises both the chronic and new poor, or a particu- while still others may have benefited from ad- larly vulnerable group justmnent measures" * target groups may comprise a subset of a 91 socioeconomic group. For instance, millet- and categories by gender, age, and disability are also maize-producing farmers may constitute target featured. groups for the purposes of policy reforms in agri- Nevertheless, the selection of target groups and culture, and each is drawn from the socioeco- their precise definition must take place at the nomic group of smallholder farmers; within the country level. They emerge through the detailed socioeconomic group of urban formal sector assessment of country situations, the establish- workers, only some public and private employees ment of government priorities, and from dialogue may be targeted by virtue of their vulnerability between governments and donors. Moreover, this * a target group may cut across several so- process is a dynamic one for two reasons. First, as cioeconomic groups, for example with rural fe- adjustment proceeds through its various phases, male-headed households. new target groups inevitably emerge. This is be- In summary, the entire population is divided into cause many policy reforms either take time to socioeconomic groups, but only some of the make their effects felt or are implemented in the population will belong to target groups. later stages of adjustment. For instance, many Throughout the discussion of policies in liberalization measures only exert their influence Chapter 8, various target groups become evident over the medium term. Second, the priority given as we assess how the design of adjustment policies to target groups changes as the poverty-focused can better incorporate the social dimension. Such measures implemented in adjustment's early target groups include smallholders producing years bear fruit. The problems of some target mainly food crops, subsistence farmers in remote groups are more readily resolved than those of areas, rural wage-workers, urban informal wage- the chronic poor. Having identified the process workers, micro-household enterprises in both ur- through which target groups are selected, we ban and rural areas, and low-wage workers dis- now take up the key issues of how help can be placed from both modern industry and the pub- given. lic service. Target groups that cut across those 92 8 Designing poverty-sensitive adjustment programs In considering how social dimensions should in- economy in the savings identity discussed in fluence the various adjustment policy instru- Chapter 6. Ignoring net factor income and trans- ments, we continue to follow the basic logic of fers from abroad, and rearranging equation (3) the conceptual framework. We begin with inter- (from Chapter 6) gives ventions that have an essentially macro profile - the overall fiscal and monetary balance. As we (1) (Cg + Ig- T) = (Sp - I: + (M - m proceed through the instruments, the roles of meso-level interventions (influencing product, where S, I, and C refer respectively to savings, credit, and labor markets as well as infrastruc- investment, and consumption, T is government ture) and micro-level support (such as the provi- tax revenue, X and M are exports and imports, sion of transfers or training) increases. and the subscripts p and g refer to private and government sectors. Since (1) is an identity, any Overall fiscal and monetary balance manipulation of the fiscal deficit (the term on the left-hand side) will either change the private sec- Fiscal discipline and the macro balances tor deficit or the external balance. As an identity, however, (1) also hides some of the complexity Macroeconomic policy under adjustment seeks behind the accounts. It is not clear, for example, to achieve a sustainable extemal balance and an whether a decrease in the fiscal deficit will cause acceptable rate of price inflation. To achieve these a compensating change in the private sector macro balances a government has a number of (through, for example, a rise in private invest- policy instruments at its disposal, notably mon- ment that "crowds out" the cut in the deficit) etary instruments such as credit controls or the leaving aggregate demand and the external bal- central bank lending rate, fiscal instruments ance unaffected, or whether the fiscal adjustment (government revenues and expenditures), and will result in a reduction in the current account trade instruments, such as the nominal rate of external deficit. This will depend on two main exchange. In manipulating these instruments to factors: first, the accompanying changes in the achieve internal and external balance, policy- other macro-policy instruments (monetary and makers face some basic accounting constraints. exchange rate policies) and second, on the expec- All successful adjustment programs.- including tations of the private sector. those that give priority to the social dimensions If the fiscal deficit reduction is accompanied by - must respect these basic macro accounts.6 a restrictive monetary policy, which raises inter- To explore the implications of this further, est rates, and by an exchange rate depreciation, it consider again the resource constraint facing the is likely to lead to a reduction in the trade deficit 93 (as described earlier). At the same time, an ex- must finance their deficits either through exter- pansionary monetary policy would lead to re- nal borrowing or through printing money. Each duced interest rates and an increase in invest- of these measures is subject to macroeconomic ment, with possibly little change in the external constraints - the former is constrained by exter- balance. Expectations will also play a key role in nal indebtedness and creditworthiness and the governing how the private sector reacts to any latter by its adverse effects on the inflation rate. fiscal adjustment. For example, if a tax change is In general, governments are therefore obliged to expected to be temporary, the private sector will reduce the fiscal deficit during periods of adjust- not adjust its expenditures. But a change that is ment to restore the macro balances. This obvi- considered permanent is more likely to lead to ously reduces their room for maneuver to protect private sector reactions. In the context of struc- various components of expenditure concerned tural adjustments in Sub-Saharan Africa, the re- with the social dimensions (such as expenditures quired changes in fiscal deficits are more likely to on health and education). But under a program be regarded as permanent, given the nature of of structural adjustment (as opposed to stabiliza- the underlying policy objectives and the reduc- tion alone - see Chapter 1 for a discussion of the tion in real incomes brought about by the terms distinction) there are possibilities for some room of trade decline (which is now perceived as being for maneuver, relaxing the macroeconomic con- permanent). straints and permitting smaller reductions in the A critical issue for all governments implement- fiscal deficit than would otherwise be the case. ing adjustment programs (and especially for those concerned with its social dimensions) is how Macro accounting under structural adjustment much room for maneuver they have in maintain- ing fiscal deficits within the resource constraints There are three principal ways that the resource that they face. This very much depends on how constraints imposed on governments seeking to governments are able to finance their deficits. maintain the macro balances can be relaxed There are three principal means at their disposal: though structural adjustment programs: by in- raising funds through printing money (that is, creasing economic growth, by encouraging through "seignorage," a government's right to structural changes in the composition of output, print money), borrowing from abroad, and bor- and by increasing the availability of external fi- rowing domestically. Each of these methods of nancing. Each of these is discussed in turn. financing the deficit creates problems with the macro balances, that must be taken into consid- Economic growth and the fiscal adjustment eration by policymakers. Printing more money can finance a budget deficit, but only at the ex- A more rapid rate of economic growth generally pense of higher inflation. Financing the deficit increases the budget deficit a government can by monetary expansion results in the private sustain without violating the basic macro bal- sector holding more money than it requires. Un- ances. The government can raise resources to wanted money balances are therefore spent on finance a budget deficit through monetary ex- consumption and investment goods. With only pansion without causing inflation. This is because limited domestic supply (assuming full employ- growth increases the demand for money balances ment), prices rise. Foreign borrowing is con- by the public, so that some monetary expansion strained by the capacity of the country to bear the will not be inflationary. If the money balances increased debt burden and its international credit held by the public are 10 percent of GNP, for each rating. Finally, borrowing from domestic sources percentage point GNP increases the government can lead to higher inteTest rates and reduced pri- can obtain 0.1 percentage points of GNP in rev- vate investment, thus contradicting the growth enue through printing money. There is thus a objectives of governments. significant difference in the macro-balance con- In most African countries there are limited op- straint on seignorage with and without economic portunities to borrow domestically, given the growth. shallow domestic credit markets and low levels But economic growth also releases the of private sector savings. If financing the public macro constraints on financing a fiscal deficit sector deficit through reducing the private sector through borrowing by relaxing the debt burden deficit is not feasible, most African governments constraint. The faster the growth rate of GNP, 94 the more accumulated debt can be incurred with- argues that recent African programs have, out aggravating the debt-GNP ratio. A through external financial support, permitted an government's room for maneuver to raise finance immediate import increase and positive growth. through borrowing will critically depend on The task of stabilization is therefore a difficult whether such borrowing will lead to unstable one. Aggregate demand reductionmeasures must debt dynamics - that is, an increasing accumu- unambiguously aim to achieve the stabilization lation of debt that will ultimately prevent any objective, while minimnizing consumption losses further borrowing. If the interest rate payable on by the poor and reflecting the need to maintain the accumulated debt is greater than the rate of crucial supply enhancing components of aggre- economic growth it will not be possible for gov- gate demand. ernments to borrow to finance a fiscal deficit without causing serious debt problems, because Fiscal adjustment with structural changes each year the debt-GNP ratio will rise. But if the rate of growth exceeds the interest rate, some Our earlier analysis shows clearly that even in permanent fiscal deficit can be financed through the absence of growth, the reduction in aggregate borrowing, with the debt-GNP ratio falling over demand required to restore the macro balances is time. Thus more rapid growth will permit extra greater if there is no output and expenditure room for maneuver for governments to finance a switching in the economy. These structural fiscal deficit through both seignorage and bor- changes in the composition of output (as opposed rowing while maintaining macro balances. to its overall rate of growth) can therefore give This has some critical implications for the na- governments more flexibility in adjusting the fis- ture of any fiscal adjustments the government cal deficit. may be obliged to make. If expenditure cuts Given market distortions, such as low producer have the effect of reducing the rate of economic prices and inefficient marketing, output is lower growth (as for example, if they result in a deterio- than it would otherwise be. The economy would ration in the infrastructure that supports produc- be within its production possibility set. Removing tive activities), they may cause a tightening of the these distortions would take total output to the macro constraints faced by fiscal planners. The frontier, raising the output of tradables most government needs to identify the combination of dramatically. If improvements in the level of expenditure cuts and revenue increases that re- output can be achieved quickly, less demand re- duces the budget deficit with the least adverse straint will be needed in the short term to achieve affects on the supply side. Insofar as the fiscal the target reduction in macro imbalances. Slow mix most favorable to output is found, less ag- supply responses to policy change may delay gregate demand contraction is needed (Chhibber improvements in output, however, so the scale of and Khalilzadeh-Shirazi 1988). demand reduction may have to be retained in the Similarly, it is possible that monetary constraint short run. The point still stands that some early may reduce output, again aggravating the re- liberalization will give more flexibility over the source constraints government faces in restoring aggregate demand target. Similarly, through macro balances. Whether monetary contraction devaluation, the reallocation of resources toward depresses output is a subject of some debate (see tradables (along the production frontier) is further Demery and Addison 1987a for a review). Khan encouraged. Given a target for the trade deficit, and Knight (1985, pp. 9-11) conclude that an av- less demand restraint is required when the cur- erage reduction of 10 percent in the growth of the rency is devalued. In summary, insofar as the money supply will tend to cut growth by just relative price changes initiated by structural ad- under one percentage point over one year. But justment lead to changes in the structure of pro- when the supply enhancing effects of measures duction and consumption, they will at the same such as devaluation are included in the calcula- time reduce the required reduction in the fiscal tion there is no general presumption that stabili- deficit that it consistent with the macro balances. zation will reduce output in developing countries (Khan and Knight 1985, p. 24). Those of the Raising extemnalfinance structuralist school are more pessimistic about the outcome, although less pessimistic about Af- Finally, the donor community can increase the rica than other regions. Taylor (1988, p. 166) government's room for maneuver by committing 95 external finance to the adjustment program. Key sumption is often severely depressed in the public and private expenditures can thereby be preadjustment period, and further consumption supported, reducing the threat that resources for declines will lead directly to lower nutrition and these items will have to come from a reduction in depreciation of human capital, thus damaging either consumption by the poor or social provi- the economy's capacity for growth. sions for their benefit. If sufficient external fi- At the same time, however, governments need nance is available, levels of real public and pri- to ensure that some other important components vate investment can be raised. This allows in- of aggregate demand do not contract. Private creased attention to be given to the growth and investment is one component, and this needs to efficiency objectives of adjustment alongside the be maintained and eventually increased, espe- macro-balance objective. Under a structural ad- cially in export sectors. Poverty-focused social justment strategy supported by adequate levels provisions by governments also need to be sys- of foreign finance, changes in the structure of tematically protected. Government expenditures output can take place within a policy framework on key infrastructure and services need similar promoting the growth of total output. priority if the economy's supply side (especially Because growth is largely a secondary objec- in tradables) is not to be damaged. The task is tive under stabilization programs, the prospects therefore a difficult one. On the one hand, the for poverty reduction are somewhat limited for following components of aggregate demand need the reasons discussed above. Much can be done, to be maintained or raised: consumption by the but the constraints are severe, and the exercise is poor, private expenditures on investment goods, rather one of damage containment. Room for and public expenditures on key economic and poverty concerns becomes greater when adjust- social infrastructure. On the other hand, the level ment design gives greater weight to the growth of aggregate demand must be cut to reduce in- and efficiency objectives, in addition to the flation and the external deficit. Some component macro-balance objective, and when external fi- of demand must therefore give way to protect nance is available to support the program. Un- priorities. This means that either consumption der these circumstances it is more likely that both by the better-off should be cut or the budget consumption by the poor and social expendi- deficit should be reduced. The latter can be tures for them can be maintained and even ex- achieved by cutting inefficient and low-priority panded. public expenditures. Consumption by nonpoverty groups can be cut through raising The composition of aggregate demand reductions taxes affecting them and charging for their use of public services. Both of these also reduce the Given the aggregate constraints that governments budget deficit through revenue. face in restoring the macro balances, there re- mains additional room for maneuver in deter- Simulating the poverty outcomes of macro-policy mining the composition of the reductions in ag- strategies gregate demand that are applied. We shall explore in some detail what this implies for the types of The preceding discussion highlighted the differ- government expenditures that are cut in the next ence in poverty outcomes between stabilization section, but first we consider at a more aggregative and growth-oriented structural adjustment pro- level some of the tradeoffs faced by governments grams. The first step in designing poverty-sen- seeking to reduce the level of aggregate demand sitive adjustment is therefore to seek external fi- in the economy. The level of aggregate demand nance to support strong supply-side measures. is important because one of its components is Much remains to be learned, however, of the best consumption by the poor. The poor can some- combination of macro instruments from the per- times reduce the value of their consumption spective of poverty outcomes. The principal without adverse nutritional effects by switching macro instruments are the level of the public from superior foods to inferior foods that are less budget deficit, the rate of monetary expansion, costly (but as nutritious). While much resilience and the exchange rate. Do some combinations of has been observed among the poor in their coping these policies generate better poverty outcomes strategies, this cannot be relied upon for securing than others, while still achieving the objectives of the objective of poverty reduction. Their con- macro balance, efficiency, and growth? In the 96 context of stabilization, it is certain that some more heat than light because country evidence in instruments will result in a more inequitable re- Africa is very thin and open to a wide variety of duction in aggregate demand than others. For interpretations." example, an attempt to reduce the fiscal deficit by raising income taxes will cause the better-off The public finance strategy groups to bear a major part of the cut in expendi- tures compared with increases in sales taxes. Our discussion so far has centered on the flexibil- Similarly, adjusting monetary instruments that ity that governments have in adjusting the over- raise interest rates will curtail spending in gen- all fiscal deficit and determining the burden borne eral, but particularly by debtor groups, who may by the various broad components of aggregate be among the poorer sections of the community. demand. We now consider in greater detail how But if money markets are liberalized at the same a government, in setting the level and composi- time, poorer households that previously bor- tion of real public revenues and expenditures, rowed in nonformal markets may not have to can pay greater attention to social dimensions. reduce their spending by as much as others. Establishing priorities for public expenditure Second, we do not know enough about the changes that favor poorer groups requires some optimal sequencing of macro-policy changes: from knowledge of which groups benefit from the a poverty perspective, should devaluation pre- various expenditure categories. For this, a macro- cede reductions of the fiscal deficit or follow it? meso-micro information system would play a key Third, are poverty outcomes affected by the role, because it would identify the main poverty timescale over which a given monetary and fiscal groups and provide information on their use of contraction is to be undertaken? Fourth, how public services. This should provide the basis should macro instruments be coordinated with for protecting expenditures in the short run that measures of market liberalization and investments benefit poor groups. It should also direct me- in structural change? These would be unimpor- dium-term public expenditure restructuring. To tant questions in a world characterized by perfect improve the poverty focus of the fiscal strategy, markets and where the private sector had perfect the following interdependent sets of issues need foresight. But in Africa, where the mesoeconomy to be addressed: establishing social and economic has been extensively disrupted and expectations priorities in public expenditures, phasing public of policy play an important role, we can expect expenditure reform, and assuring adequate fi- different outcomes for poverty depending on nancing of core expenditures benefiting the poor. these four aspects of macro-policy design. To These themes are discussed below. take one example, for Zimbabwe different com- binations of macro policies and liberalization (all Social and economic sector priorities set to achieve the same targets) have been found to have significantly different effects on output A key challenge of public expenditure reform is and inflation because of expectations (Chhibber to protect and expand core expenditures benefiting et al. 1989). Variations in poverty outcomes of the poor and to improve the cost-effectiveness of these policies are therefore to be expected as well. social service delivery. It is also important to Because African economies show considerable build a "safety-net" for the very poor, who, variations in market structures, policy approaches, through multiple deprivations, are less able or and macro problems, there is a need to employ unable to gain benefits from adjustment-led tools that facilitate the identification of poverty growth. Projects and programs need to be de- scenarios associated with different macro strate- signed for the poor in such areas as regional and gies. Each scenario should indicate whether ag- community development, credit and marketing gregate poverty is expected to rise or fall under schemes, technical assistance in agriculture and the particular policy mix, which poverty groups micro enterprises, and small-scale irrigation. The are liable to gain or lose, and the orders of magni- rehabilitation and maintenance of physical infra- tude involved. This task is essential because much structure (roads, water supplies, drainage, sew- discussion of the poverty effects of macro-policy erage, and the like) are known to be productive packages is clouded by insufficient country evi- and employment-intensive, to have short lead dence. The debate on the output effects of de- times, and to have prospects of good economic mand management has, for example, generated returns (Anderson 1987a and 1987b). 97 Core expenditures benefiting the poor must at the primary level. The population of preschool include investment in human capital (health, children is projected to grow at 3.3 percent annu- education, and targeted food-linked transfers) and ally until 2000, and the growth rate of 2.9 percent support to production and trade. In health the in primary enrollments expected by some analysts priority is to achieve higher cost-effectiveness in will not keep pace (World Bank 1988d, pp. 1-2). public services and their reorientation toward The quality of education has fallen with a dete- the needs of the poor, particularly in primary rioration in the supply of inputs (especially books health care. Resources released through produc- and equipment), and this has been reflected on tivity gains can be used in priority programs, the output side with the available evidence sug- with significant benefits for poverty target groups gesting a fall in cognitive achievement compared (World Bank 1987a and Mosley and Jolly 1987). with other developing regions (World Bank 1986c The reallocation of resources from expensive ur- and 1988d). Finally, females generally have less ban hospitals to rural and urban clinics, the re- access to education than males, especially at placement of expensive treatments by cheaper postprimary levels. and more effective alternatives, an emphasis on Targeted staple food subsidies are a key element basic drugs, and the greater use of paramedics of food security policy aimed at improving the are all ways of delivering better health care at a welfare of the poor. To achieve this goal, it must lower unit cost. Resource savings can be used for be possible not only to identify these groups, but special programs targeted to vulnerable groups, also to effectively target the assistance that is for example, the screening of children and moth- given. It is not our purpose here to rehash the ers from low-income households, appropriate problems and opportunities encountered in tar- follow-up programs for nutrition and health, geting (see, for example, World Bank 1986e and health education, and immunization against the Demery and Addison 1987b), but to note that the most prevalent diseases. Since much health care identification and appraisal of targeting possi- is provided within the household, usually by bilities is central to any policy analysis. The women, attention must be given to programs that problem with across-the-board transfers (such as increase women's resources, educational levels, food price subsidies) is that they are generally and the time that they have available for such inconsistent with the objectives of public expen- tasks. diture reform. Alternatives include restricting With public health provisions in short supply, subsidies to inferior goods or to product markets attention should be given to increasing the pro- in areas where the target group lives (World Bank vision of private facilities, both by profit-making 1986e, p. 40). Where neither of these approaches enterprises and community-based organizations. is feasible, direct food interventions may be This would entail removing limitations brought needed, for example through food programs about by the government on their capacity to supported by food aid, feeding programs through operate. Encouraging the use of private facilities schools and clinics, or the mobilization of cash by better-off households will allow more public transfers (World Bank, 1988e). There are few, if resources to be allocated to target groups. Simi- any, successful targeted food subsidy programs larly, the introduction of user charges for public in Africa operating through the market. Given services will generate resources for the funding that general subsidies are inadmissible from a of targeted programs, although user charges fiscal point of view, this is a priority area for should not be allowed to inhibit access to vital monitored experimental interventions. services by the poor. There exists a potentially difficult tradeoff be- There is a strong relationship between house- tween protecting social programs and funding hold welfare and educational attainment. Policy crucial economic sector expenditures. Clearly, actions that affect education have profound effects the solution cannot be of an "either-or' type. on present and future income.'9 But the educa- There is a need to find an appropriate balance tion systems of many African countries are cur- between expenditures of both kinds. In striking rently in a state of fundamental disequilibrium. this balance, economic policies should be designed Fiscal constraints, but also difficulties within the to maximize the benefits to poorer groups. The educational system itself, necessitate adjustments need for economic policy reform along these lines in educational policy. Enrollments across the is perhaps greatest in agriculture. Much public region have stagnated at all levels, but especially expenditure in African agriculture is toward state 98 production as opposed to private farming, com- The potential for raising the cost-effectiveness mercial farming as opposed to smallholder farm- of public expenditures benefiting the poor is un- ing, and capital intensive infrastructure invest- doubtedly greater over the long term. Greater ment rather than the small-scale infrastructure flows of external finance and technical assistance development needed to support peasant produc- can be expected, and the governments' own tion. Agricultural research and extension activi- planning capabilities (through, for example, a ties are often underfunded. Public expenditures better data base) will have been improved. And that can help raise the productivity of the benefits of investments involving long gesta- smallholders over the medium to long term tion lags (as is typical of many infrastructural should be protected along with critical social ex- developments) can be expected to come on stream. penditures. By raising the primary incomes of some poor groups, resources can then be made available to Phasing public expenditure reform assist groups where poverty is more persistent. There are both short-run and medium- to long- Financing core social and economic expenditures run dimensions to the design of a poverty-sensi- tive public finance strategy. In the short run, it is Raising govermnent revenues is essential to the essential to preserve (and even increase) core ex- objective of fiscal stabilization. But it is also im- penditures benefiting the poor by excluding them portant for poverty reduction. Although re- from budget cuts. While the exact sequence of sources to finance core expenditure programs can measures depends on opportunities and con- be found through savings in other components of straints at the country level, intensive restructur- the government budget, these may be insufficient. ing that involves raising the efficiency of spend- And the contribution to be made by such savings ing is a longer-run task. will inevitably decline over time as efficiency gains The first step in the short-run task of preserv- are made. Finally, while everything should be ing core expenditures is to identify the services done to mobilize external finance, this cannot be and programs that should be exempted from cuts. relied on to sustain core expenditures indefinitely. These typically include primary education, pri- Consequently, a widening of the government mary health care, sanitation, and targeted food domestic revenue base is needed to secure the consumption subsidies. Basic agricultural infra- expansion of economic and social provisions over structure and services should also be protected, the long run. given their role in reactivating production and External funds should preferably be made their orientation toward the sector that includes available to the general budget and earmarking most poor people. Defining which programs of such funds should be avoided. In this way, within these broad areas are to constitute the they will be subjected to the priorities established "core" requires information on their benefits to under agreed expenditure reform programs. This the poor. While this information is often lacking, is particularly true of counterpart funds. Their some preliminary assessment might be feasible earmarking for projects not already in the budget based on regional variations in the incidence of has a potentially inflationary impact unless other poverty. The strengthening of household-level commensurate expenditures are postponed. The data would obviously provide a sounder basis interests of macroeconomic stabilization and ad- for making such judgments. justment are best served by permitting counter- The short-run strategy for preserving core social part funds to be spent soon after they are gener- spending should, in the main, be limited to re- ated and to be used for any expenditure in a current expenditures, that is, program operation, budget reflecting a commitment to public expen- maintenance, and rehabilitation. Insufficient re- diture reform (Roemer 1988, p. 15). current expenditure outlays have reduced the Turningto domestic resource mobilization,taxes on productivity of available physical and human international trade account for about 43 percent capital (schools, teachers, health posts, and health of the region's tax revenues (Shalizi and Squire personnel). Public expenditure reviews in Africa 1987). Their importance has encouraged the use recommend raising the level of recurrent fund- of high tariffs, thereby contributing to industrial ing in key programs to levels compatible with inefficiency and discrimination against agricul- effective use of the existing capital endowment. ture. Because the import-substitution trade strat- 99 egy leads to increased capital intensity, employ- able to finance programs of poverty alleviation. ment growth is discouraged. Increasing tariffs to The collection of household data sets of the kind finance core expenditures would therefore be discussed earlier thus has an important role to counterproductive in poverty reduction. It has to play in the formulation of policy and in revenue be acknowledged that reducing tariffs (as part of forecasting exercises. This is one of the reasons a liberalization strategy) can have the short-run that it is important to obtain data not just on effect of tightening the macro constraints, both target groups, because it is nontarget groups who by reducing government revenue and by in- will provide most of the domestic resources to creasing the external deficit. This is another ex- finance programs. It is necessary to establish ample of the intimate connection between stabi- whether alternative antipoverty strategies have lization and structural adjustment that was dis- revenue implications with effects on the deci- cussed earlier. sions of nontarget groups that make financing of Commodity taxes, which provide an average some projects from domestic resources impos- of nearly 28 percent of revenues, do not distort sible. the efficiency of production as do trade taxes User charges are a potentially important form of (Anderson 1987b). They will become an increas- domestic resource mobilization. They can be ex- ingly important revenue source under current pected to benefit the poor if the revenue gener- tax reforms. The growth of such revenue, how- ated is used to extend the supply of essential ever, is constrained by the pace of improvements services or to raise their quality. If instituted in tax administration and collection. So in the across-the-board and at equal levels for all serv- short term their contribution to the costs of core ice users, however, they are likely to be regres- expenditures may be small. They can be made sive to the extent that the welfare loss of the poor more progressive in structure, with luxury pur- relative to income is larger than for the rich chases being subject to higher rates. Income taxes, (Gertler, Locay, and Sanderson 1987, p. 67). The which are levied on either individuals or busi- effect of user charges on the poor depends on the nesses, presently account for nearly 30 percent of price elasticity of their demand for services. If revenues. Personal income taxes are collected social service demand is elastic (as documented, from less than 5 percent of the African population, for example, by Gertler and van der Gaag 1988, compared with the 15 percent average for devel- for C6te d'Ivoire), the introduction of charges can oping countries as a whole (World Bank 1988f, p. be expected to result in a substantial reduction in 97). They offer some potential, therefore, for service use. Conversely, if the demand for serv- growth under current reforms (IMF 1981, p. 26). ices is inelastic (as argued for the case of health Most of the poor will still be outside the tax net, by de Ferranti 1985, p. 38), households will not even if the 15 percent target is reached. They greatly reduce their service use but may suffer predominate in smallholder agriculture and the considerable income and welfare loss as a result informal sector where income is difficult to of the introduction of user fees. User fees should measure and taxes are difficult to collect. not be allowed to restrict the access of poor groups It is unlikely that income taxes will perform a to vital services. While there is often untapped major redistributive role in Africa in the near potential for the collection of user fees or the future. Because of its skill-intensive nature, col- organization of communities for the purpose of lection costs are high - in the range of 10 per- service delivery and partial or full cost recovery, cent. Nevertheless, improvements can be made there are likely to be very poor groups in every in the structure of taxes to make them more eq- society that need to be exempted from user fees. uitable. One of the most important is to eliminate income tax allowances and deductions, which Monetary and financial policy primarily benefit wealthier groups. At present there is little hard evidence on the effects on Credit availability to the poor households, and thus on incentives and disin- centives, of alternative tax systems and cost re- We move now to consider some key mesoeco- covery measures. In calculating where to set nomic aspects of adjustment policy, beginning their tax rates, policymakers have few empirical with the operation of credit markets. Poor people studies to draw on. This affects their ability to generally have little access to formal credit mar- estimate the domestic revenues likely to be avail- kets. It is not clear whether this is because of 100 relatively low rates of return among poorer bor- depends on how restrictive the ceiling on total rowers or because of other factors, such as the credit expansion has to be, and thus on decisions high administrative costs involved in such lend- on whether a program oriented toward demand ing. Whatever the reason, they must rely instead or toward supply is to be implemented. on more expensive informal credit. Available Likewise, raising interest rates to market levels evidence suggests that in many cases poor farm- will reduce the crowding-out of smaller borrow- ers derive high returns from increased access to ers, and from an equity perspective, early finan- credit. In rural Kenya, for instance, poorer cial liberalization is desirable. Although poorer households have higher returns to land and capi- borrowers will pay the new (higher) interest rate tal than wealthier households, indicating that they on their formal loans, this will generally be much could profitably use more land (Collier and Lal lower than the costs of informal credit. There is 1986, p. 125). But lack of credit prevents them some evidence of these benefits having occurred from buying or renting the necessary assets (or under programs in Ghana and Kenya (Heller et inputs). Lack of credit also prevents poor farm- al. 1988, p. 16). ers from investing in the most profitable activi- ties such as tree crops and quality livestock. And Credit programs for poverty target groups by constraining consumption close to current in- come, it raises the risk of diversifying away from While adjustment in itself may direct more credit food crops. A similar situation exists in the rural to low-income borrowers, the gains to the poor- areas of many countries. est borrowers are unlikely to be large, mainly If the poor can use credit as profitably as better- because of their lack of collateral. The latter sig- off groups, policy attention should be focused on nals creditworthiness and provides lenders with why they are excluded from formal credit mar- risk insurance in case of default. Measures to kets. Government intervention in credit alloca- raise the productive assets of the poor could im- tions by the banking system might be one factor: prove their collateral. Legal and administrative large (often public sector) borrowers are favored. measures to secure their titles to assets such as This is further encouraged by the repression of land will have a similar result. In particular, borrowing rates below market levels, which leads allowing women farmers to hold title to land will to credit rationing. When rationing occurs, the remove some of their difficulty in gaining access largest borrowers are favored because of profit- to formal credit. This lack of access has limited able economies of scale of their loans. But the their cultivation of the most profitable crops, returns on lending to inefficient public enterprises contributing to efficiency losses for the economy are often less than those on lending to and to female poverty (Horenstein 1989, p. 23).7 smallholders. Inefficiency, as well as inequity, is In designing credit programs that help the poor, the outcome. it is essential to work through "semni-informal" Stabilization has the potential to rectify part of financial institutions. Africa's informal financial this problem. Even though a ceiling on total system is diverse, and includes rotating funds credit expansion is imposed, the addition of a ("susu," "tontines"), mobile bankers, and mon- credit ceiling on the public sector can release eylenders. Because lending takes place within funds for private borrowers. In many programs small communities and often within kinship net- assisted by the IMF, including those in Africa, the works, lenders have much more knowledge about growth of lending to the private sector has actu- the borrowers' prospects than any modern bank. ally increased in real terms (Heller et al. 1988, p. Credit is thus more readily obtained, especially 16). In determining credit ceilings there may also in the small amounts usually required, but which be scope for special provisions to favor loans to are too expensive for banks to administer. Accu- sectors with high levels of poverty. Some IMF rate assessments of creditworthiness, along with programs have explicitly supported credit ex- social pressures, normally ensure high levels of pansion to the agricultural sector in general, and loan repayment. Lending tends to be short term, the smallholder sector in particular. While selec- often seasonal, however, so that not much long- tive credit controls can cause major distortions if term finance is provided. Therefore, potentially maintained indefinitely, special (temporary) credit profitable investments are not made by the poor. provisions during the stabilization phase are Consequently, the promotion of semi-informal warranted. The extent to which this can be done financial institutions, which have the advantages 101 of the informal system but can provide longer- Mauritania (World Bank 1989b). Many of the term credit as well, has some appeal. The prac- principles discussed in relation to credit programs tice of group lending has been used with some for the poor apply to these schemes as well. In- success in Malawi and Zimbabwe (Schaefer- terest rates must not be set too low, otherwise Kehnert 1982), and the ILO has appraised its in- unrealistic proposals will be attracted and the troduction in The Gambia.7 Savings and loan revolving fund will risk decapitalization. To avoid associations have also proved effective in Westem a proliferation of credit institutions, each with Cameroon. The success of the Grameen Bank in high transaction costs, it is desirable to base such Bangladesh, especially in reaching rural women, schemes in existing institutions where this is fi- has prompted much interest in these schemes nancially prudent. For example, a credit scheme (Hossain 1988). Women in poor communities in Ghana is being operated by local banks with often form savings and loan clubs, and these are funds from the central bank. At the same time, a natural candidates for group lending schemes. single agency should be established to oversee all Their high repayment rates make such groups credit schemes. This is particularly important useful for delivering credit to the poor without because the schemes for retrenched workers will, jeopardizing the restoration of financial sound- by the nature of the problem, be only temporary. ness. In this way poorer people will find it easier Moreover, having all credit schemes under one to adapt their enterprises to policy reforms and institution, implies that - at the margin - credit will have more chance of entering the high-return should go to the best use, irrespective of whether activities promoted by adjustment. it is for the rural poor or for retrenched employ- Group lending schemes depend on the mobili- ees. zation of communities themselves. While their Assistance and training may also be needed for pace of development can be promoted by gov- workers to prepare satisfactory loan applications. emnment and extemal agencies, drawing particu- For example, the Indigenous Advisory Service larly on nongovernmental organization (NGO) (IBAS) in The Gambia gives such help prior to the experiences, their expansion can become too rapid final round of project submission. If such schemes under the influence and finance of outsiders. are to be effective, then the process of considering Group accountability and local sanctions on debt and granting a loan must be as rapid as is finan- default will then be diminished, enhancing the cially prudent. Existing schemes have often been sustainability of the schemes themselves (World too slow because of stifling bureaucracy. Further Bank 1989a, p. 117). Moreover, the development technical assistance from specialized agencies of literacy skills is an essential prerequisite for would be desirable in this effort. Some existing such schemes to be administered through local schemes have already suffered from low repay- communities. Hence, they depend on the avail- ments. One example is FIRVA in Mauritania, ability of appropriate social infrastructure. Fi- which has a 31 percent repayment rate (World nally, since groups select their own members, the Bank 1989b). The use of severance pay as collat- very poorest may be excluded. This can occur if eral may improve repayment rates. Since many the very poorest people are subject to cultural proposals come from business partnerships, prejudices. Special lending groups for these group lending would be an appropriate instru- people would then need encouragement. Alter- ment to reduce loan default. Such schemes are natively, the very poor may truly be bad risks, obvious candidates for donor assistance. IBAS in because of chronic malnutrition and disablement, The Gambia is already supported by the United for instance. Other forms of help, rather than Nations Development Program (UNDP) and the encouraging them into debt, would then be ap- International Labor Organization (ILO). The ILO, propriate. together with the U.S. Agency for International Development (USAID), is assisting the creation Credit programs for retrenched employees of a similar scheme in Senegal. The ILO has also appraised and advised on a number of other such Workers who have lost their jobs in the public or ventures. The appraisal and implementation of private sectors as a result of adjustment and who such schemes takes time, and since their phasing wish to start small businesses can be assisted with policy reforms is crucial for their effective- through credit schemes. This is now being done ness, interagency collaboration is important. in The Gambia, Ghana, Guinea, Senegal, and 102 Mobilizing the savings of the poor capital is relocated in the longer run. It is well known that when the economy is near full em- The low population densities in most rural areas ployment the real-wage effect of devaluation is have made it difficult to bring modem banking ambiguous, depending in the short run on facilities to most of the African population. The whether workers consume more tradables than cost to banks of operating in rural areas is often nontradables, and in the long run on relative not justified by the business that is generated. factor intensities in the sectors (Corden 1985). While traders in rural areas frequently provide When unemployment exists in the preadjustment informal deposit facilities, no interest is usually period, devaluation unambiguously reduces the paid (since the deposits are not used as the basis real wage. But employment is certain to rise, for loans, as in a bank). The poor are thus denied provided that the devaluation leads to an in- the interest income that provides some compen- crease in output (which implies that any defla- sation for inflation. The potential for mobilizing tionary effects are insignificant). Finally, when local communities is again apparent in this area. labor market imperfections exist, the real-wage Rwanda's Banques Populaires, for instance, with effects are more complex, because not all workers villagers as members, have reached remote areas are affected in the same way. If it is accepted that that are not covered by the commercial banks the formal labor market is predominantly in the (World Bank 1989d, p. 172). In addition, linking nontradables sector, mainly the public sector, a informal savings clubs to formal financial insti- devaluation will tend to narrow the wage differ- tutions improves the access of the poor to formal ential between formal and informal sectors, which deposit-taking facilities (and loans as well). In is generally equitable. This does not preclude the Congo, informal savings clubs have been able falls in the real wage: it may be that formal sector to make deposits in local savings and credit co- wages simply fall more than informal wages. operatives. This has facilitated greater access by Assistance may be needed where tradables form villagers to the formal financial system than if the main consumption items of the urban poor. they had acted individually. Finally, interest rate The need will be particularly acute when tradable liberalization is crucial so that nominal rates match food crops constitute the mnain wage goods. De- the rate of inflation. This will encourage deposits, valuation, together with increases in official pro- the growth of financial intermediation, and con- ducer prices, raises food prices where these are tribute to growth. It will also be equitable because tradables. The best initial help will include tar- the poor, unlike the wealthy, have no access to geted food aid or subsidies, the promotion of foreign goods or capital to protect them from periurban farming among the urban poor, and domestic inflation. assistance to those remigrating to rural areas. Less assistance will be needed to cope with ef- Exchange rate policies fects on nutrition when the urban poor are al- ready purchasing large amounts of food in paral- Devaluation affects the poor as producers, as lel markets. Prices in parallel markets are closer suppliers of labor, and as consumers. It does this to world parities, so that devaluation will leave through raising the prices of tradables relative to these prices unaffected. Similarly, if other im- nontradables and by influencing the real wage. portables consumed by the poor have been sub- Since urban and rural poverty groups are affected ject to quota restrictions and economic rents have in different ways, the policy implications for each been earned by traders having access to the li- are discussed in turn. censes, the consumer price effect of devaluation coupled with liberalization may be small. Nev- Prospects for the urban poor ertheless, food policy interventions as part of an overall strategy to tackle urban poverty may still For devaluation to motivate producers to move be needed. into tradables production, the wage rate must fall relative to the price of tradables and must rise Prospects for the rural poor relative to the price of nontradables. This en- courages labor transfer into tradables. The con- In most African countries tradables are mainly traction of production in nontradables and its agricultural commodities: export crops and most expansion in tradables is further facilitated as food crops. Some subsistence food crops are 103 nontradables in remote areas.' By raising the issue of raising the capacity of farmers to re- prices of tradables, devaluation will tend to ben- spond to the new price structure, however, not efit the rural poor as producers. This beneficial only applies to farmers previously producing effect, however, depends on the presence of an nontradables. To benefit fully from the devalua- efficient product market to communicate the tion, tradables farmers must also raise output beneficial price changes to the farmers. If this is levels. If they are constrained from doing so - not the case, farmers may not gain the full po- for example, if the economic infrastructure is tential offered by the new incentive structure. poorly developed - there is an obvious case for There is some evidence that this problem does policy intervention. Again, it should be noted exist (see Thomas 1989 and Thomas and Weide- that a policy directed at involving poor farmers mann 1988), with monopsonistic product mar- in the process of adjustment- in this case, rais- kets preventing farmgate prices from benefiting ing their capacity to respond to price incentives from devaluation. Where this problem does arise, - not only increases their incomes, but also en- devaluations should be combined with selective hances the adjustment program itself. If supply policy interventions to improve the functioning response among the poor is low, this does not of rural product markets. lead to the conclusion that devaluation should be The role of the meso economy in determining abandoned as a policy tool. Rather, it reaffirms how devaluation affects the rural poor is therefore the importance of the ancillary measures dis- critical. The chronic poor in remote regions who cussed here. In particular, it points to the impor- produce nontraded foods will see no improve- tance of reducing food insecurity, thereby reduc- ment in their prices as a result of devaluation, but ing the risk constraint that partly holds back the no decline either. As investments under adjust- participation of the poor in export crop produc- ment improve the meso economy, remote regions tion. In this sense, the drive to improve food will be drawn into national markets. Local prices security is complementary to meeting the objec- will increasingly be affected by national (and tives of devaluation. world) market conditions. In turn, nontradable As rural wage earners for either estates or foods will become more tradable in character, smallholders, the poor are likely to gain in the and the chronic poor might benefit as producers medium to longer term as the demand for their if they can market some surplus. Price measures labor (in the cultivation of tradable crops) rises. such as devaluation, therefore, will benefit the Rural labor frequently has a low supply elasticity poor as producers if phased with infrastructural in most African countries because of the relative investments in remote areas, together with absence of landless laborers. Most labor is part projects to raise the ability of poor farmers to time by farmers, and the seasonality of agricul- produce marketable surpluses of tradables. Some ture forces the wage up at the busiest times, when of these benefits may occur quickly as the import everyone has a labor shortage. The real-wage support associated with adjustment improves effect of devaluation for rural wage earners would rural transportation, but the gains from project thus tend to be beneficial over the longer term. interventions will usually be reaped over the This is especially true where the adjustment pro- medium term. Because the pace of devaluation gram induces sustained agricultural growth. may be slower under structural adjustment than In the short term, however, rural wage labor under stabilization as a result of a higher level of might suffer a real-wage fall. This is particularly external finance in support of the program, the the case where the estate wage falls relative to the chronic poor may be helped during this extended price of the goods produced by the estates. period to a position where they gain greater Whether smallholders would require any such benefits as producers. real-wage fall to induce their expansion of trad- If poor households produce nontradables, or if able production is uncertain, and complicated by they make intensive use of imported intermedi- their primary provision of wage labor to each ate inputs such as fertilizers, they will be adversely other. The short-term effects of devaluation on affected by devaluation. The key to raising their real wages are therefore somewhat ambiguous. incomes lies in creating alternative production In sum, some food intervention to the rural poor, patterns, which may require complementary especially on estates, will be warranted, although policies such as targeted extension services that the scale of this will be much smaller than that encourage farmers to change crop mixes. The needed in urban areas. Donors such as the World 104 Food Programme already have extensive experi- face of constraints on export expansion and terms ence in food assistance to estate workers. And of trade changes. This issue raises the question of food interventions to rural populations more how much potential exists to raise output and generally have improved under programs meet- incomes among export producers in Africa, many ing food insecurity caused by drought. These of whom are poor. Second, a comparison (from programs may therefore provide appropriate the poverty point of view) is made between de- avenues for food assistance under adjustment. valuation and import restrictions as instruments for correcting an external deficit. Finally, the Should the speed of devaluation be altered for reasons problems encountered in trade liberalization are of poverty? reviewed from the poverty perspective. Delaying devaluation when labor market imper- Export-demand constraints fections exist is generally a second-best solution One of the motivations underlying the choice of to the ensuing poverty problems. Devaluation an inward-looking trade strategy is export pessi- should in most cases release enough efficiency mism rooted in the conviction that the net barter gains to permit assistance to poor wage earners terms of trade are in inexorable decline (the and still leave a positive net gain from the reform. Prebisch/Singer hypothesis) and that the world Rigidities in product markets are a different market absorptive capacity for additional African matter. If supply-side rigidities are found to sig- exports is low (the so-called fallacy of composi- nificantly slow the required switch of resources tion). If true, these hypotheses cast shadows on to tradables so that the main effect of devaluation the possibility of restoring growth under adjust- is to raise the overall price level, a slower pace of ment and the scope for poverty reduction through devaluation may be warranted. With only limited increased production of tradables. Reality is both supply-side responsiveness, domestic recession more complex and, on balance, less discouraging. threatens, because tradables output shows only It is necessary to distinguish between a secular small improvement, while the production of decline in the net barter terms of trade and price nontradables falls. If the price rise induced by variability over time. There is now widespread devaluation touches off inflationary expectations consensus among analysts that the claim of a so that all prices ratchet further upward, the re- secular decline in terms of trade is not sustainable. cession would be more severe. Urban producers Conclusions critically depend on the chosen pe- of nontradables would be negatively affected, riod of observation. Unresolved measurement and a limited responsiveness of rural tradables problems (for example, accounting for quality output would provide insufficient employment change in industrial goods) make the net barter for labor shed from urban occupations. Down- terms of trade a less meaningful statistic than ward pressure on wage rates throughout the might be assumed at first. The income terms of economy would result. trade (that is, the total receipts from exports di- In avoiding inflationary expectations, a slower vided by import prices) may be a more relevant rate of devaluation will therefore attain the given measure since it is welfare-related and takes into balance of payments objective with less short-run account productivity gains made by a country cost in terms of lost output and incomes. In (World Bank 1989c). The nonfuel primary com- countries where infrastructure is depleted, these modity income terms of trade for Sub-Saharan points reaffirm the need for an externally financed Africa, while fluctuating, displayed neither a de- structural adjustment program focusing on the clining nor a growing trend during the 1965-85 supply-side of the economy. It is clear that dif- period, the lack of an improvement being as- ferent assumptions about supply-side expansion, cribed to the region's slow growth in productiv- business expectations, and policy combinations ity and export production (World Bank 1989c). have an important bearing on what theory pre- The policy implication calls for measures to raise dicts.73 productivity and output, including adjustment to correct the effects of import-substituting poli- Trade policies cies that were put in place on the basis of the Prebisch/Singer hypothesis, which, by discrimi- In this section we address three basic issues. The nating against agriculture, have been largely first concerns the overall trade strategy in the antipoor. 105 Whereas during the 1960s African agricultural restrictions have more favorable effects on poverty exports grew in volume at about 2 percent a year, than devaluations. Specifically, because tariffs they have since declined. Africa's market shares only increase the price of importables, while de- of major agricultural exports have declined sig- valuation increases the price of exportables as nificantly. World Bank calculations show that if well, the reduction in real income, particularly Sub-Saharan Africa had maintained its 1970 real wages, will be less under tariffs than under market shares of non-oil primary exports from devaluation. Tariffs can be concentrated on developing countries, and prices had remained 'luxury" goods, and the tariff revenue can be the same, export earnings of the region would used to cut taxes, minimizing the overall change have been US$9-10 billion a year higher in 1986- in the cost of living of the poor. Finally, if quotas 87 (World Bank 1989c, p. 20). Prices might, of on luxury goods are sufficient to reduce the trade course, have declined somewhat as a result of balance deficit, no major cost-of-living increase increased supplies if other suppliers had not re- need occur for the poor. duced their share in total exports. It has been This line of reasoning has led to attempts by argued (Koester, Schafer, and Valdes 1989) that many African countries to resolve their balance whether or not world market prices and exporters' of payments problems without recourse to de- foreign exchange earnings decline as a result of valuation. Since import restrictions have been an an increase in agricultural exports depends on integral part of most industrialization strategies, the rate of this increase relative to the growth in the use of such restrictions to cope with recent world demand, the price elasticity of world de- balance of payments difficulties seemed appeal- mand, and both the world market share of a ing. Nevertheless, while quotas and tariffs were particular country or group of countries and the often intended to cut only luxury imports, the supply response of competing countries to size of the terms of trade decline has forced re- changes in world prices. For example, the main- ductions in basic imports as well. Foreign ex- tenance of earlier African agricultural export change savings through import substitution have growth might well have led more advanced de- not been rapid enough to prevent the emergence veloping countries to diversify their production of this pattern. Indeed, in many cases they have and export patterns, thus diminishing the com- not been large enough to maintain the interme- petition faced by Sub-Saharan Africa. diate imports essential for the functioning of the If world demand for a particular export crop is domestic economy. price inelastic, price and total revenue (foreign In practice, therefore, unless balance of pay- exchange earnings) vary directly; that is, a price ments difficulties are temporary and small, which decline resulting from a given increase in export Africa's are certainly not, import restrictions will volume would lead to a reduction in foreign ex- inevitably induce large changes in the prices of change earnings from that crop. This would be basic goods. Moreover, attempts to ration basic serious in the case of countries that have achieved imports - at subsidized prices to poor consum- a dominant position in the world market for a ers alone - have had little success in Africa, single crop. This would appear to be the case judging by the prevalence of parallel marketing. with C6te d'Ivoire and its experience with cocoa. The claim that import restrictions have less un- It is unlikely, however, to be a serious problem in favorable cost of living effects compared with countries with a more diversified export base. devaluation is therefore weak in current circum- These considerations are of some significance for stances. The major problem with such a strategy our present concern. If only a limited potential is the damage it does to export production, since exists for an expansion in commodity exports in import restrictions motivate investment in the Sub-Saharan Africa, the scope for raising the in- protected sectors where the apparent returns are comes of poor smallholder exporters would also highest. Real wages in thes sectors are thus ef- be restricted. Countries facing less constrained fectively maintained through taxing exports export markets will have more room for maneu- (Corden 1987, p. 18). Raising import restrictions ver to enhance the incomes of export farmers. tends to perpetuate a dualistic pattern of em- ployment - limited employment in the high- Are import restrictions better for the poor than wage, protected sector and underemployment devaluation? elsewhere. In contrast, export activities encour- aged by devaluation would tend to be in labor- The argument is sometimes made that import intensive industries, since Africa's comparative 106 advantages lie mainly in labor-intensive prod- dustries, requires the development of a policy ucts. environment conducive to private investment and Small farmers are also worse off under an im- the coordination of investment with the neces- port restricting strategy compared with a devalu- sary public infrastructure and support services. ation; the latter raises agricultural prices, since None of this will be easy or trouble-free. most crops are tradable. Efficient producers of In particular, capital and labor released from import substitutes are also discouraged since they formerly protected activities cannot be expected are least likely to obtain protection, and invest- to move easily and rapidly into new uses. Protec- ment in protected importables gives higher re- tion is often concentrated on a narrow range of turns. This applies particularly to food crops, industries, many of- which are geographically many of which are unprotected importables. And concentrated. Special problems for those with market scarcities under quotas encourage the in- industry-specific skills or capital will thus arise, vestment of resources in rent-seeking to the det- together with highly localized unemployment. riment of productive activities. Although the The process through which the longer-term ben- higher tariff revenues can finance government efits of trade reform are realized may thus gen- programs that offset some of the undesirable ef- erate adverse employment and income conse- fects of protection, it is almost certain that de- quences in the short to medium terrn. How can valuation will generate higher revenues beyond trade liberalization policies be designed to mini- the medium term through its effects on output, mnize these short-run problems? Should correc- and thus on the size of the tax base. tive policies change the nature of the liberaliza- Trade liberalization tion process itself (that is, change the macro-level policy instruments), or should they be applied at Although protection has encouraged a degree of the meso or even micro levels? There are three domestic industrialization through import sub- dimensions to such design: stitution, it has reduced the pressure for firms to * the speed at which liberalization should be improve efficiency. Inefficiency has been further implemented protected by the increase in import restrictions in a the stages that liberalization should go response to balance of payments difficulties in through the late 1970s and early 1980s. A clear case thus * the relationship that trade liberalization exists for reforming existing structures of protec- should have to other economic policies (Michaely tion, given the objectives of securing macro bal- 1986, p. 44). ance, raising growth, and improving equity. Im- port liberalization removes the disincentives to SPEED OF TRADE LIBERALIZATION. Theoretical cases exports caused by import protection and initiates can be made for both increasing and decreasing a flow of resources out of protected importables the pace of trade liberalization. On the one hand, into exportables. It can be expected to raise in- there are circumstances in which it is advisable to comes since exportables in Africa are predomi- slow down the speed of liberalization. When nantly more labor-intensive than protected labor and capital cannot readily be moved out of importables. The latter are mostly import-sub- theprotectedindustry,neitherimprovedefficiency stituting industries with above-average capital- nor higher output will be gained by reducing the labor ratios, reflecting past strategies of cheapen- level of protection at a rate faster than necessary to ing capital through currency overvaluation?4 discourage new workers from acquiring skills Nevertheless, while trade reforms offer the specific to the protected industry or new capital potential for considerable efficiency and equity from being invested (Mussa 1986, p. 95). Simi- gains, there remains legitimate concern about the larly, when labor is mobile but capital is not, labor transition period. The long-term goals of trade could not take up new employment when pro- reform require fundamental shifts in the alloca- tectionisremovedfromitsoldoccupationsbecause tion of capital and labor. Entire systems of indus- of the absence of complementary factors such as trial and manpower planning, constructed over specific physical capital and managerial capacity two decades, are at present under review. Much (Edwards 1988a). Too fast a liberalization of the money and effort has been spent in building up sector would cause undue "frictional unemploy- industries whose activities are now being re- ment" in these circumstances (Michaely 1986, p. structured, and in some cases closed down. The 45). The extent to which these problems are likely other side of the coin, the nurturing of new in- toarisedependsonacountry'sindustrialstructure 107 and the sector-specificity of its factors of produc- prevented, these reforms should be sequenced tion. Fewer problems may be expected when together with import liberalization. Thus some reallocations take place within firms, but greater alteration in the speed of liberalization may still "frictional unemployment" can be anticipated be implied. The guiding principle that is estab- when movement is necessary across widely dif- lished is that the macro-policy intervention (im- ferent activities and locations. port liberalization) should be phased with Some impediments to factor mobility may be mesoeconomic policy interventions (first-best re- caused by government interventions, for instance, moval of market distortions). In most African minimum wages and regulations on hiring prac- adjustment programs, however, major trade re- tices. Some may be improved by government forms are sequenced to take place after the stabi- interventions, for example, through worker re- lization phase. Removal of distortions impend- training. The first-best policy in such cases is ing mobility could thus begin in the former phase, often to tackle the problem directly, and not to implying in most cases no delay in implementing modify the speed of trade liberalization, which second-phase liberalization. Finally, it must be would be determined by the efficiency objective emphasized that African policymakers generally alone. In other words, the guiding principle are not faced with a distinct choice between mov- would be to retain as much as possible the mac- ing immediately to free trade as against taking a roeconomic policy stance, but to apply selective more graduated approach. Rather they are faced interventions at the meso level to ease the transi- with a range of choices involving different speeds tion process. of graduated liberalization. Reforming compli- Nevertheless, the presence of some market dis- cated systems of trade protection takes time, and tortions implies that the liberalization process it may be difficult to speed up the process to the should be speeded up rather than slowed down degree required by equity considerations. Exter- if frictional unemployment is to be minimized nal assistance can therefore be valuable in reduc- (Choksi and Papageorgiou 1986, p. 7). This may ing the impediments in this area. occur when the adjustment of the economy to the existing speed of trade liberalization is too slow. STAGES OF TRADE LIBERALIZATION. Most African For example, corporate taxation may discourage countries use quotas and tariffs to protect imports, the movement of capital out of the protected sec- and liberalization will entail the reduction of both tor. Alternatively, the credit market may operate forms of protection simultaneously. Alternatively, in favor of protected sectors, so that firms will it is now common practice to convert import quo- lose these benefits by adjusting their activities. If tas into equivalent tariffs, thereby making the these distortions cannot be removed directly, and system of protection more "transparent." Tariff thus act as a drag on adjustment to trade liberal- reductions are then applied in a second stage. ization, then there is a case for speeding up lib- There are potentially two benefits for social di- eralization to accelerate adjustment by firms mensions in this process. First, the rents earned by and reduce frictional unemployment. those holding quotas are converted into tariff Similarly, high frictional unemployment can revenues for the government, some of which may arise when the private sector doubts the beusedforsocialexpendituresorpovertyprojects. government's commitment to liberalization and Second, increased transparency in the protective is reluctant to move capital for fear of losses if the structuremakesitclearerhowthatstructureaffects policy is reversed. Accordingly, rapid liberaliza- poor people, as either producers or consumers. tion, rather than a graduated program, may be This makes it easier to design a liberalization necessary to ensure policy credibility. Adverse program that maximizes gains to the poor. If the expectations and market rigidities may even structure of market distortions is such that fast warrant a policy of reducing the level of protec- liberalization would minimize frictional unem- tion to below its long-run optimal level to speed ployment, the time involved in converting the up the movement of resources out of protected protective system will diminish the importance of sectors (Mussa 1986, p. 84). Protection would these advantages. then be raised to the optimal level after capital If it is accepted that quotas have been con- had moved in sufficient amounts. verted into tariffs, how can reductions of tariffs Although tackling mobility impediments di- be applied in such a way that gains to the poor rectly may be the first-best policy, this may itself are maximized? We have already discussed how take time, implying that if unemployment is to be the speed of liberalization may be altered to re- 108 duce frictional unemployment. When impedi- The frictional unemployment costs of trade lib- ments to factor mobility are not the same across eralization can also be minimized by appropriate all sectors, which is likely to be the case, mini- macro-policy coordination. Although liberaliza- mizing frictional unemployment implies that tion encourages the movement of resources into liberalization will not take place at the same speed exportables, real devaluation is essential to across all the protected sectors. This has the strengthen this process if employment losses in disadvantage of increasing variability in rates of protected industries are to be swiftly matched by protection, which may lead to resource realloca- employment gains elsewhere (Corden 1987, p. tions between the protected sectors, but not out 20).76 If an especially strong signal to potential of them. Such a tendency would need to be exporters is required, there is an equally strong discouraged by appropriate policies. The so-called case for the initial devaluation to be a large one. "concertina" method of tariff reduction would Note, however, that devaluing prior to liberal- seem to do the least damage to employment, ization, while it encourages resource flows from judging from its success in other developing re- nontradables to exportables, also encourages such gions.75 flows into protected importables. It is essential There may also be scope for giving first priority to discourage the latter by making it clear that to the liberalization of imports that are most im- any new investments in protected importables portant to the poor as consumers. Import facili- will incur a loss when liberalization takes place ties financed by donors could support this, but (Corden 1987, p. 20). A firm public commitment modifying the liberalization program in this to liberalization is thus essential. Imposing tem- fashion may not be the best way of assisting the porary restrictions over the direction of private poor. Using import facilities to support the lib- investments and other disincentives (such as eralization of imports in general may actually credit restrictions on loans to protected grant more benefits to the poor through raising importables) may also be needed while liberal- economic efficiency and thus medium-to-long- ization of the product market takes hold. term labor demand, while their problems as It must be emphasized that potential employ- consumers could be dealt with through other ment problems will be reduced if trade reforms means. Similarly, the liberalization of intermedi- take place within a growing economy. In such ate imports before finished goods is superficially cases, the output of import-substituting indus- an attractive idea given that low-income produc- tries may take a lower share of total output but ers have been hit by the inefficiency, and thus may not fall in absolute terms. Accordingly, high prices, of domestic producers of intermedi- employment in formerly protected industries will ate goods. This could, however, actually raise grow more slowly than before, but it will not the effective protection of the producers of fin- suffer any decline. Moreover, the efficiency gains ished goods, when the desired outcome is con- resulting from the reform process will, through traction (World Bank 1987b, p. 110). higher output and income, increase the demand for nontradables. Insofar as trade reforms suc- CooRDiNAInoN OF LIBERALIZATION wrMomTHER PoU- ceed in raising foreign earnings, nontradable csi. The phasing of liberalization with other output does not have to be compressed to the policy reforms and interventions is another criti- extent otherwise necessary. Employment in pri- cal issue that policymakers must address (De- vate sector nontradables will thus be more buoy- mery and Addison 1987a). Our discussion of the ant. The employment effects of liberalization pace of liberalization emphasized that this deci- will thus be more favorable if such measures take sion is partly based on the degree to which meso- place alongside other policies that promote and micro-policy reforms can be undertaken to growth. assist factor mobility. While such reforms may go far to reduce frictional unemployment problems, Policies toward self-employment they may be insufficient to enhance the occupa- tional mobilityof the poor. The constraints linifting Sectoral employment trends the freedom of action of the latter are severe and must be tackled by a range of interventions. Social Agriculture provides an average of about 78 action programs may need implementation percent of employment in Africa, ranging from alongside liberalization measures to enhance their over 90 percent in countries such as Rwanda and mobility. Niger, to 70 to 75 percent in Cameroon and Zim- 109 babwe, and to 60 to 65 percent in the Congo and duction has grown at negative per capita rates Cote d'Ivoire.77Small-scale and micro enterprises for many years, and foreign food dependency account for about 16 percent of the region's em- has increased. African agricultural exports grew ployment, while modem wage employment cov- at an annual rate of 0.9 percent between 1982 and ers just 6 percent (World Bank 1989d, p. 41). The 1987, compared with 7.4 percent in Asia (FAO share of agricultural employment has fallen since data), and it is well known that Africa's world the 1960s, reflecting a process of structural trans- market shares for major export commodities have formation. But the shift of labor out of agricul- declined in the past 20 years. Agriculture's po- ture has been less than that in either Asia or Latin tential contribution to development and poverty America, where agriculture's share is 32 percent alleviation is far from being realized in many and 66 percent respectively (ILO 1989a, p. 52). African countries. Although small and micro enterprise employ- While the causes of Africa's agricultural decline ment and modern wage employment are pro- are complex, there is wide agreement that they jected to grow annually at 6 percent and 3.4 per- include inadequate price and institutional policies cent, respectively, as against 2.5 percent for agri- of the kind that structural adjustment programs culture, their growth is from a much smaller base seek to correct. Agriculture is the main sector than in agriculture.78 Based on current predic- producing tradables in most African economies tions, agriculture's employment share will not - in the form of export crops and food com- fall to the Asian level until early in the next century modities for import substitution. (The degree to (World Bank 1989d, p.41). which food crops are tradable depends on trans- These are medium- to long-term trends. In the port costs between points of production and the short term, the adjustments caused in the em- market.) "Urban-biased" economic policies that ployment structure by recent shocks and policy promote industrial import-substitution through responses have raised agricultural employment trade restrictions tend to discriminate against growth above the underlying trend level. Micro- agriculture by "closing" the national economy enterprise employment (and informal service and raising the price of nontradables relative to employment) have also risen above the trend. tradables (an appreciation of the real exchange Both of these employment adjustments are re- rate). Indirect agricultural taxation associated with sponses to the depression of formal sector wage overvalued exchange rates in Africa (as elsewhere) employment. The latter has been particularly af- has often been accompanied by high levels of fected by foreign exchange shortages in industry, direct taxation in the form of export levies and with inputs that are import-intensive. In addition, monopsonistic, state-controlled food procure- the withdrawal of subsidies from loss-rnaking ment. The disincentive effects associated with public enterprises, combined with a reduction in these price policies have been reinforced by the the protection of industry, contributed to the labor low priority accorded agriculture in public in- shakeout. Reductions in public employment - vestment policies, particularly in the areas of the main form of modem wage employment in transport, appropriatedly scaled irrigation, and most countries - have been extensive. Informal agricultural research and extension. employment sources have therefore tended to During the 1980s the effect of trade and mac- grow, although much of the labor in this sector in roeconomic policies on agricultural incentives and urban areas is underemployed. private agricultural investment behavior has been It is important that policy reforms maximize studied in depth for some 25 countries in Africa, employment benefits in agriculture and micro Asia, and Latin America (Krueger, Schiff, and enterprises, particularly for the chronically poor Valdes 1988). It is clear that these policies, which, who are overwhelmingly found in these activities. among others, determine the level of the real This section focuses on these issues for the self- exchange rate, play a major role in determining employed. The next section discusses in further the profitability of farming relative to detail how wage employment might be encour- nonagricultural activities. The negative impact aged further, especially in the modern sector. of "closed economy" trade and exchange rate policies on agricultural incentives has in many Agricultural self-employment countries been found to outweigh the effect of selective sector- and crop-specific promotional Agricultural performance is disappointing in policies, such as input subsidies. Africa. In Sub-Saharan Africa as a whole, pro- Adjustment, as described elsewhere in this vol- 110 ume, aims to depreciate the real exchange rate The effect of devaluation and trade liberaliza- through reducing import restrictions, nominal tion on net sellers of agricultural commodities devaluation, and reducing inflation. Adjustment would generally be expected to be positive, not is, therefore, a necessary condition for the rever- only because of the attendant price improvements, sal of the process of agricultural decline referred but also because of the reduction of the tax implicit to above, as long as an adequate share of price in the exchange rate regime when farmers are gains related to devaluation is passed on to forced to sell at an official, overvalued rate, while farmers. Given the structural rigidities that buying consumer goods and inputs at prices re- characterize African agriculture, however, real flecting dearer parallel market rates. Trade liber- exchange rate reform alone will not suffice. It alization can be expected to lead to an increase in must be accompanied by appropriate public in- the availability of consumer or "incentive" goods, vestment in agriculture and the development of and thus a higher supply response (Bevan, Collier, institutions capable of serving the needs of rural and Gunning 1989 show that rationing of con- people through input distribution, product mar- sumer goods severely constrains peasant labor keting, and social services, which together provide supply). It is true that producers are negatively an 'enabling environment" (World Bank 1989d). affected by increases in the cost of imported inputs Long-run consistency in macroeconomic man- caused by devaluation. But against this should agement and inflation control is essential to en- be set the cost of overvaluation, including long- courage private investment in agriculture and run resource misallocations and shortages of help raise the aggregate supply response. consumer goods and inputs resulting from Price incentives for agriculture appear to have chronic foreign exchange shortages. Because of improved in Sub-Saharan Africa during the sec- the low import content of their input mix, peas- ond half of the 1980s. Countries undergoing ant farmers (who make up the bulk of Africa's macroeconomic reform have devalued their cur- rural population) would seem to benefit dispro- rencies substantially, and these nominal devalu- portionately from devaluation relative to larger ations have proved effective. The real effective farmers. At the same time, risk-averse producers exchange rate for Sub-Saharan Africa has declined may perceive increases in the prices of purchased by about 20 percent in the second half of the inputs as increasing the risk of innovations ca- 1980s (Jaeger and Humphreys 1988). Agriculture pable of enhancing productivity. This would, is a slowly growing sector, and it is too early for a however, appear to be the case only where mac- full assessment of the production response to roeconomic adjustment does not translate into improved incentives. In a preliminary analysis real devaluation. of the effect of adjustment on agriculture in Africa, Cleaver (1988, para. 44ff.) found that aggregate Enhancing the welfare of the agricultural poor agricultural production in adjusting countries "performs significantly better than in countries While there are strong reasons to believe that not in adjustment," but even in adjusting coun- most of the rural poor will benefit from the policy tries agricultural production growth has not (yet) reforms under an adjustment program, this is not caught up with population growth. Moreover, necessarily and always the case. To design ap- food production responds more slowly to adjust- propriate policy interventions for groups that do ment than cash crop harvests. While price im- not gain from adjustment, it is essential first to provements, as stated above, are far from suffi- understand why they are bypassed in this way. cient to invigorate African agriculture, Balassa's We shall consider three broad cases here: first, (1988b) findings, which document a strong asso- when poor groups have insufficient access to ciation between the real exchange rate and agri- productive assets and infrastructure; second, cultural exports for Sub-Saharan Africa, are note- when marketing imperfections distort the price worthy. That in the short run, food production signals of adjustment; and finally, when poor recovery lags behind production of export crops groups fail to switch into high-return activities. may be explained (in part) by reductions in de- The most serious problem encountered in as- mand arising from real wage declines associated sisting poor rural groups arises from inadequate with devaluation. If this is so, targeted food holdings of productive assets. Securing the claims subsidies on domestically produced staples would to ownership that poor people have over their not only strengthen demand and improve nutri- assets is an important first step to ensuring that tion, but also help sustain production incentives. poorer rural groups benefit from an adjustment ill program (Chambers 1988, p. 3). Land rights are amounts of land have been redistributed from often tenuous in customary areas, and the poor large commercial farms to smallholders. frequently have few channels for redressing their Land reforms have not, in general, been asso- grievances. Extending property rights will raise ciated with adjustment lending because their time their incentives to maintain and improve those frame has usually been longer than the adjust- assets. When land tenure interventions are made, ment program. An exception is their incorpora- care must be taken to protect the traditional rights tion in SAL agreements with Kenya, with the of women to cultivate land for food. Such mea- intention of giving land rights to squatters. Simi- sures in the past have often designated men in lar measures have been undertaken in Asia in the household as title holders, thus weakening association with adjustment lending (see Demery the claims of women. This also puts child nutri- and Addison 1987b for a review). Agricultural tion at risk if the women of the household are the sector loans by multilateral and bilateral donors main providers. Cote d'Ivoire, Ethiopia, Kenya, can provide one vehicle for this purpose. and Zimbabwe have now given women the right Improving the access of poor groups to infra- to inherit and own property (World Bank 1986d, structure can be as beneficial in many cases as p. 40). increasing their productive assets. Infrastructure Raising the access of the poor to quality land is investments often bypass areas containing high an important second step. Although low popula- concentrations of poor people. This raises their tion densities suggest that Africa is land-abun- production costs and acts as a barrier to gains dant (compared with South Asia, for example), from greater specialization. For example, while much unused land cannot yield even a subsistence adjustment programs in Ghana and Kenya have living without major investments. Competition sent out clear signals for the expansion of tradable for quality land is increasing in many countries. activities, many of the poorest farmers have faced The FAO estimates that only 30 percent of the difficulties in achieving this because of their loca- labor force lives in countries that have unused tion (Heller et al. 1988, p. 20). In making new land with potential yields equal to those of land infrastructural investments, policymakers need already cultivated (Higgins et al. 1982). In these to give more weight to actions that assist the cases it may be feasible to make the land avail- poor. In many areas transport infrastructure has able to rural target groups without the necessity broken down, so that some local markets are of large investments. The important target groups poorly integrated with the national market, lead- will be found not only among the rural poor, but ing to large regional price differences (Ahmed also among the urban poor who can be resettled. and Rustagi 1987, p. 109). In designing a program Unused land held under communal and tradi- of infrastructure rehabilitation, greater priority tional titles may provide the first source. Inter- for areas ill-served by transport infrastructure, vening in communal tenure systems to create but with a high concentration of poor people, rights for target groups can be difficult to achieve, may be called for. however, and politically hazardous for govern- A second set of issues is raised when the failure ments (Feder and Noronha 1987). The target of poor groups to benefit from adjustment arises group may be outsiders to the local community from the weakening of price signals. Many of these where the unused land is located, or they may have already been discussed under our consider- suffer from cultural and gender discrimination ation of the effects of devaluation. In cases when within the community. Communities are in- this is the result of weak marketing structures, creasingly placing restrictions on the settlement the terms of trade facing poor farmers may be of outsiders. For example, in some areas of Ghana raised through improving the efficiency of official and Cameroon outsiders are not allowed to plant marketing organizations. In Asia, for example, cash crops (Feder and Noronha 1987, p. 154). 75 to 90 percent of the consumer price of food Moreover, the rights of the target group have to grains is paid to the farmer, while in Africa the be secured in some way - through the grant of proportion is only 35 to 60 percent (Ahmed and legal title, for example - and compensation per- Rustagi 1987, p. 115). Nearly 30 percent of the haps paid to others. Alternatively it may be pos- difference in margins is due to the lower effi- sible to transfer unused land held under legal ciency of African marketing organizations. In title to the target group (for example, that held by many countries the growth of marketing costs modern estates). In Zimbabwe substantial has been as important as currency overvaluation 112 in causing low producer prices (see Harvey 1988, poorest 50 percent of the village studied by Col- p. 221, on Tanzania and Zambia, for example). lier et al. (1986, p. 75) were dependent on subsis- Marketing reforms are now benefiting poor tence crops for 70 percent of their income - this farmers. In Mali, for instance, a restructuring of is double the share for the better-off half. In C6te the marketing system has shifted output and in- d'Ivoire approximately 44 percent of the rural put prices in favor of farmers (Tuinenburg 1987, poor cultivate cocoa or coffee, compared with 65 p. 503). Improvements in the efficiency of mar- percent of the rural population as a whole keting organizations can reduce the conflict of (Glewwe and de Tray 1987, p. 20). In Kenya the interest that exists between the poor who produce probability of a household being poor falls if it food and poor consumers over food prices. One grows tea and coffee - the main export crops study found that, for a sample of African coun- (Greer and Thorbecke 1986). Although cotton is tries, reducing the marketing marginby 25 percent mainly grown by poor households in C6te would result in a 49 percent increase in farm d'Ivoire, their participation in other, more profit- prices and a 13 percent fall in food prices, given able crops, is lower (Glewwe and de Tray 1987, p. reasonable assumptions about demand and sup- 14). Overall, female-headed households are less ply elasticities (Ahmed and Rustagi 1987). likely to cultivate cash crops because the alloca- Even with marketing and infrastructure im- tion of the necessary land, credit, and inputs does provements, some poverty groups may still have not favor women. Altering the product mix of imperfect market access (for instance, from private female target groups will necessitate interven- and public monopsonies in marketing in remote tions in the supply of factors of production to regions). Their terms of trade will thereby benefit them. less from adjustment than the average. Further In summary, raising cash-cropping by farmers deregulation may be warranted, if, for example, offers an important way in which their benefits regulations block greater competition in market- from adjustment and growth can be increased. ing structures. But this may not be enough for In some cases the investments in infrastructure areas with special difficulties. In this case the and the improvements in marketing that have needs of the poverty target group may warrant been discussed may be adequate for target groups the establishment of a special component of the to raise their incomes sufficiently. In such in- official marketing system (perhaps with some stances, policy interventions must clearly be di- explicit subsidy). Alternatively, the provision of rected at the micro level. In other cases a more resources and assistance to members of the target comprehensive package of measures involving group may allow them to reduce their marketing marketing and infrastructural services may be problems. Women farmers may receive lower needed. Higher cash-cropping may be dependent prices than their male counterparts because of increasing access to productive assets in the ways poor market access, and can thus merit special discussed previously. help (Henn 1983, p. 1050). For example, the ILO A major constraint on expanding the incomes is assisting women's cooperatives in the Gambia of target groups in this way is the ecology of the to market food crops, particularly from remote country concerned. This plays a large part in rural areas. determining whether farmers in a given region Finally, it is possible that some poor groups fail are able to cultivate the most profitable cash crops. to benefit from adjustment simply because they For instance, in many West African countries there do not switch production toward high-return ac- is a marked division between ecological regions tivities. This may arise initially from a perception in food production and export crop production. lag - farmers may take time to observe relative In Ghana, for example, cocoa is cultivated in the price changes and to expect them to continue. Or coastal region, while farmers in the northern sa- the failure to change output patterns may be more vanna derive most of their income from domestic deep-seated, arising from either ecological con- food production. In Cote d'Ivoire maize is the straints or risk aversion. The participation of dominant crop in the savanna region, and while poor households in producing the most profit- cotton is an important cash crop, cocoa and cof- able cash crops is usually below average. Such fee are concentrated in other regions. Conse- participation is a strong determinant of income quently, it may not be possible to help a poverty differences across households in most countries. target group in a particular region through the In Tanzania, for example, the income of the expansion of particular cash crops because of 113 ecological constraints leading to prohibitive proj- Under adjustment the prospects for earnings ect costs. In such cases, assisting the entry of the from urban micro enterprise may be dim. First, poor into activities with a higher return could workers displaced by the liberalization of indus- necessitate relocation, and issues of mobility that try and the contraction of public employment are discussed later in this section. seek out urban informal employment. Using their Some countries have already shown how par- severance payments for capital investments, the ticipation in the growth process can be raised in a returns in the more advanced forms of relatively short period of time, provided that micro enterprise can be expected to fall. Second, comprehensive packages of assistance are em- micro enterprises producing nontradables will ployed. To take only one example, in Zimbabwe be unfavorably affected (mainly in services). smallholders supplied only 5 percent of marketed Third, if adjustment depresses urban incomes, maize in 1980. But the subsequent targeting of then the demand for informal products will fall. marketing and support services, together with Fourth, as modern industries recover their capac- favorable price policies, allowed them to raise ity utilization, some consumers will switch back their share to 30 percent by 1983 (Mellor, Delgado, to them since their products are preferred. Fi- and Blackie 1987, p. 353). Given the right kind of nally, some micro enterprises traded in scarce sectoral interventions and an appropriate policy commodities before adjustment, and policy re- environment, similar participatory growth will forms will reduce incomes from rent-seeking. become feasible in other countries. While urban informal employment and earn- ings may contract under adjustment, the pros- Micro-enterprise self-emnployment pects for rural micro enterprises are brighter. As policy distortions against agriculture are cor- Recent years have seen a reversal in official atti- rected, the demand for goods and services by tudes to the informal sector, and a number of farmers will rise as their incomes increase. Much strategies have been advanced to improve the of this will be met by rural-based micro enter- performance and profitability of the prises, since farms generally require local suppli- micro enterprises within the sector (ILO 1989b, p. ers to maintain farm equipment, for example. 65). These strategies focus on training, credit, Similarly, much construction work requires and the provision of supporting infrastructure. knowledge of local materials and requirements. In particular, licensing and registration formali- In Africa each dollar increase in agricultural in- ties may be unduly restrictive for small business come generates about $0.50 in rural nonfarm development. In many countries these formalities earnings (Haggblade and Hazell 1988, p.10). Ag- are extensively evaded and are thus of little merit. ricultural growth in Asia has been sustained for But operating illegally makes it more difficult for long enough periods to show the kind of inten- small entrepreneurs to obtain formal credit and sive nonfarm rural economy that can develop to use public facilities. These points also apply to (Mellor 1985). Asian farmers typically spend a formalities applied, notably to women (such as large proportion of their additional income on requiring a male relative's signature on business locally produced goods and services, including licenses). Adjustment offers an opportunity to textiles, transportation, housing, and health serv- review these restrictions, especially since a thriv- ices (Hazell and Roell 1983). Similar expansion ing business sector is crucial to tradables expan- can be expected in Africa as households who sion. operate micro enterprises as sidelines to farming Urban informal employment growth was rapid increasingly specialize as demand grows. In some during the 1980s, although many in the sector are countries the expansion of such off-farm em- underemployed since ease of entry into the low- ployment can be expected to reduce rural pov- est grade occupations has provided a last resort erty because poverty groups engage dispropor- for those displaced from the modern sector. The tionately in this activity.79 But in other countries effect of inappropriate policies on rural livelihoods (for instance, Lesotho, Tanzania, and Uganda) has also swelled the numbers of informal workers. nonfarm income is disproportionately concen- As a result, the profits of informal micro enter- trated among better-off households. Off-farm prises have declined in free-entry occupations, employment is also a major income source for although in activities where substantial capital is women, who dominate "traditional" female ac- required, operating margins have probably been tivities such as food preparation (Haggblade, better. Hazell, and Brown 1988).w° 114 The opportunities for informal manufacturers nontradables. While the latter category includes in towns to supply the increase in rural demand labor-intensive activities such as service employ- will be limited in many cases to areas in the ment, it also includes industries that are capital- immediate vicinity of towns. The improvement intensive because of previous distorting policies. of transport networks under adjustment may give Since agriculture is the dominant tradable in most greater scope than before, but the nature of many countries, and since it is very labor-intensive in of the goods and services needed by farmers will Africa, a rise in the average labor intensity of probably preclude informal urban suppliers. production can therefore be expected to result High transport costs to remoter regions also make from adjustment. it difficult for urban suppliers to compete against These orthodox expectations, however, are de- local producers. Thus, if urban informal pro- rived under the assumption of full employment. ducers are to gain the higher returns offered by In Africa unemployment is high - averaging the rural market, they will need to migrate, and over 15 percent across the region (ILO 1989b, p. assistance for their mobility can be considered. 17)- so real wage improvements may not result, For many who have lost their access to tradi- or may be slow in materializing. Unemployment tional lands, such nonfarm activity may offer their in the African labor market arises from a mix of best means of returning to the rural sector. Sec- factors, including the preadjustment situation of ond, because demand for informal products will industries with foreign exchange constraints, shift toward rural micro enterprises, assistance to short-term labor market responses to adjustment the informal sector in general requires review to in some countries, structural employment prob- ensure that strategies are adapted to meet the lems that have long afflicted the region, and high needs of rural micro enterprises. population growth rates. Real wages will im- prove at a faster rate if adjustment promotes out- Policies toward wage employment put growth, not just a change in its structure. An emphasis on growth and efficiency in program Wage and employment outcomes under adjustment design is thus crucial to prospects for wage earn- ers. Current policy reforms in Africa are aimed at Design of the macro strategy also affects the significantly altering the structure of production short-run outcome for real wages - in particular and consumption, and thereby the use of produc- the relative importance attached to demand re- tive factors. The impact of these processes on straint (emphasized in stabilization programs) wage earners partly depends, therefore, on the versus supply expansion (emphasized in struc- size and character of the resource reallocations tural adjustment). Although structural adjust- required. In the African context the shift is most ment may cause some frictional unemployment often from nontradables, such as urban services through market liberalization, its emphasis on and quota-protected industries, to tradables, such growth is more conducive to employment than as export and food crops together with competi- stabilization. Under the latter, domestic demand tive manufactures. Over the longer term, the falls when monetary and fiscal restraints are ap- outcome for wage employment and earnings de- plied, nontradables production contracts, and pends on the relative labor intensities of the ex- tradables production expands. But in the short panding versus the contracting sectors, and term, nontradables usually contract faster than whether the structural shift in production takes tradables expand. Labor demand therefore place in the context of overall output growth. If weakens, putting downward pressure on wages structural change causes a fall in the overall labor (even if long-term wage prospects are good). intensity of production, and GDP is growing only Whether wages are flexible or not can also affect slowly, downward pressure on wages is inevitable the short-term employment effects of adjustment. (if labor absorption is to be maintained). But With real wage rigidity, some workers may pro- with strong GDP growth, even when contracting tect their living standards at the cost of unem- sectors are more labor-intensive than expanding ployment for others, whereas a fall in wages sectors, the negative (static) effect on labor de- would allow the employment of the latter de- mand may be outweighed by the dynamic ben- spite weakened economic activity. In general the efits. current weakness of organized labor in Africa In Africa, exportables and unprotected import- prevents much of the real-wage rigidity experi- ables are generally more labor-intensive than enced in semi-industrialized and industrialized 115 countries. Real wages have fallen considerably Discrimination by employers on the basis of in Africa - both before and during adjustment gender or culture is the most important type of in many countries (Jamal and Weeks 1988). For "endogenous" labor market inefficiency. Gov- unskilled labor, real wages may be close to a rate ernments can prohibit such discrimination that provides only a minimum nutritional level. through legislation, and this can have substantial In such situations further wage declines could benefits if properly policed. Regulation is most not be expected, and would not therefore per- effective in the public sector, but less so among form their function of equilibrating the labor private enterprises, especially where labor re- market (as discussed earlier). cruitment is informal. Governments can begin In cases where there is scope for wage flexibil- by reviewing their own hiring practices for dis- ity, a government may call for temporary wage crimination. Action against discrimination be- restraint both to encourage a rise in tradable pro- comes even more important during adjustment, duction and to maintain employment. Whether because the disadvantaged are often the first to such restraint can be successfully negotiated with be dismissed and the last to be hired. While labor depends on whether labor perceives the inappropriate regulations can be eliminated, of- situation to be fair. Governments that attempt to ten at the stroke of a pen, cultural prejudices are spread adjustment costs equitably (through, for much more persistent (Serageldin 1989, p. 30). example, economizing on inessential public ex- The latter may render well-meaning legislation penditures in order to protect social services) will ineffective, or even counterproductive. Ultimately have more credibility with labor than those that interventions in the labor market may be less act otherwise. Governments that resort to limit- successful than empowering the disadvantaged ing the rights of workers (such as freedom of through other means, such as improved education association and collective bargaining) are likely and asset distributions. Aside from enhancing to lose public confidence in the integrity of the their self-employment, such assistance may, adjustment program itself. There are a number through improving their human capital, weaken of examples of successful incomes policies built prejudices against them in the labor market. through dialogue among government, labor, and In general, market inefficiencies caused by firms (Kyloh 1989). In such situations workers governments are easier to rectify than those in- effectively trade present wage benefits for future herent in the market itself. Obviously govern- employment and wage growth. And specialized ment interventions are prime candidates for re- agencies such as the ILO have much experience view at times of adjustment. Regulations on in assisting the promotion of tripartite dialogue employment practices, working conditions, and (ILO 1987, p. 55). minimum wages should be assessed to establish whether they are having their desired benefits. Improving the functioning of labor markets Such regulations may improve worker "quality," and thereby productivity and employment. They The labor market is a key component of the may effectively correct employment practices that meso economy. As such, its performance is a the market is too weak to rectify automatically. critical determinant of how the costs and benefits Or they may unduly encourage capital intensity, of adjustment are distributed across society. Yet and harm employment (Krueger 1988, p. 365). this performance is often far from satisfactory. Alternatively, they may have unintended effects, Both inappropriate government interventions and such as limiting the job prospects of young people. the workings of the market itself can cause ineffi- Because the links between such regulations and ciencies. Unemployment may exist because of the performance of both workers and firms is government regulations on hiring practices, for complex, they should be reviewed on a case-by- example, or because the qualities of some types case basis (ILO 1989b, p. 44). While wage levels of workers are not effectively signaled to em- are rightly seen as the key measure of labor wel- ployers by the market. Consequently, policy fare, working conditions, such as health and choices at the level of the macro strategy are safety, are also an important component in their necessary, but not sufficient, for maximizing the own right. growth of wage employment. Complementary There are strongly differing opinions about actions at the level of the meso economy, particu- minimum wages. One view holds that formal larly in the labor market, are also needed. sector employment will be reduced if employers 116 are required to set wages above market-clearing largely unaffected. This occurs when it is not levels. The displaced labor either becomes un- profitable for employers to reduce the wage they employed or, more likely, is absorbed into the pay if the legal minimum is removed. Removing informal sector (where minimum wages cannot the minimum wage in these circumstances will be enforced). Informal wage levels are thus re- not affect the number of jobs lost from the formal duced. Wage dispersion, along with inequality, nontradables sector, or the extent to which the increases, and a dualistic pattern of employment informal wage falls.8' But if during adjustment is perpetuated (which in turn encourages rural- the minimum wage is removed from a to-urban migration). The formal sector wage dif- monopsonist in nontradables, employment will ferential between skilled and unskilled workers contract further than if it is retained. The is reduced, and with it the incentive to accumu- monopsonist argument has more validity, but a late human capital (Psacharopoulos 1986, p. 54). limited application. Estates in the agricultural The alternative view defends minimum wages sector come nearest to being monopsonists in the on two grounds. First, minimum wages can in- African context, especially where smallholder la- crease employment and output by raising labor bor markets are thin. A minimum wage for the productivity. This may be important where estate sector may be justified in some cases. worker efficiency is constrained by malnutrition and poor health. Rather than diminishing prof- Public sector employment policy itability, minimum wages raise it. The require- ment for wage legislation, presumably, is based The main burden of wage restraint tends to fall on imperfect perception on the part of employ- on public employees, reflecting a need both to ers. Second, it is well known that when there is curtail public expenditures and to transfer re- only one employer (a monopsonist) in the market, sources out of nontradables. Although public both employment and the wage level are set be- sector wage freezes have been a feature of donor- low their competitive market levels (Brown, assisted programs, they have also occurred as the Gilroy, and Kohen 1982). Introducing a minimum domestic revenue base has contracted during wage (equal to the competitive market level) will earlier recession. For these reasons, the scope for raise employment without affecting output. further reductions in public expenditures through The real value of minimum wages has fallen public wage freezes may be limited. Indeed, in drastically across Africa, because the minimum many cases remunerations have fallen so low has not been adjusted in line with inflation. In that they neither keep low-paid workers out of some countries therefore, market wages may lie poverty nor attract and retain skilled manpower. above the mninimum, thus making the latter irrel- The problems of absenteeism, demoralization, and evant at present. But in countries where mini- corruption affecting many public administrations mum wages do "bind," should they be retained, are familiar and need no reiteration here. These raised further, or removed? The case for remov- limit the capacities of African governments to ing them rests on the view that minimum wages carry out key policy and management tasks nec- exacerbate the adverse wage and employment essary for successful adjustment programs effects of adjustment. As nontradables contract (Nunberg 1988). Hence there is not only a "social (either because of deflation or devaluation or both) dimension," but an important institutional effi- there is no possibility of retaining some employ- ciency problem as well. ees by downward money-wage adjustments. Given the need to redirect public expenditures They are displaced into the informal sector, which to key investments and services that support the forces informal wages down. Those who remain adjustment effort, it is almost impossible to en- employed in formal nontradables thus protect visage a large improvement in the remunerations their wages at the cost of depressing the living of all public employees. The public wage bill is standards of informal workers, many of whom very high in some countries, and will need to be are poor. Removing the minimum wage during reduced further if priority investments are to take adjustment would raise formal employment, and place. In doing this, care is needed not to erode thus reduce poverty. public sector wage differentials in order to pre- When the minimum wage has raised produc- serve incentives and retain highly skilled em- tivity among competitive firms, its removal may, ployees (Klitgaard 1989). Comprehensive re- under certain circumstances, leave employment structuring of public employment has been un- 117 derway throughout Africa since the mid-1980s les the root of the poverty problem. In such and earlier. The introduction of performance cases, if nothing else is done, PWPs can end up pay, linked to improvements in govemment making large demands on public resources in- revenue collection and the efficiency of govern- definitely, with no significant improvements. For ment programs, can be an appropriate solution instance, in situations where policy reforms are to the problem of keeping salary levels within expected to reduce the growth of urban employ- bounds while at the same time maintaining in- ment over the long term, the best solution is to centives. Revenue-raising agencies are prime help the urban poor into rural employment candidates because their performance is easiest through rural projects and mobility assistance. to measure. For example, in Bolivia the pay of For such groups, PWPs should be only one com- revenue officers is now related to the amount of ponent of the strategy. revenue collected, and a large increase in rev- There are a number of other important issues enues has resulted. In addition there are cases relating to the design of PWPs. First, in practice where user charges have been shared with em- better-off groups often benefit the most from the ployees (for example, Nigeria). On the expendi- new infrastructure (since it raises the retums on ture side, railway, highway, and port authorities productive assets, most of which are owned by in Ghana have, with World Food Programme the better-off), and the gains to the poorest are (WFP) assistance, linked subsidized food to mea- confined to the income earned in their construc- surable results for each worker and for the corpo- tion phase. While appropriate project selection ration as a whole (Klitgaard 1989, p. 452). may overcome this, it may often be unavoidable until complementary programs have improved Public works programs the asset base of the poor. This also implies, however, that the better-off would bear the brunt At their best, public works programs (PWPs) of user charges to recoup the PWPs' costs. Such provide significant assistance (in either cash or revenues can be used to finance further PWPs food) to the underemployed, while creating in- and to cover infrastructure maintenance, thus frastructure of lasting benefit. They are now being continuing employment benefits to the poor. adapted to help poor groups cope with the impact A second issue of recurrent concem is that poor of adverse policy reforms (see, for example, Re- project design, makeshift administrative ar- public of Ghana 1987, p. 5). Given their micro rangements, and lax project supervision have too orientation, they can be tailored to the problems often contributed to low cost-effectiveness among of specific poverty groups whose employment PWPs (Guha 1986 and World Bank 1986e). On problems are insufficiently resolved by actions at average, nonlabor costs account for higher shares the macro- and meso-policy levels. Well-designed of total costs in African PWPs than in Asian PWPs PWPs can impart major improvements in human (Stewart 1987b, p. 202). In Latin America, PWPs capital, both through improved nutrition and on- have addressed adjustment-related unemploy- the-job training. The general improvement in ment, but many of these projects have involved labor quality has important spin-offs for the prof- activities with low returns for either the adjust- itability of the household's own activities (which ment objective or social goals. In such cases the are further enhanced if program participants are government is effectively financing a reexpan- the main beneficiaries of the infrastructure cre- sion of unproductive nontradables. An "uncon- ated). The time profile of household activities ditional" transfer system would be more cost- can also be favorably affected: for instance, up- effective if projects with reasonable returns cannot grading a road may reduce travel time, thereby be found. releasing more time for productive employment A third important design issue concerns the and childcare. Insofar as PWPs contribute to level at which the PWP wage (cash or kind) should raising production (especially of tradables) they be set. The higher the wage, the greater the assist in meeting adjustment objectives. benefit to the participants, but setting the wage Unfortunately, PWPs do not provide a lasting too close to the market wage may discourage solution to groups facing poor long-term em- participants from seeking private sector employ- ployment prospects. PWPs, while they provide a ment. Payments below the market wage effec- useful income supplement, are no substitute for a tively target the program to those who need it comprehensive program of assistance that tack- most (World Bank 1986e, p. 38). This means that 118 if the wage has fallen under adjustment, the cor- low quality, of limited coverage, and have low responding PWP wage may be low, and the PWP female participation rates. Their improvement is income supplement to participants will not re- one of the prerequisites for bettering the limited duce their poverty sufficiently. If it is thought employment prospects of young people. Pro- that the nature of the work will deter those out- grams should move away from the traditional side the target group from participating, then a focus of preparing the young for wage employ- market-level wage may be appropriate (Thomas ment, and toward promoting their skills for self- 1986, Kinsey 1987). But otherwise the PWP will employment, for example, through in-service be limited as an antipoverty measure. Finally, training (ILO 1987). Curricula need to be orien- PWPs need to be designed in such a way that the tated toward the skills required by the priority very poorest can participate. Because of their sectors under adjustment. For example, employ- malnutrition, they may be unable to undertake ment in labor-intensive clothing businesses has physically demanding work unless the PWP is risen in several countries (Madagascar and coordinated with suitable nutrition interventions Mauritius, for instance) in response to policy re- and skill enhancement. Similarly, PWPs need to forms. To ensure their rapid growth and to en- be targeted to the needs of women, whose house- courage the production of high value-added hold responsibilities often reduce their participa- items, improvement in labor skills is essential. tion compared with men. Complementary actions A number of countries now have retraining such as the provision of childcare facilities are schemes in operation, or in the pipeline (includ- required. ing Ghana, Guinea-Bissau, Senegal, and Mada- gascar). These are mainly for retrenched public Training and retraining employees, although they should also include those affected by the liberalization of private in- As enterprises adjust their product lines to new dustries. Retraining is often linked to providing policy incentives, they will undertake some re- other support, such as credit. Early implementa- training of employees. But government sponsor- tion is important if such schemes are to mitigate ship of vocational training is also needed to meet the effects of adjustment. Closer coordination is the demands of employers, especially in export needed between the designers of policy reforms sectors. Such schemes should be protected and training managers if unexpected demand is alongside basic education in the public finance not to overwhelm the training institutes. strategy. Training systems at present are often of 119 Concluding perspective There are at least two major problems in deciding first establishingmacro-meso interactionsand then on a conceptual framework for assessing the so- dealing with meso-micro effects. This approach cial dimensions of adjustment. First, the policy also applies at the empirical level. research problem is itself a major undertaking. The most important conclusion that the theory The analytical challenges that are faced in estab- offers (apart, thatis, from indicatingthatthe subject lishing how macro- and sector-level policies af- is inherently difficult, and that the theory is in- fect households and the well-being of individuals evitably inconclusive) is that adjustment policies are serious and should not be underestimated. can set in motion changes in the economy which One of the major outcomes of the theory is that have profound and pervasive effects on markets, there are no ready answers - the problem is infrastructure and households. Attempts to raise highly complex and can only be resolved in the household incomes which ignore (or even run last analysis at the empirical level. And, of course, counter) to these deep-seated changes, are both the empirical problems that are raised are just as foolish and counterproductive. While the theory challenging (Chapter 6). is of little use in analyzing the myriad changes that The second problem arises from the heteroge- adjustment programs inevitably involve, itisat its neity of the Sub-Saharan African region to which mostusefulinuncoveringthesemorefundamental this framework must apply. It must at the same economywide effects. Without it, the policymaker time be general enough to apply to all the diverse would not be able to see the wood for the trees. economic and social circumstances of the region, The document has sought to provide an over- and specific enough to be of use in guiding the view of the range of policy measures which re- policy and investigative initiatives at the country quire attention in the course of formulating pov- level. Because of this, we have persistently issued erty-oriented adjustment programs. As the focus warnings that the analytical framework we have of adjustment policies has shifted during recent presented is not meant to be strictly applied in years to encompass wider objectives than simply each and every case. For some cases, the as- attaining macroeconomic equilibrium, so too has sumptions we have made are a reasonable ap- the realization that the impact of macro policies proximation, but for others, there may be a need on the well-being of households and individuals for further refinement. Our purpose, however, is is complex, often pervasive, and, above all, not not to sell any specific model with its associated well understood. Economic theory must now range of assumptions, but rather to establish a give way to empirical application as a result of structured way of thinkingabout the problem. At the wide variety of socioeconomic conditions in the heart of this structure is the simple device of the countries of Sub-Saharan Africa. The rela- dividing the research problem into two stages: tionship between structural adjustment and pov- 120 erty presents even more hurdles for policy de- transitionally affected by the adjustment process. sign, due in no small measure to the great varia- A number of measures were discussed in the tions in the kinds of poor households found in areas of public finance, monetary and financial most African countries, and to the diversity of policy, exchange rate and trade policy, and em- their income and expenditure patterns as both ployment policy which can shift the thrust of consumers and producers. The effects of various adjustment toward supply expansion and away macro-policy instruments on such households are from demand restraint, in order to achieve the thus difficult to determine, much less predict. dual objectives of sustainable growth and pov- Yet some understanding of this interaction is erty reduction. Through shifts in the mix and necessary for the design of effective policy inter- sequencing of policy instruments available un- ventions. der adjustment, the poor and vulnerable can be Even as African countries begin to build up assisted without causing major distortions in their database on household social and economic economic mechanisms, which might threaten indicators, it is possible to undertake policy ac- macroeconomic discipline. In fact, policy reform tions which encourage greater participation by provides an opportunity to address the problems the poor in the newly emerging economic envi- of these groups anew. ronment, and which protect others who have been 121 Appendix: The social accounting matrix A SAM is a square matrix, or table, representing duction, consumption and accumulation) plus an the transactions taking place in an economy accountfortransactionswith the restof the world. during a certain period of time (Pyatt and Round, Figure A.1 is a schematic version of a very basic 1977). Each account of the system is represented SAM and has only five accounts. However, it by a row and a column of the matrix, which is does serve to illustrate the principal features of why it is square, with the convention that outlays the matrix accounting approach and also helps to are shown down the columns and receipts along tie in with our earlier discussion of the main the rows. The essence of the matrix accounting economic aggregates. The more interesting and approach is that each transaction is represented important features of SAMs are discussed subse- by a single entry. Thus if a producer pays wages quently when these five accounts are further dis- to a household, then the payment is shown at the aggregated to show more of the interdependence intersection of the producer's outlay account between households and the rest of the economic (column) and the household's income account system. For the present it can simply be noted (row). This is in contrast to the more conven- how the five accounts are structured to show the tional 'T' account system where this payment main elements of the circular flow of income be- would be recorded in two places, once in the tween institutions, production activities and the producer's outlay account and once in the rest of the world. Institutions are, of course, the household's income accounts, hence the "double principal transactors of the system. They are the entry" nature of that system. A further property economic units which are capable of owning as- of a SAM is that, because the accounts have to sets (real assets or financial claims) and incurring balance, row sums must equal column sums, liabilities on their own behalf. The individual providing of course that the rows and columns institutional units can be grouped to form broad follow the same ordering. These are simple yet institutional sectors, such as households, corpo- basic properties and are fundamental to all SAMs. rate enterprises and government. In Figure A.1 If a matrix does not obey these properties, then it we simply show one overall account for the com- is not a SAM which is why not all input-output bined institutions but we distinguish between tables are SAMs. They represent only the pro- their current account (their incomes and expen- duction part of the whole structure of transac- ditures) and their capital account. tions of the economic system. There are two accounts shown for production. In principle, a SAM can be any dimension al- One is an account for "products" which records though the minimum possible order would be the receipts from the sale of products and the four so that accounts can be specified for each of subsequent outlays of value added to factor in- the three basic forms of economic activity (pro- come recipients, while the second is a separate 122 Figure A.1 A basic SAM Outlays Institutions Production Rest of Current Capital Factors Products world Total 1 2 3 45 Current I Current FactorCurn Current tansersincome income t _ C t 1 hCraunrsefenrts (GNP) a Capital 2 Savings Capital Gross Capital 2 (S)transfers savings Domestic Net factor Gross Factors 3 product income from national A a _Y) abroad (NFP) product Products 4 Consumption Investments Intermediate Exports Use of (C) (I) products (X) products Net current Current External Rest of 5 transfers account Imports accornat world abroad (NTR) surplus (M) balnuce (BOP)banc Current ~~~~Gross Etra Total Current Gross national Supply of External outlay investment product products balance (GNP) account for "factors of production" which serves generation of income from domestic production. to receive the factor income generated either do- The outlay account for products shown as col- mestically or from abroad before paying this to umn 4 pays out the aggregate domestic product institutions. Thus the combination of the five to the factors account in row 3. The domestic accounts shows the real essence of the circular product therefore appears in the matrix as cell flow of income. In particular, the five show the (3,4). This is augmented by net factor income interaction between production and institutions, from abroad, cell (3,5), to forrn gross national and between the domestic economy and the rest product and this is recorded as the row total for of the world. They also show something of the the factors account. Gross national product is interrelationship between the two key markets of then distributed to form the primary income of the system which are important for empirical institutions, cell (1,3), although if incomes are analysis, i.e. for commodities and for the factors simply added across institutions, then there of production. would be double counting because the total would Let us briefly examine how the transactions exceed GNP by an amount equal to the aggregate can be read from the matrix and start with the current transfers between institutions. In a con- 123 solidated account for institutions as a whole, cur- first is to say more about the nature of transfers rent transfers simply net out from both their in- occurring in cell (1,1) while the other concerns comes and their expenditures, but they are in- the interactions between the household and fac- cluded here and appear as a diagonal entry in cell tor accounts in cell (1,3). The remaining cells in (1,1) of Figure A.1 to help facilitate subsequent the structure of the household accounts are fairly discussion. In a similar fashion, the institutions self-evident. For example, household expendi- capital accounts and the products accounts also tures are represented by cell (4,1) so that in a very contain diagonal entries. Capital transfers of both obvious sense and with appropriate disaggrega- real and financial assets take place between insti- tion, this cell would actually be a submatrix map- tutions and could be added to both gross savings ping expenditures from different types of house- and gross investment without disturbing the holds across different categories of products. overall balance. Similarly, in total, the sales and Similarly, household savings are shown in cell purchases of products for intermediate use net (2,1). out from both sides of the product balance. If we Cell (1,1) of the SAM, representing current ac- did this it would give us the basic expression count transfers, assumes a special significance shown as equation (1) in Chapter 6, but it could when institutions (and in particular, households) just as easily be left as a diagonal element. Again are disaggregated. The kinds of income transfers this is useful to do because in a more disaggre- that are captured are best illustrated by expand- gated framework these transactions enter the ac- ing cell (1,1) into a 3 by 3 submatrix to distinguish counts in the form of the intermediate transac- accounts for the household, corporate sector and tions of an input-output table. In all other re- government sectors. This submatrix is set out in spects, the cell entries are fairly self-explanatory. Figure A.2. For the household account the cells One can note a direct comparison between aggre- show that, in addition to factor income, house- gates in the cells in Figure A.1 and those which holds may receive transfer income in the form of appear in Chapter 6. distributed profits from the corporate sector, The basic structure of the SAM shown in Fig- certain cash benefits from government (say in the ure A.1 can be disaggregated in a variety of ways form of social security payments, together with depending upon the particular analytical and possible transfer income from other households policy focus. The most interesting disaggrega- which arise through various family support sys- tion of all is to explicitly identify households tems (Kusnic and Da Vanzo, 1980). Clearly the within the broad group of institutions and to pattern of income sources will vary considerably consider further disaggregation of that account across households, which will be central infor- across different household groups. In an ideal mation for any empirical analysis of the social world where information is readily available we dimensions of adjustment. Although the patterns could perform parallel sets of disaggregation clearly begin to emerge even with a fairly across both the current and capital accounts of aggregative SAM structure, an even finer disag- households (as well as those of other institutions), gregation of the accounts would provide more but on practical grounds it is usually preferable information on the nature of the interdependen- to limit disaggregation to the current accounts cies (or the lack of them), and how different types and to leave the capital accounts as a single con- of households depend to a greater or lesser degree solidated account. This is a point which will be on a variety of sources of income. taken up in later discussion in connection with The second component requiring special men- incorporating the flow of funds. For the present, tion is the submatrix represented by cell (1,3). if we view the first account in Figure A. 1 as a set This records the incomes received by households of household accounts then what we have is a and other institutions in return for the services schematic description of the sources of house- they provide in the factor markets. For house- hold incomes and the destination of their out- holds this will consist of labor income, income lays. It is this potentially quite detailed display from unincorporated enterprises and rent from of incomes and outlays within a framework of dwellings. But each household may include more general interdependence which makes the SAM than one individual who is an income earner. formats so attractive as a data framework. Moreover, these individuals may belong to quite There are two key parts of the basic schematic separate labor markets so that the composition of SAM which need to be discussed further. The household factor income could depend quite 124 Figure A.2 Institution current transfers submatrix of the SAM Institution outlays Total Households Corporate Government Interhousehold Distributed Goverrunent Household: Households transfers profits transfers, soial domestic security benfits receipts Goverrnment Corporate: Corporate 0 O transfers to domestic ol companies receipts Direct taxes, Direct taxes, Government: Government social security trading surplus of O domestic contributions public enterprise receipts Household Corporate GoveTrnment Total domestic domestic domestic outlays outlays outlays critically on any change in circumstances in the least in principle if not always in practice. For various labor markets. Technically, the same instance, a judicious choice of products accounts would be true of those individuals who have would reveal something of the full and detailed more than one job which is a phenomenon that circular flow that connects the distribution of in- becomes increasingly prevalent when expendi- come to the structure of production. Further- ture cuts in the public sector lead to retrench- more, it is now fairly standard practice to distin- ment and cuts in wage rates of government em- guish in the products accounts between com- ployees, at least in real terms. It is clear, there- modities and the activities which produce them. fore, that if the classifications are carefully cho- This distinction is made for a number of good sen and adequately reflect the main types of fac- conceptual and practical reasons, not least be- tor markets then this submatrix could represent a cause it means activities can be identified accord- rich source of information about the household ing to criteria other than the kind of commodity composition of income (Grootaert, 1982b). they produce (e.g., level of technology, owner- This brief synopsis of the SAM as a basic ship, etc.) and it also helps in handling the treat- framework for macroeconomic data is no more ment of secondary products and the valuation of than a sketch of its main features. As already commodity transactions. Similarly, while the emphasized the SAM really becomes a useful above discussion has primarily focused on disag- data construct when the very broad accounts il- gregations of the current institution, product and lustrated in Figures A.1 and A.2 are disaggre- factor accounts, it is important to note that the gated further. In this discussion we have empha- capital accounts can also be disaggregated fur- sized the institutions, households and factors ac- ther. Thus, they might show the generation of counts, and have alluded to their disaggregation savings by institution, cell (2,1); investment by because of their significance to the issues under- product, cell (4,2); and capital transfers between lying social-dimensions concerns. At the same institutions, cell (2,2), in a way that is analogous time it is clear that other parts of the matrix could to current transfers in cell (1,1). But it is some- also be disaggregated in a number of ways, at times useful to distinguish between types of 125 capital finance, in which case cell (2,2) would inconsistencies. But even if there is a genuine record the flow of funds between domestic insti- conflict between estimates from two key sources tutions (Greenfield, 1985). then the (statistical) ramifications of adopting one It should not be inferred from all of this that a rather than the other can be explored by con- fixed SAM design should be developed for every structing alternative SAMs. country or even that there should be a single (iii) Modelng A SAM is a data framework and framework to be used for all purposes in any one not a model of the economy. Indeed any particu- country. As a data construct, it is clear that a lar SAM could be consistent with a wide range of wide variety of different SAMs could be as- models, depending upon the behavioral relation- sembled depending, for instance, on the degree ships posited. But it has been demonstrated quite and extent of imputations one might choose to convincingly (de Melo, 1988; Pyatt, 1989) that carry out, or the purposes to which the SAM SAMs are a useful way to formulate as well as to might be put. Thus, for example, in some in- calibrate economywide models. However, it must stances we might require quite detailed financial be said that the matrix representation of accounts flows while in others a consolidated capital ac- is not of itself a sufficient condition for it to be count might be sufficient. It is this inherent flexi- useful and informative about structure. bility in the structure of a SAM that seems to offer The main limitation of a SAM is clearly its po- most scope in its role as an organising framework tentially voracious appetite for data, although as and data construct at the macro, meso, and micro we have seen, SAMs can be quite compact data levels. constructs. The main problem is being able to obtain, from available data sources, enough in- Advantages and limitations of SAMs formation to articulate the flows and therefore to show both origins and destinations of each set of There is now considerable practical experience transactions in accounting terms. Hence, for ex- gained in constructing SAMs across a wide vari- ample, if we want to examine the composition of ety of country situations. Based on this we can household income and trace this back through its identify three quite distinct advantages of pur- sources to production structure and commodity suing a SAM approach and really very few dis- demand, then we need quite specific information advantages or limitations, compared with al- about the various components of factorial and ternative systems. It must be emphasized that a transfer income. This is not very easily obtained. SAM is simply a way of representing macroeco- It should be noted, however, that the same kind nomic flows and, therefore, will not provide all of problem was faced in the early days of compil- the macro-level information we might require for ing input-output tables and, as a result, a meth- analysis as part of a single, unified framework. odology has now been established in carrying For example, information on stocks and assets out surveys of production establishments to cater would have to be compiled and shown sepa- for this. In the limit, if data do not exist then it rately, and the same would be true of all the may not be possible to construct a SAM of any relevant social and demographic data. substance. But equally, the lack of data would Consider first the advantages of the SAM ap- also severely limit the kinds of analysis one could proach: perform on any issues which require knowledge (i) Description Matrix accounts are generally about the circular flow of income. As such, ex- considered to be a very good schematic way of planations of social-dimensions issues would be visualizing the transactions and transfers taking among the first to suffer. place in an economy. The essential key to this is that the accounts are "articulated", by which it is Data requirements meant that both the origin and destination of transactions are identifiable. As already noted, one of the principal features of tii) Data assembly The accounting constraints a SAM is its role as an organizingframework for both inherent in a SAM provide an excellent basis for data and models. It can help us to identify both identifying inconsistencies between data derived gaps and weaknesses in the available data sources. from different and, sometimes, quite disparate In many instances these will already be apparent data sources. More controversially perhaps, it but so much depends on the degree of detail can sometimes help in resolving some of these required and the general taxonomy that has been 126 chosen before the full extent of data requirements it then allows a distinction between modem and and availability can be properly assessed. traditional technologies in production and be- In addressing the effects of macroeconomic tween those sectors which are predominantly policies on the well-being of different socioeco- under public rather than private ownership. nomic groups, it is clear from our discussion so Beyond these obvious but fairly crucial data far that considerable information is required to requirements the compilation of SAMs can pro- estimate the flows between the households, factors ceed on the basis of the same range of data sources and products accounts in particular. It is also usually required for standard national account- clear that a well-conducted and comprehensive ing purposes. Keuning and de Ruijter (1988) household survey is crucial in this regard. Given provide a useful and fairly detailed set of guide- that the main needs of data collection are toward lines for the construction of a SAM. These were households, it follows that the SAM is a conve- briefly reviewed earlier and would broadly con- nient way of integrating these data within a con- form to the requirements for any macroeconomic sistent macroeconomic framework. Furthermore, analysis. As so often happens, apart from identi- it demonstrates how, in a hierarchical data sys- fying complete gaps in data availability, one also tem, we observe data at the micro level being fed finds potentially useful data sets which cover into the system to generate the macro-level data only part of the desired universe or which relate constructs, and which brings out so clearly the to a different year from the one being considered. importance of the meso level at the interface be- There is no easy solution to how these data sets tween the two. should be incorporated, but it usually means that A second principal requirement is for some compromises, adjustments and assumptions have reasonably detailed information in order to fill to be made and then documented in case they out the products accounts. As indicated in our affect subsequent analysis. In the limit it is some- earlier review of data already available and the times suggested that the stylized facts about an particular difficulties most African countries face, economy (or, more fundamentally, "informed both conceptually and practically, the immediate guesses") may have to be resorted to in some problems are nontrivial. The data sources will be instances, especially in the short term and before wide-ranging but will constitute production sur- a program of data collection can get underway. veys of one form or another. It should be em- If this is so, then assembling such data in the phasized again, however, that it may be quite form of a SAM can help narrow down the range unnecessary to compile a highly disaggregated of choice and eliminate possible inconsistencies set of products accounts. Still, it is useful to along the way. distinguish commodities from activities because 127 Endnotes 1 The term "meso" is derived from the 8 The effects of adjustment on poverty un- Greek, "mesos" or middle. It therefore describes der quantity rationing are discussed briefly be- those elements which come between the macro low, but see also Demery and Addison (1988). and the micro. 9 Note, here "size" concerns the relative im- 2 Some countries have also suffered from portance of the country's exports and imports in military and civil instability, which have been relation to world markets. economically destabilizing. 10 For a discussion of the endogeneity of the 3 Africa's experience is discussed in detail tradables/nontradables and exportables/im- in World Bank (1981, 1984, 1986a and 1989) and portables divisions see Dombusch (1980: 94-5) World Bank/UNDP (1989). See also Balassa and and Timmer (1986, Chapter 4). McCarthy (1984), Liebenthal (1981), and Zulu and 11 While food is sometimes classified as a Nsouli (1985). nontradable good (see for example Lal (1986) on 4 Thus while African countries took some the Philippines), most analyses of Africa treat it exchange rate action over 1980-82, the real ex- as a tradable commodity - see for instance Lip- change rate nevertheless registered a 31 percent ton (1987), Norton (1987) and Bevan et al (1987a). increase between 1969-71 and 1981-83 for a repre- 12 For example Devarajan and de Melo (1987) sentative group of 14 countries (World Bank, in their model of three Francophone countries 1986b: 67). describe domestic manufactures as semitradables, 5 Botswana's rapid tightening of monetary because of the high rates of protection behind policy after the second oil price shock is one rare which they operate. For the use of this distinction example (see Harvey 1985). in applied general equilibrium models, see Dervis 6 In their survey of adjustment experiences et al (1982, Chapter 6) and Shoven and Whalley over 1980 to 1983, Zulu and Nsouli (1985:14) cite (1984: 1034), and Michel and Noel (1984: 21) for a number of examples of fund-supported pro- applications to the CMte d'Ivoire. grams during that period whose success was 13 The model we use here has its origins in limited by unforseen factors - including for ex- Dornbusch (1974, 1980), but follows closely the ample Madagascar (fall in coffee prices), Malawi extension and refinement by Collier (1988). (failure of the maize crop), Sierra Leone (unfa- 14 For convenience, money supply is meas- vorable weather), and Zimnbabwe (disruption of ured in units of foreign exchange in this model, the transport system). since it means that an x percent devaluation is 7 Corden (1986) considers that the flex-price equivalent to an x percent increase in domestic dependent-economy model is appropriate in money supply. many developing-country situations. 15 Comparing point C (in Figure 2.2) with 128 point A, it is clear that since P. is fixed, both P. 24 These relate to GDP growth, investment and P. must be higher, so that the demand for performance, savings performance, export money will be higher at C than at A. growth, real exchange rate, current account bal- 16 For simplicity, we are assuming that the ance, budget deficit, inflation, and extemal debt. loci drawn hold even when other markets are in These nine indicators measure performance in disequilibrium. This means that NN continues to four areas of policy concern - growth, external signify market-clearing values of relative prices balance, internal balance and external debt. even when the money market is characterized by 25 Balassa (1988: 17-20) also found evidence nonzero excess demands. In other words, NN for expenditure switching in SSA in the better and LL are assumed to signify effective as well as performance of agricultural output growth and notional equilibrium values. import substitution. 17 P/PnJis given by the slope of a ray through 26 Compare, for example, the results reported the origin. in Table 2.4a of World Bank (1988a) with those in 18 Edwards (1988a) shows that a terms of Tables 19 and 20 of World Bank/UNDP (1989). trade deterioration (improvement) has similar ef- 27 The reason why a devaluation is a neces- fects in the domestic economy to an increase (de- sary complement to a cut in absorption lies in a crease) in tariff rates (or their equivalent). The downward inflexibility in P.. If it were fully only difference is that the income effect is greater flexible, a demand contraction would be a suffi- under the former. cient corrective, since it would induce a real ex- 19 This assumes, of course, that the export- change rate depreciation through a fall in P.. But able sector is more labor intensive than either with Pg inflexible, the real exchange rate depre- importables or nontradables. ciation must be induced through an exchange 20 Where the rate of interest on loans is fixed rate devaluation - i.e., an increase in Pt. by the government rather than by the market (as 28 This assumes that there is no unemploy- in so-called 'repressed' banking systems), this ment in the domestic labor market, so that in- effect will work through the real loan interest rate. creased employment in tradables has to be drawn Thus, while the nominal rate remains fixed, the from nontradables labor. real cost of credit will fall in proportion to the 29 Long-run movements in the wage can also rate of intlation. be illustrated, with the flow of capital into the 21 It should be emphasized that these results tradables sector shifting the labor demand curves are largely illustrative, and depend on the as- - VMP, upward and VMP. downward. The fi- sumptions that were made. For example, they nal equilibrium would depend on the relative depend on the assumption that the deflationary factor intensities in the two sectors. effects of the terms of trade decline were more 30 This formal/informal-tradable/nontrad- than offset by the fiscal expansion; or that the able combination is obviously only one of a exportables sector is the most labor intensive, etc. number. Addison and Demery (1989b) analyze Our concern is to provide a structured basis for the full range of possibilities, including the case our thinking on these issues. What actually oc- where tradables and nontradables are both pro- curs in each country situation will of course de- duced in the formal and informal sectors. Again, pend on which of a number of important as- our purpose here is mainly illustrative - to dem- sumptions applies. onstrate how the framework can address this is- 22 Note that if devaluation were carried out sue. without fiscal and monetary contraction, the NN 31 This Harris and Todaro (1970) construct is curve would remain at N'N', so that excess de- now well established in the literature (e.g., mand will emerge for nontradables, E being to Edwards, 1988a), although the underlying as- the left of N'N'. This will raise P,, taking the sumptions it implies are not always made explicit. economy back to D. For devaluation to restore In particular, it assumes that all workers have an equilibrium at E, therefore, it must be combined equal chance of formal-sector jobs and that the with a contraction in aggregate demand shifting turnover rate in the formal sector is unity (see NN to the left, and LL upward. Addison and Demery 1989b for details). 23 On the theory of DUPs, see Bhagwati, 32 This is because the real wage in terms of Brecher and Srinivasan (1984), Srinivasan (1985), nontradables has risen, but it has fallen in terms which build on the earlier work of Krueger (1974). of tradables - recall that the rise in W,/P. is less 129 than the increase in P,/P., so that W,/P, falls. 43 See Guyer (1986,1988) and Fapohunda 33 This is consistent with one of the stylized (1988) for an anthropologist's perspective, and facts observed for developed countries, that in Jones (1986) and Dey (1981) for evidence from recession the formal sector takes the brunt of the North Cameroon and The Gambia respectively. employment adjustment and that the informal- 44 The items produced by household labor- formal wage gap widens (see McDonald and time are termed 'Z' goods in the literature, thus Solow 1985). distinguishing them from purchased commodi- 34 Financial sector policy conditions were not ties, usually called 'X' goods (Deaton and particularly prominent in adjustment lending in Muellbauer 1980: 245). Thus for example, SSA according to the World Bank (1988a) study. household healthcare (a Z good) is 'produced' 35 The real exchange rate appreciates under using household labor-time, and purchased import controls or tariffs through two processes: commodities - such as medicines (X goods). first, the nominal exchange rate will appreciate in The distinction between Z and X goods is impor- the face of the induced balance of payment sur- tant because it highlights the crucial role of plus; and second, the prices of nontradables will homework in transforming X goods into tend to rise as a result of import restrictions as consumables. consumers switch from importables to 45 See, for example, Yotopoulos and Lau nontradables. (1974), Barnum and Squire (1979), and Singh et al 36 Note, we confine ourselves here to the (1986b). See also Ellis (1988: 128) for a useful short-run case. We know that in the longer run, summary of these models. the effect on factor returns will depend on rela- 46 It is assumed that the household's endow- tive factor intensities. ment of the fixed factor/s of production is con- 37 An alternative approach in the structural- stant within the production period. Thus, the ist tradition would be to assume fix- and flex- curvature of the HPF reflects diminishing mar- price sectors (Taylor 1983). ginal returns to labor. 38 This may have some significance in the 47 The allocation of time between homework context of developing countries. For example, if and leisure can be shown by defining a separate the exportables sector is subject to some form of household production function for homework excess capacity, and if labor there is underem- (see King and Evenson 1983: 55). ployed, a devaluation can raise exportables out- 48 Incompleteness arises when private mar- put without requiring a labor transfer process kets fail to deliver a good or service, even though out of nontradables. The additional labor is ob- the cost of provision is less than what individuals tained from the unemployment (or rather, un- are willing to pay. deremployment) pool. 49 The analysis of such situations often gives 39 Depending on the length of the import- prominence to the demographic structures of the compression phase, there will have been some household since variations in the ratio of loss of productive capacity. For example trans- nonworkers to workers in the household will port and manufacturing equipment may have give rise to differences in the preference of work deteriorated through an inability to obtain im- against leisure, and thus different equilibrium ported spare-parts. Similarly, soil-quality and points for households otherwise faced with iden- crop-yields may have deteriorated. Thus, some tical production functions (Thorner et al 1966 and measure of reinvestment will be required, and it Nakajima 1969). may thus take several years for output levels to 50 However, there are exceptions - women recover. provide 35 percent of the labor on tobacco in 40 N*T* is therefore to be considered the 'ef- Swaziland and 37-47 percent of the labor on cot- fective' production frontier, since it is subject to ton in Malawi, while men provide 45-60 percent the output-reducing effects of the domestic rela- of the labor on upland rice in Sierra Leone (Guyer tive price distortions within the tradables sector. 1986:296). 41 The labels applied to the different con- 51 This assumption is made to simplify the cepts of the family are presented by Sen (1983a), analysis. In practice, it is common for the entire reprinted in Sen (1984a). household to work certain fields collectively, and 42 For a more extensive discussion of these the food produced or the income earned is dis- problems see Schultz (1989). tributed by the male household head. Individual 130 members also cultivate separate fields. The pro- to the concepts of the SAMs. duce is controlled by the cultivator concerned, 61 Note that NFP and NTR could be positive and often stored apart from the crop of the coop- or negative, but they are entered with these signs erative fields. For example in Burkina Faso, to be consistent with the treatment in Helmers McMillan (1987) in one area under study found and Dornbusch (1988; p. 380) that jointly worked fields accounted for 60 per- 62 Pyatt and Thorbecke (1976) discuss the cent of the total area planted and 60 percent of merits of wealth, sociological criteria and loca- recorded labor hours. Individually worked fields tion as a means of classifying households. accounted for the remainder. 63 On defining the objectives of adjustment 52 For example McSweeney (1979) reports policy, see World Bank (1988c), ECA (1989), ILO that in Upper Volta (now Burkina Faso) rural (1987) and Fischer (1987). women work an average total of 9.78 hours per 64 This is the main thrust of the alternative day as against 7.55 hours for men. Eicher and approach to adjustment suggested by UNICEF Doyle (1982) report evidence that women work (see Cornia et al 1987). nearly 30 percent more total hours over the year 65 On "categorizing" the poor under adjust- than men in rural Tanzania. ment, see World Bank (1988e) and Zuckerman 53 See Kuznets (1976) on using household (1988). income as the appropriate welfare unit when in- 66 The number of people in the subsistence come is pooled. economy who are relatively unaffected by ad- 54 Applying this technique to LSMS data on justment has almost certainly declined since 1980. the C8te d'Ivoire, Deaton (1987: 4,18) finds no Their self-reliance has been diminished consid- evidence that males are treated more favorably erably by the increased frequency of drought than females in the intrahousehold allocation of across the region. They have thus been forced to food. However, the allocation of adult goods is rely on outside help precisely at a time when heavily biased towards adult males: women, old governments have been least able to assist. men, and particularly old women, appear to have 67 For a discussion on the role of fiscal policy much less access. Sen (1984b) cites African ex- in achieving macroeconomic balance, see World amples of bias against both females and children Bank (1988f). This section draws on a recent (see also Crawford and Thorbecke 1980, on paper by Fischer and Easterly (1989). Kenya). Much of the evidence on the existence of 68 For instance, adjustment programs were such biases comes from South Asia (again see repeatedly blown off course in the early 1980s by Sen 1984b). fresh shocks (Zulu and Nsouli 1985). Do we as- 55 The group's elders often regulate marriage cribe output contractions to the shocks or to the transactions as well. Control over women, and policies put in place to respond to the shocks? therefore over the reproduction of future work- 69 For example, in Tanzania, one study has ers, is central to the success of labor-intensive shown that education is second only to cattle in agricultural systems (Swindell 1985: 38). determining household income variation within 56 Note, those neither buying nor selling in villages (Collier et al 1986). In C6te d'Ivoire the the initial equilibrium (that is consuming and educational attainment of the household head is producing at A) will gain, and will become net found to have a strong positive correlation with purchasers of labor services following adjustment. household consumption expenditures (Glewwe, 57 This implies that the marginal utility of 1987:18). And in Ghana the risk of food poverty leisure is zero over the relevant range. This as- falls as households' education rises (Kyereme and sumption is simply required to keep the exposi- Thorbecke 1987:1196). tion simple-in two dimensional space. 70 For instance, in Kenya it is estimated that 58 Note, in Figure 19, W/Pi (i = t,n) gives the only 10 percent of all borrowers from formnal credit real product wage (and therefore the real pur- institutions are rural women (World Bank 1989a). chasing power of the wage) in terms of each 71 The groups themselves select their own commodity. members, and the outside lending institution re- 59 See for example, Jarnison and Moock duces its risk and transaction costs by lending to (1984), Jamison and Lau (1982), Lockheed et al the group, which then on-lends to its members (1980) and Moock (1980,1986). (Desai 1983). 60 See the Appendix for a brief introduction 72 This means that their prices are mainly 131 determined by local market conditions, since re- 78 ILO (1986) and World Bank (1989d: 41) moteness implies little trade with other markets, estimates for 1985-2020. Out of 168 million em- and thus little influence on local subsistence food ployed people in 1985, agriculture employed pricesof changes in market conditions elsewhere. 131m, small/microenterprises 27m, and the 73 Chhibber et al (1989), for instance, show in modem wage sector lOin. By 2020 these num- their study of Zimbabwe how relatively small bers will be 311m, 206m, and 32m, respectively, macro-models can be used to simulate these ef- on current trends. fects. Although wage adjustments enter into their 79 Such off-farm employment is more impor- model, larger models involving greater disaggre- tant for poorer households than for the wealthy gation may be needed to deal with income distri- in Botswana, Nigeria and The Gambia. Evidence bution effects. from northern Nigeria, Sierra Leone, and Malawi 74 Although the initial wage improvement shows that off-farm income accounts for 50 per- may be small when unemployment characterizes cent of total income for the smallest landowners, the preliberalization situation. Note that mining and under 25 percent for the largest (Haggblade may be the exception with regard to the high et al 1987:12). labor intensity of exportables in general. 80 In Ghana and Zambia, for instance, women 75 Under this method all tariffs above a cer- account for over half those employed in nonfarm tain ceiling are reduced to that ceiling. In the enterprises. But in some societies their participa- next stage, a new lower ceiling is set, and all tion is low because of cultural factors. tariffs are reduced to it, and so on (World Bank 81 In effect, the wage rigidity becomes en- 1987b: 110 and Michaely 1986). dogenous to the market. 76 Of course, a devaluation must also accom- pany trade liberalization to maintain external balance. 77 Total agricultural employment figures are for 1985 (World Bank 1989d: 41), while the coun- try data refer to 1980 (see ILO 1986 and World Bank 1988f: 282). 132 Bibliography Addison, T., and L. Demery (1989a) "The Impact Balassa, B., and F. D. McCarthy (1984) Adjustment of Liberalisation on Growth and Equity," in G. Policies in Developing Countries, 1979-83: An Renshaw (ed.) Market Liberalisation. Equity and Update, World Bank, Washington D.C., Staff Development, Geneva, International Labour Working Paper No. 675 Office, p. 1-88. Barnum, H., and L. Squire (1979) A Model of an - (1989b) "The Poverty Effects of Adjustment Agricultural Household, Washington D.C., with Labor Market Imperfections," paper pre- World Bank. sented at World Bank EDI!Warwick Univer- Barro, R., and H. I. Grossman (1971) "A General sity DERC Conference on Labor Markets in an Disequilibrium Model of Income and Employ- Era of Adjustment, Warwick University, 8-10 ment," American Economic Reavew, Vol.61, p. August. 503-524. Ahmed, R., and N. Rustagi (1987) "Marketing Becker, G. (1965) "A Theory of the Allocation of and Price Incentives in African and Asian Time," Economic Journal, Vol.75 (September). Countries: A Comparison," in D. Elz (ed.) Ag- Becker, G. (1981) A Treatise on the Family, Cam- ricultural Marketing Strategy and Pricing Policy, bridge, Mass., Harvard University Press. World Bank, Washington D.C.: 104-18. Behrnan, J. (1988) "The Impact of Adjustrnent Anderson, D. (1987a) "Infrastructure Pricing Programs," in D.E. Bell and M.R. Reich (eds.) Policies and the Public Revenue in African Health, Nutrition and Economic Crisis, Auburn Countries," World Bank, Washington D.C., House Publishing Company, Dover, Massa- Africa Department, Processed (May). chusetts. (1987b) The Public Revenue and Economic Besley, T., and S.M.R. Kanbur (1988) "Food Sub- Policy in African Countries, World Bank Dis- sidies and Poverty Alleviation," in Economic cussion Paper No. 19, Washington D.C., (No- Journal, Vol. 98. vember). Bevan, D., P. Collier, and J. Gunning (1987a) Armington, P. (1969) "A Theory of Demand for "Consequences of a Commodity Boom in a Products Distinguished by Place of Produc- Controlled Economy: Accumulation and Re- tion," IMF Staff Papers, Vol. 16. distribution in Kenya 1975-83," The World Bank Balassa, B. (1988a) "Quantitative Appraisal of Economic Review, Vol.1 No. 3, (May), p. 489- Adjustment Lending," PPR Working Paper No. 514. 79, DECVP World Bank, Washington D.C. Bevan, D., A. Bigsten, P. Collier, and J. W. Gun- (August). ning (1987b) "Peasant Supply Response in (1988b) "Incentive Policies and Agricultural Rationed Economies," World Development, Performance in Sub-Saharan Africa," PPR Work- Vol.15 No.4 (April), p. 431-440. ing Paper No. 77, World Bank, Washington Bevan, D., P. Collier, and J. W. Gunning (1989) D.C. (August). Peasants and Governments: An Economic Analy- 133 sis, Oxford: Clarendon Press. rica," Statistical Appendix in World Bank Bhagwati, J. N., R. A. Brecher, and T. N. Srina- (1989d). vasan (1984) "DUP Activities and Economic Chambers, R. (1988) "Sustainable Rural Liveli- Theory," in D. Colander (ed.) Neo-classical Po- hoods: A Key Strategy for People, Environ- litical Economy, Cambridge Mass., Ballinger. ment and Development," in C. Conroy and M. Binswanger, H. (1989) "How Agricultural Pro- Litvinoff (eds.) The Greening of Aid: Sustainable ducers Respond to Prices and Government In- Livelihoods in Practice, London, Earthscan in vestments," paper Presented at First Annual association with the International Institute for World Bank Conference on Development Eco- Environment and Development: 1-17. nomics, Washington (April 27- 28). Chhibber, A., J. Cottani, R. Firuzabadi, and M. Binswanger, H. P., and J. McIntire (1987) "Be- Walton (1989) Inflation, Price Controls, and Fis- havioral and Material Determinants of Pro- cal Adjustment in Zimbabwe, Washington D.C., duction Relations in Land-Abundant Tropical PPR Working Paper No.192 (April). Agriculture.' Economic Development and Cul- Chhibber, A., and J. Khalilzadeh-Shirazi (1988) tural Change, Vol. 36 No.1 (October), p. 73-100. Public Finance in Adjustment Programs, World Birdsall, N., and W. P. McGreevey (1983) Bank, Washington D.C., PPR Working Paper "Women, Poverty and Development," in M. No.128 (December). Buvinic, M. A. Lycette, and W. P. McGreevey, Choksi, A.M., and D. Papageorgiou (1986) "Eco- (eds.) Women and Poverty in the Third World, nomic Liberalization: What Have We Baltimore, Johns Hopkins University Press, p. Learned?" in A. M. Choksi and D. 3-13. Papageorgiou (eds.) (1986) Economic Liberal- Bond, M. (1983) "Agricultural Responses to Prices ization in Developing Countries, Oxford: Basil in Sub-Saharan African Countries," IMF Staff Blackwell, P. 1-14. Papers, Vol. 30 No.4 (December), p. 703-726. Cleaver, K. (1988) "Agricultural Policy Reform Bourguignon, F., W. H. Branson and J. de Melo and Structural Adjustment in Sub-Saharan Af- (1989) "Macro-Micro Simulation Model," rica: Results to Date," World Bank, Washington OECD Development Center, Paris (mimeo). D.C. (August). Braverman, A., and J. Hammer (1986) "Multi- Collier, P. (1988) Macro-economic policy, employ- market Analysis of Agricultural Pricing Policies ment and living standards in Malawi and Tanza- in Senegal" in I.J.Singh et al (eds), 1986. nia, 1973-84, World Employment Programme (1988) "Computer Models for Agricultural Research Working Papers, International Em- Policy Analysis," Finance and Development Vol. ployment Policies Working Paper No.18, 25, No. 2 (June). International Labour Office, Geneva (March). Brown, C., C. Gilroy, and A. Kohen (1982) "The Collier, P., and D. Lal (1986) Labourand Povertyin Effects of the Minimum Wage on Employment Kenya 1900-1980, Oxford, Oxford University and Unemployment," Journal of Economic Lit- Press. erature, Vol. 20 (June), p. 487-529. Collier, P., S. Radwan, and S. Wangwe with A. Buiter, W. H. (1986) Macroeconomic Responses by Wagner (1986) Labour and Poverty in Rural Developing Countries to Changes in External Eco- Tanzania, Oxford, Oxford University Press. nomic Conditions, London, Centre for Economic Corden, W. M. (1985) Inflation, Exchange Rates and Policy Research, Discussion Paper No.93, the World Economy, Oxford, Clarendon Press, (February). Third Edition. Casley, D.J., and K. Kumar (1987) Project Moni- _ (1986) The Relevance for Developing Countries toring and Evaluation in Agriculture, John of Recent Developments in Macroeconomic Theory, Hopkins University Press, Baltimore. Canberra, Australian National University, Re- _ (1988) The Collection, Analysis and Use of search School of Pacific Studies Working Pa- Monitoring and Evaluation Data, John Hopkins per No.86/1. University Press, Baltimore. Corden, W. M. (1987) Protection and Liberalization: Casley, D. J., and D. A. Lury (1987) Data Collec- A Review of Analytical Issues, Washington D.C., tion in Developing Countries, Oxford, Clarendon IMF Occasional Paper No. 54 (August). Press, Second Edition. Cornia, G.A., R. Jolly, and F. Stewart, eds (1987) Chander, R. (1989), "Strengthening information Adjustment with a Human Face: Vol.1 - Protect- systems and basic statistics in Sub-Saharan Af- ing the Vulnerable and Promoting Growth, Ox- 134 ford, Clarendon Press for UNICEF. nomic Review, Vol. 1 No. 3 (May). Crawford, L., and E. Thorbecke (1980) "The Dey, J. (1981) "Gambia Women: Unequal Partners Analysis of Food Poverty: An Illustration from in Rice Development Projects?" in Journal of De- Kenya," Pakistan Development Review, Vol.19. velopment Studies, Vol.14, p. 109-122. Cuddington, J.T. (1980) "Fiscal and exchange rate Dixit, A.K. (1978) "The Balance of Trade in a policies in a fix-price trade model with export Model of Temporary Equilibrium with Ra- rationing" Journal of International Economics. tioning," Review of Economic Studies, Vol. 45, p. (1981) "Import substitution policies: a two- 393-404. sector, fix-price model," Review of Economic Dombusch, R. (1974) "Tariffs and Non-Traded Studies, Vol.48 pp 327-342. Goods" Journal of International Economics, No.4, Cuddington, J. T., P. Johansson, and K. G. Lbfgren p. 177-185. (1984) Disequilibrium Macroeconomics in Open (1980) Open Economy Macro-Economics, New Economies, Oxford, Basil Blackwell. York, Basic Books Inc. Deaton, A. (1987) The Allocation of Goods Within Dornbush, R, F. Leslie, and C.H. Helmers (1988) the Household: Adults, Children, and Gender, The Open Economy, EDI Series in Economnic Living Standards Measurement Study, Development, Oxford University Press for the Working Paper No. 39, World Bank, Washing- World Bank. ton D.C. Drud, A., W. Grais, and G. Pyatt (1986). "Macro- Deaton, A., and A. Case (1987) Analysis of economic Modelling based on Social Account- Household Expenditures, Washington D.C. ing Principles," Journal of Policy Modelling, Vol. World Bank, LSMS Working Paper No. 28. 8, No. 1. Deaton, A., and J. Muellbauer (1980) Economics Economnic Commission for Africa (1989) African and Consumer Behavior, Cambridge, Cambr- Alternative to Structural Adjustment Program- idge University Press. mes: A Framework for Transformation and Recov- de Ferranti, D. (1985) Paying for Health Services in ery, Addis Ababa, ECA-UN, (mimeo) (April). Developing Countries: An Overview, World Bank Edwards, S. (1988a) "Terms of Trade, Tariffs,, Staff Working Papers No. 721. and Labor Market Adjustment in Developing de Melo, J. (1988) "SAM-based Models: an Intro- Countries," World Bank Economic Review, Vol.2 duction," Journal of PolicyModelling,Vol. 10,No. No.2 (May) p. 165-87. 3, Fall. - (1988b) Exchange Rate Misalignment in De- Demery, L., and T. Addison (1987a) "Stabiliza- veloping Countries, World Bank Occasional Pa- tion Policy and Income Distribution in Devel- per No. 2/New Series. oping Countries," World Development, Vol. 15 Eicher, C. K., and D. C. Doyle (1982) Research on No. 12 p. 1483-1498. Agricultural Development in Sub-Saharan Africa: - (1987b) The Alleviation of Poverty Under A Critical Survey, Michigan State University, Structural Adjustment, World Bank, Washing- Dept. of Agricultural Economics, MSU Inter- ton D.C.. national Development Paper No.1. ____ (1988) Adjustment and Income Distribution: Ellis, F. (1988) Peasant Economics, Cambridge, the Role of Labour Markets, Development Eco- Cambridge University Press. nomics Research Centre, Working Paper No. Fapohunda, E.R. (1988) "The Nonpooling 81 (May), Warwick University. Household: A Challenge to Theory" in Dervis, K., J. de Melo, and S. Robinson (1982) D.Dwyer and J. Bruce (eds.) A Home Divided: General Equilibrium Models for Deuelopment Women and Income in the Third World, Stanford, Policy, Cambridge, Cambridge University Press Stanford University Press. for the World Bank. Feder, G., and R. Noronha (1987) "Land Rights Desai, B. M. (1983) "Group Lending in Rural Ar- Systems and Agricultural Development inSub- eas" in J.D. Von Pischke et al (eds.) (1983) Saharan Africa," World Bank Research Observer, Rural Financial Markets in Developing Countries, Vol.2 No.2 (July), p. 143-170. Baltimore, Johns Hopkins University Press for Fields, G. (1980) Poverty, Inequality and Develop- the World Bank, p. 284-88. ment, Cambridge, Cambridge University Press. Devarajan, S., and J. de Melo (1987) "Adjustrnent Fischer, S. (1987) "Economic Growth and Eco- with a Fixed Exchange Rate: Cameroon, C6te nomic Policy," in V. Corbo, M. Goldstein, and d'lvoire and Senegal," The World Bank Eco- M. Khan (eds.) Growth-Orientated Adjustment 135 Programs, Washington D.C., IMF/World Bank, Guha, S. (1986) "Employment Promotion through p. 151-78. Labor-intensive Public Works Programmes: Fischer, S., and W. Easterly (1989) "The Econom- Experiences, Problems, and Prospects in Af- ics of the Government Budget Constraint" rica" in ILO, The Challenge of Employment and (mimeo) Washington D.C., The World Bank Basic Needs in Africa, Nairobi, Oxford Univer- (November). sity Press, JASPA, p. 113-122. Foster, J., J.Greer, and E. Thorbecke (1984) "A Guyer, J. I. (1986) "Women's Role in Develop- Class of Decomposable Poverty Measures," ment" in R. J. Berg and J. Seymour Whitaker Econometrica, Vol. 52 No. 3. (eds.), Strategies for African Development, Gertler, P., L. Locay, and W. Sanderson (1987) Council on Foreign Relations and Overseas "Are User Fees Regressive? The Welfare Im- Development Council, p. 393421. plications of Health Care Financing Proposals _ (1988) "Dynamic Approaches to Domestic in Peru," Journal of Econometrics, 36, p. 67-80. Budgeting: Cases and Methods from Africa" Gertler, P., and J. van der Gaag (1988) Measuring in D.Dwyer, and J. Bruce (eds) A Home Divided: the Willingness to Payfor Social Services in Devel- Women and Income in the Third World, Stanford: oping Countries, World Bank, Living Standards Stanford University Press. Measurement Study, Working Paper No. 45. Guyer, J. I., and P. E. Peters (1987) "Introduction" Glewwe, P. (1987) The Distribution of Welfare in (to a special issue on Conceptualizing the the Republic of C6te d'Ivoire in 1985, World Bank, Household: Issues of Theory and Policy in Af- Living Standards Measurement Study, Work- rica), Development and Change, Vol. 18 No.2 ing Paper No.29. (April), p. 197-214. Glewwe, P., and D. de Tray (1987) "The Poor Haddad, L., and R. Kanbur (1989) How Serious is During Adjustment: A Case Study of the C6te the Neglect of Intra-Household Inequality? (What d'Ivoire," World Bank, PPR (mimeo), Novem- difference does it make to the measurement and ber. decomposition of inequality and poverty?), Uni- Greenfield, C. (1985) "A Social Accounting Ma- versity of Warwick, Development Economics trix for Botswana, 1974-75," in Pyatt and Round Research Center, Discussion Paper 95 (May). (eds.) Social Accounting Matrices: A Basis for Haggblade, S., and P. Hazell (1988) Prospects for Planning, World Bank, Washington D.C. Equitable Growth in Rural Sub-Saharan Africa, Greer, J. and E. Thorbecke (1986) "Food Poverty World Bank, Washington D.C., Agriculture and Profile Applied to Kenyan Smallholders," Rural Development Department, AGRAP Economic Development and Cultural Change, Economic Discussion Paper No.3 (February). Vol.35 No.1 (October), p. 115-141. Haggblade, S., P. Hazell, and J. Brown (1988) Gronau, R. (1973) "The Intrafamily Allocation of Farm/Non-Farm Linkages in Rural Sub-Saharan Time: The Value of Housewive's Time," Africa: Empirical Evidence and Policy Implications, American Economic Review, Vol. LXIII, No. 4. World Bank, Washington D.C., Agriculture and Grootaert, C. (1982a) The Conceptual Basis of Rural Development Department, AGRAP Measures of Household Welfare and their Implied Economic Discussion Paper No. 67 (May). Survey Data Requirements, Living Standards Harris, J., and M.P. Todaro (1970) "Migration, Measurement Study, Working Paper No.19, Unemployment and Development: a Two World Bank, Washington D.C. Sector Analysis," American Economic Review, (1982b) The Labor Market and Social Account- Vol. 60 p. 126-142. ing: A Framework of Data Presentation, Living Harvey, C. (1985) The Use of Monetary Policy in Standards Measurement Study, Working Paper Botswana, in Good Times and Bad, Brighton, No. 17, World Bank, Washington D.C. University of Sussex, Institute of Development _ (1986) Measuring and Analyzing Levels of Liv- Studies, Discussion Paper No. 204 (May). ing in Developing Countries: An Annotated Harvey, C., ed. (1988) Agricultural Pricing Policy Questionnaire, Living Standards Measurement in Africa, London, Macmillan. Study, Working Paper No. 24, World Bank, Havrylyshyn, 0. (1977) "Towards a definition of Washington D.C. non-market activities," Review of Income and (1987) "The Measurement of Levels of Liv- Wealth, Series 23, No. 1. ing for Policy Analysis," Development Re- Hayden, C.G., and J.I. Round (1982) "Develop- search Department (mimeo), World Bank, ments in Social Accounting Methods as Ap- Washington D.C. (June). plied to the Analysis of Income Distribution 136 and Employment Issues," World Development, - (1989b) The Challenge of Adjustment in Africa, Vol. 9, No. 6. Geneva, Tripartite Symposium on Structural Hazell, P.B.R., and A. Roell (1983) Rural Growth Adjustment and Employment in Africa (Octo- Linkages: Household Expenditure Patterns in Ma- ber). laysia and Nigeria, Washington D.C., Intema- IMF (1981) Taxation in Sub-Saharan Africa, Wash- tional Food Policy Research Institute, Research ington D.C., Occasional Paper No. 8. Report No.41. Jaeger, W., and C. Humphreys (1988) "The Effect Heller, P. S., A. L. Bovenberg, T. Catsambas, K-Y. of Policy Reforms on Agricultural Incentives Chu, and P. Shome (1988) The Implications of in Sub-Saharan Africa," American Journal of Fund-Supported Adjustment Programs for Pov- Agricultural Economics Vol. 70 No. 5 (Decem- erty: Experiences in Selected Countries, Wash- ber). ington D.C., International Monetary Fund, Jamal, V. (1988) "Getting the Crisis Right: Miss- Occasional Paper No.58 (May). ing Perspectives on Africa" International Henn, J. K. (1983) "Feeding the Cities and Feed- Labour Review, Vol.127 No.6, p. 655-678. ing the Peasants: What Role for Africa's Women Jamal, V., and J. Weeks (1988) "The Vanishing Farmers?," World Development, Vol.11 No.12 Rural-Urban Gap in Sub-Saharan Africa," (December), p. 1043-1055. International Labour Review, Vol.127 No.3, p.271- Herz, B., and R. Measham (1987) The Safe Moth- 92. erhood Initiative: Proposals for Action, World Jamison, D.T., and P.R.Moock (1984) "Farmer Bank, Washington D.C. Education and Farm Efficiency in Nepal," Higgins, J. M., et al (1982) Potential Population World Development, Vol. 12 No. 1, p. 67-86. Supporting Capacities of Lands in the Developing Jamison, D.T., and L.J.Lau (1982) Farmer Educa- World, Rome, Food and Agriculture Organisa- tion and Farm Efficiency, Baltimore, Johns tion. Hopkins University Press. Hill, T.P. (1977) "Do-it-yourself and GDP," Review Jones, C. W. (1986) "Intra-Household Bargaining of Income and Wealth, Series 25, No. 1. in Response to the Introduction of New Crops: - (1979) "On goods and services," Review of A Case Study from North Cameroon," in J. L. Income and Wealth, Series 25, No. 3. Moock (ed), Understanding Africa's Rural Horenstein, N. R. (1989) Women and Food Secu- Households and Farming Systems, Boulder: rity in Kenya, PPR Working Paper No. 232, Westview, p. 105-123. World Bank, Washington D.C. (June). Kahn, M.S., and M.D. Knight (1985) Fund Sup- Hossain, M. (1988) Credit for Alleviation of Pov- ported Adjustment Programs and Economic erty: The Grameen Bank in Bangladesh, IFPRI Growth, IMF Occasional Paper No. 41, Wash- Research Report No. 65, Washington D.C. ington D.C. (November). lFAD (1988) Structural Adjustment and Small- Kanbur, S. M. R. (1987a) "Measurement and Alle- holders in Sub-Saharan Africa: an IFAD Per- viation of Poverty: With an Application to the spective, background paper produced for the Impact of Macro-economic Adjustment," IMF JCGP high-level meeting, New York (Febru- Staff Papers. ary). - (1987b) "Transfers, Targeting and Poverty," (1986) Labour Force Estimates and Projections, Economic Policy, No. 4 (April), p. 111-148. 1950-2000, Geneva. (1987c) "Structural Adjustment, Macroeco- ..(1987) Background Document of High-Level nomic Adjustment and Poverty: a Methodol- Meeting on Employment and Structural Ad- ogy for Analysis," World Development, Vol.15 justment, Geneva (23-25 November). No.12 (December). - (1988a) Recent Trends in Employment, Equity, - (1988) "Poverty and the Social Dimensions and Poverty in African Countries, Geneva. of Structural Adjustment in C6te d'Ivoire" - (1988b) Distributional Aspects of Stabilisation (mimeo), Africa Region/SDA Unit, World Programmes in the United Republic of Tanzania, Bank, Washington D.C. (April). 1979-84, Geneva, ILO Mission Report. Keuning S., and W. de Ruijter (1988) "Guidelines - (1988c) International Labour Rezview, Special for the Construction of a Social Accounting Issue: The African Crisis, Food Security and Matrix," Review of Income and Wealth, Series 34, Structural Adjustment, Vol.127 No. 6. No.1. - (1989a) African Employment Report, 1988, Keuning S., and E. Thorbecke (1989) "The Impact Addis Ababa, JASPA. of Budget Retrenchment on Income Distribu- 137 tion in Indonesia: A Social Accounting Matrix Staff Working Paper No. 636, Washington D.C. Application," OECD Development Centre, _ (1986) "Stolper-Samuelson-Rybczynski in Paris (mimeo). the Pacific: Real Wages and Real Exchange King, E., and R. E. Evenson (1983) "Time Alloca- Rates in the Philippines, 1956-1978,' Journal of tion and Home Production in Philippine Ru- Development Economics, Vol.21 No.1 (April), p. ral Households," in M. Buvinic, M. A. Lycette, 181-204. and W. P. McGreevey (eds.) (1983) Women and _ (1988) "A Simple Framework for Analyzing Poverty in the Third World, Baltimore, Johns Various Real Aspects of Stabilisation and Hopkins University Press, p. 35-61. Structural Adjustment Policies" Journal of De- Klitgaard, R. (1989) "Incentive Myopia," World velopment Studies, Vol. 25 No. 3 (April) p. 291- Development, Vol.17 No.4 (April), p. 447-59. 312. Knight J. B. (1976) "Devaluation and Income Dis- Lele, U. (1985) "Termns of Trade, Agricultural tribution in Less-Developed Economies" Ox- Growth and Rural Poverty" in J. Mellor and G. ford Economic Papers, Vol. 28 No. 2, p. 200-227 Desai (eds) Agricultural Change and Rural Pov- (July). erty, Baltimore, Johns Hopkins University Press Koester, U., H. Schafer, and A. Valdes (1987) for IFPRI, p. 161-180. "Structural Adjustment and the Outlook for _ (1987) "Agriculture and Infrastructure," Agricultural Export Earnings in Africa," re- paper presented at the "Symposium on Trans- port prepared for Swiss Development Coop- portation and Structural Adjustment" orga- eration and Humanitarian Aid Agency (De- nized by the Transportation Department of cember). the World Bank, Baltimore, MD (6-8 May). Korayem, K. (1989) "Termns of Trade Shocks: A Liebenthal, R. (1981) Adjustment in Low-Income Graphical Analysis," (mimeo), Africa Region/ Africa, 1974-78, Staff Working Paper No. 486, SDA Unit, World Bank, Washington D.C. World Bank, Washington D.C. (August). Krueger, A. 0. (1974) "The Political Economy of Lipton, M. (1985) Land Assets and Rural Poverty, the Rent-Seeking Society" American Economic World Bank Staff Working Paper No. 744, Review, Vol. 64 (June). Washington D.C. - (1988) "The Relationship between Trade, (1987) "Limits of Price Policy for Agriculture: Employment, and Development," in G. Ranis Which Way for the World Bank?" Development and T. P. Schultz (eds.) The State of Develop- Policy Review, Vol. 5 No. 2 (June), p. 197-215. ment Economics: Progress and Perspectives, Ox- Lockheed, M., D. Jamison, and L. Lau (1980) ford, Basil Blackwell, p. 357-83. "Farmer Education and Farm Efficiency: A Krueger, A., M. Schiff, and A. Valdes (1988) "Ag- Survey" Economic Development and Cultural ricultural Incentives in Developing Countries: Change, Vol. 29 No. 1 (October). Measuring the Effect of Sectoral and Economy- Low, A. (1986) Agricultural Development in South- Wide Policies," World Bank Economic Review, ern Africa: Farm-household Economics and the Food Vol. 2 No. 3. Crisis, London, James Currey. Kusnic, M.W., and J. DaVanzo (1980) Income Ine- Mellor, J. W. (1985) "Determinants of Rural Pov- quality and the Definition of Income: The Case of erty: The Dynamics of Production, Technology, Malaysia, the Rand Corporation, R-2416-AID. and Price," in J. W. Mellor and G. M. Desai Kuznets, S. (1976) "Demographic Aspects of the (eds.) Agricultural Change and Rural Poverty, Size Distribution of Income" Economic Devel- Baltimore, Johns Hopkins University Press for opment and Cultural Change, Vol.25 No.1 (Oc- the International Food Policy Research Insti- tober), p. 1-94. tute, p. 21-40. Kyereme, S. S., and E. Thorbecke (1987) "Food Mellor, J. W., C. L. Delgado, and M. J. Blackie Poverty Profile and Decomposition Applied (eds.), (1987) Accelerating Food Production in to Ghana," World Development, Vol.15 No.9 Sub-Saharan Africa, Baltimore, Johns Hopkins (September), p. 1189-1200. University Press. Kyloh, R. H. (1989) "Flexibility and Structural McDonald, I.M., and R.M. Solow (1985) "Wages Adjustment through Consensus" International and Employment in a Segmented Labor Mar- Labour Review Vol. 128 No. 1, p. 103-123. ket," Quarterly Journal of Economics, Vol. C. No. Lal, D. (1984) The Real Effects of Stabilization and 4 (November), p. 1115-1142. Structural Adjustment Policies: an Extension of McMillan, D. E. (1987) "Monitoring the Evolu- the Australian Adjustment Model, World Bank tion of Household Economic Systems over 138 Time in Farming Systems Research," Develop- Norton, R. (1987) Agricultural Issues in Structural ment and Change, Vol.18 No.2, p. 295-312. Adjustment Programmes, Rome, FAO Economic McSweeney, B. G. (1979) "Collection and Analy- and Social Development Paper No.66. sis of Data on Women's Time Use," Studies in Pacey, A., and P. Payne (eds.) (1985) Agricultural Family Planning, Vol.10 No.11/12. Development and Nutrition, London, Hutchin- Meerman, J. (1979) Public Expenditures in Malay- son for FAO and UNICEF. sia: Who Benefits and Why, Oxford University Peters, P. E. (1986) "Household Management in Press for the World Bank, Washington D.C. Botswana: Cattle, Crops, and Wage Labor," in Meyers, K. (1986) A Reappraisal of the Sectoral Inci- J. L. Moock (ed.), 1986, p. 133-154. dence of Government Expenditure Cutbacks, CPD Pitt, M., and M.Rosenzweig (1986) "Agricultural Discussion Paper No. 1986-11, World Bank, Prices, Food Consumption, and the Health and Washington D.C.. Productivity of Indonesian Farmers," in Michel, G., and M. Noel (1984) Short-term Re- I.J.Singh et al, 1986. sponses to Trade and Incentive Policies in the Ivory Psacharopoulos, G. (1986) "Welfare Effects of Coast, World Bank Staff Working Paper No.647, Government Intervention in Education," Con- Washington D.C. temporary Policy Issues, Economic Analysis for Michaely, M. (1986) "The Timing and Sequenc- the Decision Maker, Vol. 4 No.3 (July), p. 51-62. ing of a Trade Liberalization Policy" in A.M. Pyatt, G., (1987) Accounting for Time Use, War- Choksi, and D.Papageorgiou (eds.) (1986), wick Economic Research Papers No. 316, Uni- Economic Liberalization in Developing Countries, versity of Warwick, United Kingdom. Oxford: Basil Blackwell, p. 41-59. Pyatt, G., and J.I. Round (1977) "Social Account- Moock, J.L. (1980) "Education and Technical Effi- ing Matrices for Development Planning," Re- ciency in Small-farm Production" Economic view of Income and Wealth, Vol. 23 No. 4. Development and Cultural Change, Vol. 20 No.4 _ (1985) Social Accounting Matrices: A Basis for (October), p. 723-39. Planning, World Bank Symposium Series, _ (ed.) (1986) Understanding Africa's Rural World Bank, Washington D.C. Households and Farming Systems, Boulder: Pyatt, G., and E. Thorbecke (1976) Planning Tech- Westview. niques for a Better Future, International Labour Mosley, W. H. and R. Jolly (1987) "Health Policy Office, Geneva. and Programme Options: Compensating for Republic of Ghana (1987) Programme of Actions to the Negative Effects of Economic Adjustment," Mitigate the Social Costs of Adjustment, Accra, in A. Cornia et al (eds.) (1987: 218-31). November. Mussa, M. (1986) "The Adjustment Process and Roemer, M. (1988) The Macroeconomics of Counter- the Timing of Trade Liberalization" in A.M. part Funds, HIID Development Discussion Pa- Choksi and D. Papageorgiou (eds.), Economic per No. 262, Cambridge MA, February. Liberalization in Developing Countries, Oxford: Rottenberg, S. (1986) "On Compensation for Basil Blackwell. Losses and Policy' in A.M. Choksi and D. Nakajima, C. (1970) "Subsistence and Commer- Papageorgiou (eds.) (1986) Economic Liberali- cial Family Farms: Some Theoretical Models zation in Developing Countries, Oxford: Basil of Subjective Equilibrium," in C. R. Wharton Blackwell, p. 128-148. (ed.) Subsistence Agriculture and Economic De- Ruggles, N., and R. Ruggles (1986) "The Integra- velopment, London: Frank Cass. tion of Macro and Micro Data for the House- Nath, S. (1988) "Welfare Economics and Struc- hold Sector," Review of Income and Wealth, Se- tural Adjustment" (mimeo), Dept. of Econom- ries 32, No. 3. ics, University of Warwick. Sahn, D.E. (1989) "Towards Understanding the N'cho-Oguie, C. (1989) "Integrating Social Di- Implications of Macroeconomic Adjustment on mensions in Macroeconomic Adjustment Poverty and Food Security in Sub-Saharan Af- Policy Analysis: Towards a Modelling Strat- rica" (mimeo), Cornell University Food and egy" (mimeo), Africa Region/SDA Unit, World Nutritional Policy Program, Washington D.C. Bank, Washington D.C. (September). (January). Neary, J.P. (1980) "Non-Traded Goods and the Schaefer-Kehnert,W.(1982) "SuccesswithGroup Balance of Trade in a Neo-Keynesian Tempo- Lending in Malawi," Development Digest, rary Equilibrium," Quarterly Journal of Eco- (January) p. 10-15, reprinted in J. D. Von nomics, Vol.95, p. 403-29. Pischke et al (eds.) (1983) Rural Financial Mar- 139 kets in Developing Countries, Baltimore, Johns Stewart, F. (1987b) "Supporting Productive Em- Hopkins University Press for the World Bank, ployment among Vulnerable Groups" in p. 278-83. Cornia et al (eds.) (1987:197-217). Schultz, T.P. (1989) Women and Development: Ob- Stiglitz, J.J. (1988) "Economic Organization, In- jectives, Frameworks and Policy Interventions, formation and Development" in H.B. Chenery World Bank, PPR Working Paper, Women in and T.T. Srinivasan (eds.) Handbook of Develop- Development: Washington D.C. (April). ment Economics: Volume 1, Amsterdam, North- Scobie, G.M. (1989) Macro-economic Adjustment and Holland, p. 93-160. the Poor: Toward a Research Strategy, CFNPP Stone, J. R. N. (1949) "Comptabilite sociale, ag- Macrograph 89-1, Ithaca, New York. gregation et invariance," Economie Applique, Scott, C. (1989) "Estimation of Annual Expendi- Vol. 11 No. 1. ture from One Month Cross-Sectional Data in Swindell, K. (1985) Farm Labour, Cambridge, a Household Survey" (mimeo), Africa Region/ Cambridge University Press. SDA Unit, World Bank, Washington D.C. Taylor, L. (1983) StructuralistMacroeconomics,New Sen, A.K. (1983a) "Economics and the Family," York, Basic Books. Asian Development Review, Vol.1 No.1, p. 14- Taylor, L (1988) Varieties of Stabilization Experi- 26, reprinted in A. Sen (1984a: 369-385). ence, Clarendon Press, Oxford. _ (1983b) "Poor, Relatively Speaking," Oxford Thomas, E.S. (1989) "Zaire's Economic Liberali- Economic Papers, Vol.35 No.2 (July), p. 153-169. zation and its Impact on the Agricultural Sec- ____ (1984a) Resources, Values and Development, tor," Development Policy Review. Oxford: Basil Blackwell. Thomas, E., and W. Weidemann (1988) "Impact = (1984b) "Family and Food: Sex Bias in Pov- of Zambia's Economic Policy Reform Pro- erty," Chapter 15 of Sen, 1984a, p. 346-368. gramme on the Agricultural Sector," Develop- (1987) The Standard of Living, Oxford, Basil ment Policy Review, Vol. 6 No. 1 (March). Blackwell. Thomas, V, and A. Chhibber (eds.) (1989) Ad- Serageldin, I. (1989) Poverty, Adjustment and justment Lending: How It Has Worked, How It Growth in Africa, World Bank, Washington D.C. Can Be Improved, World Bank, Washington Shalizi, Z., and L. Squire (1987) "A Framework D.C.. for Tax Policy Analysis in Sub-Saharan Af- Thorbecke, E. (1985) "The Social Accounting Ma- rica" (mimeo), Country Policy Department, trix and Consistency-Type Planning Models," World Bank, Washington D.C. (March). in G. Pyatt and J.1. Round (eds.), Social Ac- Shoven, J. B., and J. Whalley (1984) "Applied counting Matrices: A Basis for Planning, World General-Equilibrium Models of Taxation and Bank Symposium Series, World Bank, Wash- International Trade," Journal of Economic Lit- ington D.C., p. 207-256. erature, Vol.22 No.3 (September), p. 1007-1051. Thorner, D., B. Kerblay, and R. E. F. Smith (1966) Singh, I., L. Squire, and J.Strauss (1986a) "A Sur- Chayanov on the Theory of Peasant Economy, vey of Agricultural Household Models: Re- Homewood, Illinois, Richard D. Irwin. cent Findings and Empirical Implications," Timmer, C. P. (1986) Getting Prices Right: the Scope World Bank Economic Review, Vol.1 No.1, p. 149- and Limits of Agricultural Price Policy, Ithaca, 179. Cornell University Press. Singh, I., L. Squire, and J.Strauss (eds.) (1986b) Tuinenburg, K. (1987) "Experience with Food Agricultural Household Models: Extensions, Ap- Strategies in Four African Countries," in J. P. plications, and Policy, Baltimore, MD., Johns Gittinger, J. Leslie and C. Hoisington (eds.) Hopkins University Press. Food Policy: Integrating Supply, Distribution and Smith, V., and J. Strauss (1986) "Simulating the Consumption, Baltimore, Johns Hopkins Uni- Rural Economy in a Subsistence Environment: versity Press for the World Bank, p. 497-508. Sierra Leone" in I.J. Singh et al (eds.), 1986. UNICEF (1985a) Within Human Reach: A Future Srinivasan, T. N. (1985) "Neo-classical Political for Africa's Children, New York, United Nations Economy, the State and Economic Develop- Children's Fund. ment," Asian Development Review, Vol.3 No.2. (1985b) Adjustment Policies and Programmes Stewart, F. (1987a) "Monitoring and Statistics for to Protect Children and Other Vulnerable Groups Adjustment with a Human Face," in Cornia et in Ghana, Accra. al (eds.), (1987, 257-272). (1987) Annual Report, New York. 140 _ (1989) The State of the World's Children, 1989, _ (1987b) World Development Report, 1987, Oxford, Oxford University Press for UNICEF. Washington D.C. UNICEF/WFP (1986) Food Aid and the Well-Being _ (1988a) "Report on Adjustment Lending," of Children in the Developing World, New York. World Bank, Washington D.C., Country Eco- United Nations S.O. (1975) Towards a System of nomics Department, PPR. Social and Demographic Statistics, Studies in _ (1988b) Adjustment Lending: An Evaluation of Methods, Series F, No.8, New York. Ten Years of Experience, Washington D.C. - (1968) A System of National Accounts, Studies Country Economics Department, Policy and in Methods, Series F, No.2, Rev. 3, New York. Research Paper No.1 (December). United Nations DTCD (1989) National Household _ (1988d) Education in Sub-Saharan Africa: Survey Capability Programme, Household Income Policies for Adjustment, Revitalisation, and Ex- and Expenditure Surveys: A Technical Study, pansion, Washington D.C. Department of Technical Cooperation for De- (1988e) Targeted Programs for the Poor During velopment and Statistical Office, New York. Structural Adjustment, Washington D.C. - (1989) National Household Survey Capability _ (1988f) World Development Report, 1988, Programme, Measuring Literacy through House- Washington D.C. hold Surveys, Department of Technical Coop- _ (1989a) Kenya: The Role of Women in Economic eration for Development and Statistical Office, Development, Washington D.C., Country Study New York. (May). Verma, V., T. Marchant, and C. Scott (1988) _ (1989b) "How Adjustment Programs can Evaluation of Crop-cut Methods and Farmer Re- Help the Poor: The Experience of the World ports for Estimating Crop Reduction. Results of a Bank," Washington D.C., Strategic Planning Methodological Study in Five African Countries, and Review Department, processed (August). LONGACRE Agricultural Development Cen- _ (1989c) "Prospects for Prinary Commodities of tre Report, London. Relevance to the African Continent," IECCM, World Bank (1981) Accelerated Development in Sub- processed. Saharan Africa: An Agenda for Action, World _ (1989d) Sub-Saharan Afica: From Crisis to Bank, Washington D.C.. Sustainable Growth, Washington D.C. _ (1984) Toward Sustained Development in Sub- World Bank/UNDP (1989) Africa's Adjustment and Saharan Africa: A Joint Program of Action, World Growth in the 1980s, Washington D.C., World Bank, Washington D.C.. Bank. _ (1986a) Financing Adjustment with Growth in World Bank/WHO/UNFPA (1987) Preventing the Sub-Saharan Africa, 1986-90, Washington D.C. Tragedy of Maternal Deaths: a Report on the Inter- World Bank. national Safe Motherhood Conference, Nairobi - (1986b) World Development Report, 1986, (February). World Bank, Washington D.C.. Yotopoulos, P., and L. Lau (1974) "On Modeling ____ (1986c) Financing Education in Developing the Agricultural Sector in Developing Coun- Countries, Washington D.C. tries" Journal of Development Economics, Vol. 1 ____ (1986d) Population Growth and Policies in p. 105-27. Sub-Saharan Africa, Washington D.C. Zuckerman, E. (1988) Poverty and Adjustment: Is- _ (1986e) Poverty and Hunger: Issues and Op- sues and Practices, World Bank, Washington tions for Food Security in Developing Countries, D.C., processed. Washington D.C. (February). Zulu, J. B., and S. M. Nsouli (1985) Adjustment _ (1987a) Financing Health Services in Develop- Programs in Africa: The Recent Experience, ing Countries: An Agenda for Reform, Washing- Washington D.C., International Monetary ton D.C. Fund, Occasional Paper No.34. 141 The World Bank Headquarters European Office Tokyo Office 1818 H Street, N.W. 66, avenue d'lena Kokusai Building Washington, D.C. 20433, U.S.A. 75116 Paris, France 1-1, Marunouchi 3-chome Chiyoda-ku, Tokyo 100, Japan Telephone: (202) 477-1234 Telephone: (1) 40.69.30.00 Facsimile: (202) 477-6391 Facsimile: (1) 47.20.19.66 Telephone: (3) 214-5001 Telex: WIUI 64145 WORLDBANK Telex: 842-620628 Facsimile: (3) 214-3657 RCA 248423 WORLDBK Telex: 781-26838 Cable Address: INTBAFRAD WASHINGTONDC Cover design by Joyce C. Petruzzelli ISBN 0-8213-1640-0