Official PDF , 59 pages - World bank documents

12 downloads 6669 Views 3MB Size Report
The economy is assumed to consist of two classes of agents: the poor with income ... equity is assigned a relatively high weight because oL concerns about racial ..... Cut in socidal nvmenls (schools, clinirs) ng in 1986 .... Cte d'lvoire (S1,180t.
Public Disclosure Authorized Public Disclosure Authorized Public Disclosure Authorized Public Disclosure Authorized

Policy Research

WORKING PAPERS ยง

Policy ~~~Trade

CountryEconomicsDepartment Ti.e Wor!dBank November1991 WPS810

Poverty and IncomeDistribution duringAdjustment Issuesand Evidence fromthe OECD Project FrancoisBourguignon Jaime de Melo and Christian Morrisson

Adjustment programs will fail when they do not recognize the interdependence of the three criteria of tfficiency, welfare, and political feasibility. These programs must be tailored to both the political and economic environments of each country.

Policy Research Working Papers disseminate tcliefndings of work il progress and encourage the exchange of ideas among t3ank staff and all others interested in developmentissues. These papers, distributed by the Research AdvisoTyStaff, carry thc nanes of the authors, reflect only theirviews, and should be used and cited accordmngly.Thcfindings, interpretauons,and conclusions are the authors' own.T hey should not be attributed to the World Bank, its Board of DJirectors,its management, or any of its member countries.

PolicyResearch

Trade Policy WPS 810

This paper- a product of the Trade Policy Division, Country Economics Department- is part of a larger effort in the Bank to analyze the effects of alternative adjustment packages on poverty and on the distribution of income. This research was funded by the World Bank's Research Support Budget, "Trade Reforms in SALs: A Positive Analysis of Performance and Sustainability" (RPO 675-32). Copies are available free from the World Bank, 1818 H Street NW, Washington DC 20433. Please contact Dawn Ballantyne, room N10-019, extension 37947 (53 pages). November 1991. Drawing lessons from country studies, Bourguignon, de Melo, and Morrisson examine the effects of adjustment policies on the distribution of income in Chile, Cote d'lvoire, Ecuador, Indonesia, Malaysia, and Morocco. After analyzing the issues that must be confronted in designing adjustment programs with a focus on poverty, they synthesize the main conclusions of the different country studies. With simulation exercises they explore the effects of the design of the adjustment packages on poverty and on the sustainability of the measures undertaken in these countries. These

exercises show considerable diversity in the evolution of income distribution during adjustment. They also expose the fatal flaws of narrowly designed adjustment programs. Adjustment programs - whether focused on efficiency or on welfare - will fail when they do not recognize the intk fJependenceof the three criteria of efficiency, welfare, and political feasibility. Adjustment programs must be carefully packaged to fit country circumstances, taking into account both the political and economic environments.

The Policy Research Wo-king Paper Series disseminatcs the findings of work under way in the Bank. An objective of thc series is to get these findings out quickly, even if presentations arc less than fully polished. The findings, interpretations, and conclusions in these papers Lo not necessarily repfesent official Bank policy. . Produced by the Policy Research Dissemination Center

Table ot Contents Page No.

1.

Introduction

2.

Issues in the Designof AdjustmentPrograms with a Poverty Focus

3

The Case Studies: A Diversityof Initial Conditions and Shocks

16

4.

DistributionalOutcomesof Adjustment: A Synthesis

25

5.

Lessons from the Countirfactual Simulations

34

6.

Conclusions

43

Notes

48

References

51

3.

List of Figr Figure 1

Eq. y-EfficiencyTradeoffs

6

Figure 2

ExpenditureSwitching,ExpenditureReduction, and Structural Adjustment

9

Figure 3 Figure 4

Adjustmentand the Factoral Distribution of Income

12

Real ExchangeRate Depreciation,Tastes, and Incowe Distribution

15

List of Ta&h Table 1 Table 2

Country Studies: Shocks,Adjustment,and DistributionalIndicators

Is

MacroeconomicIndicators During Adjustment: A Comparisonwith Other Countries

24

Supportfrom the OECDDevelopmentCenter and the World Bank under RPO 675-32 is gratefully acknowledged. The views are those of the authors, not those of their respectiveaffiliations, nor those of the supporting agencies. We thankAlain de Janvry, Patricio Meller, ElisabethSadoulet, and Erik Thorbecke for manyhelpful suggestionsand Rebecca Sugui for unfailing logistic support.

1. INTRODUCTION

It is widely acceptedthat investmentin human capital and social services, when accompanied by growth, remain the primary means to reduce poverty and improve the quality of life. However, during the turbulent 1980swhen the majorityof developingcountries faced severe macroeconomic crises, growth slowed markedly,to the point where many countriesno longer registered positive per capitagrowth. In this environment,living standardsoften fell sharply, particularlyamong the poor. Early on, while most of internationalinstitutionswere still focusingon how the developingworld was going to reestablishmacroeconomicbalanceand service its external debt, the OECD Development Centre launched a study of the effects of adjustmenton incomedistrib!.ion and poverty. When the OECD project was launched, little was known about how adjustmentprograms affected ik.comedistribution. To begin with, no economiccrisis like this one had occurred before. For the first time, policies to alleviatepovertyhad to take into accountthe need to stabilizethe economyand to promote restructuringto restore (or establish)long-run growth through improved resource allocation. Secornd,the tools neededto analyzethe joint effectsof stabilizationand microeconomicstructural adjustmentpolicies we- not yet developed.' Becausethe macroeconomiccrises had pt,- .sive effects, analyzingthem required an economywideapproachthat went beyondthe descriptiveapproachused, by Cornia, Jolly, and Stewart (1987) among others, in their UNICEF study.2 Most adjustmentprograms, whetherthey were supportedby the IMF and the World Bank or not, emphasizedthe simultaneousapplicatior of stabilizationand structural adjustmentmeasures. Stabilizationpolicies emphasizeddemand management,while structural adjustmentprograms emphasizedmicroeconomicreforms designedto have supply-augmentingeffects. Despitethis difference, however, stabilizationand structural adjustmentmeasures are not always easy to separate; exchangerate policies, for example, are a fundamentalelementof both IMF-supportedstabilizationpackagesand Bank-supportedstructural

2 adjustmentpackages. Consideringnot only these analyticalissues but also the difficultyof obtaining reliable informationon povertyand incomedistritution - especiallybefore and after adjustment(and the emotionallychargednature of the topic of incomedistribution), it is understandablethat we were poorly informedabout the likely distributionaleffects of adjustment. The OECD project was designedto remedy this situation. A two-prongedapproach was used that combined countrystudies and simulationanalysis. For the countrystudies, countrieswere selectedto provide wide geographicalcoverage and variationin initial conditions,to give a relatively detailedpicture of the variety of adjustmentexperiences. Authors were asked to draw on all available informationon distributionalshifts and povertyindicatorsduring adjustmentin their interpretationsof events. Country authors were also encouragedto rely on simulationmethodsand counterfactual analysisto compensatefor the general lack of informationon distributionalindicators during adjustment and to provide a more informed and dispassionateassessmentof each country's adjustmentexperience. These methodscould also be used to isolate the effects of various policies. This symposiumbrings together a selectionof the country studies commissionedby the project. In this introductionto the symposium,we first discussthe channelsthrough which the stabilization/structuraladjustmentprograms are likely to have affected the distributionof income. This analyticaldiscussionis essentialfor understandinghow adjustmentis likely to get translated into distributionalshifts across householdsand how the linkagesbetween stabilizationand structural adjustmentpolicies were modeled in the simulationanalyses. We then discussthe six country studies presented in the volume,3 recallingin section 3 the initial conditionsand diversityof disequilibria facing each country. In section4, we synthesizethe main conclusionsfrom the inerpretative descriptionin the country studies. Then in section 5, we turn to the lessons from the simulation exercises used extensivelyin all but one country study (Chile) to evaluatethe effectsof the selected

3 adjustmentpackageson povercyand incomedistributionand to examinewhether alternativepolicies would have been preferable. Conclusionsfollow in section6.

2. ISSUESIN THE DESIGNOF ADJUSTMENTPROGRAMSWITH A POVERTYFOCUS

This section introducesthe issuesthat had to be confrontedin the design of adjustment packageswith demand managementand supply-sidecomponents. Emphasisis on the tradeoffs faced and on the role of rigiditiesin increasingdistributionalshifts and makingadjustmentboth more difficultand less sustainable,becauseof resistanceto cuts in living standards. We start with a discussionof the tradeoffs betweenequity and efficiency,at one point in time and then intertemporally,and close with a discussionon the selectionof poverty alleviationmethods. This discussionalso identifiesthe informationthat wouldbe needed to comprehensivelyevaluatethe longrun distributionaleffects of adjustmentpackages. Even though not enoughtime has elapsed to evaluatethe long-run effects of adjustmentpackageson income distribution, the emphasison growth is warrantedbecause these programs emphasizesupply-sideeffects. Only by focusingon long-run growth are we able to see the eventualtradeoffsfaced by the growth-orientedpackages of the 1980s. Next, we considerthe short-run adjustmentto the externalshocksthat were at the origin of the crisis, concentratingon the role of initial conditions,rigidities, and tastes in determiningthe impactof adjustmenton the factoraldistributionof income. In many ways, short-run tradeoffsdominatedthe measuresadopted, and much of the discussionin the remainderof the paper will deal with identificationand analysisof these tradeoffs, since these are the only ones that can be identifiedat this stage.

4 (a) Growthand distributionin the long-run: tradeoffsbetween efficiencyand equity'

As we argue in more detail below (in 3ect.Un2b), the difficultyfaced by the majority of developingcountriesduring the 1980swas how to adjust to a particularly strong external shock that combineda deteriorationin the terms of trade with an increase in interest rates on foreign debt and for many countries, the foreclosingof access to foreign funds. for economieson a normal growth path with resources at or close to full employment,adjustmentrequired a shift in resources toward the tradables sector. For economieswith idle capacity, adjustmentrequired, in addition, structural reforms with supply-augmentingeffects. An efficientresource transfer towards tradables would result in a net foreign exchangesavings and wouldallow for interestpaymentson the foreign debt, which couldno longerbe rolled over because of insufficientforeign financing. To induce a shift in resources toward the tradables sector, the relative price of tradables had to rise. Also, absorptionhad to be reduced relative to income. In the absence of access to foreign borrowing abroad, the easiest way to speed up adjustmentis to increasethe rate of investmentby increasingthe rate of domestic saving. But since the overal!ievel of expendituresalso had to be reduced, such a strategy wouldhave resulted in a drastic cut in consumptionexpenditures, which would already have fallen becauseof income loss resultingfrom the externalshock. Poverty would certainlyincrease. These efforts to achieveefficientoutput levels are likely to conflictsharply with efforts to alleviatepoverty during the adjustmentperiod.5 We now illustrate the dilemmasfaced under those circumstances. As the discussionof the countrystudies will show, the relative importanceattached to equity and efficiencyin the policymakers'intertemporalsocial objectivesvaried: concern for equity (increasingthe consumptionof the poorest) weighedmore heavily in, say, Malaysiaand Indonesia than in Cote d'Ivoire and Morocco which emphasizedincreasingefficiency(growth in per capita

5 consumption)by reducing macroeconomicdisequilibria. It is therefore useful to begin our discussion of the tradeoffs faced during adjustment. Redistributenow or later? If now, at wi it distortionary cost and through what means - by transfers or by investmentin the poor? Consider the simple analyticrepresentationin figure 1 of the three tradeoffsmentionedabove. The economy is assumedto consist of two classes of agents: the poor with incomey. and the nonpoor with income yn. Assumethat populationshares (np,n,,) are constant, so that different levelsof aggregateincome are representedby the family of iso-incomelines with constant slope /n.1 . Now supposethat if no policy decisionis taken, adjustmentwil, be distribution-neutral(i.e., the economy will move along the ray ON. (Later we will deal in more detail with the mechanismsthrough which adjustmentitself affects the factoraldistributionof income.) The economyis initiallyat Al. The social indifferencethrough A, reflects the policymakers'evaluationof equity. Consider now a policy aimed at reducing povertythrough a transfer of current incometo the poor. The static tradeoff betweenpoverty reduction and efficiencyis representedby the movefrom A, to B,. At B1, the incomeof the poor has risen, but the mean income of the populationis lower since the economyis now on a lower iso-incomeline. The differencebetween the slope of the segment A1 B, and that of the lines L representsthe contemporaneousmarginalefficiencycost of reducingpoverty, that is both the leakagedue to imperfecttargeting and the distortionarycost of transfers. So long as B, lies on a higher social indifferencecurve than A1 as in figure 1, current transfers to the poor are preferred to noninterver..ion. It is already clear that evaluatingwhether adjustmentpolicies should be accompaniedby an activist redistributivepolicy is going to be difficult. For example, in economieslike Malaysia, where equity is assigneda relativelyhigh weight becauseoLconcernsabout racial imbalances,the social indifferencecurve through A1 is likely to be steep and so redistributivepolicies are likely to be undertaken. By contrast, in economieswith relatively weak administrativeand fiscal systems, the

Figure 1 Yn

Equity-EfficiencyTradeoffs Distribution-neutral Growth: or one-shot non-intervention

N

Income of

current

transfer

the non-poor

Capital transfer

A~ ~~

~

~

~

strategy

~

~ ~ IS-ncm

L

lines (L)

2~~~~~~~~~~~~~~~~~~~~~~

2~~~~~~~~~~~~~~L0 Social / Indifferencj

, >,,

\

02 ~~~~L

1~~~~~~~~~~~~~~~~~~~~~~~~~L /2.

\

Income of the poor B

=

L B2

Y LAi

current transfer; C1- capital transfer {Investment in human capital of the poor)

P

7 segment A, B1 is likely to be steep, and redistributivepolicies will not be undertakenbecause of the need to re!y on highly distortionarytax policies to effect the current transfer.6 Consider now the intertempora'tradeoff. The no-interventionpath leads to A2,whi!e B2 representsthe outcomeof having redistributedcurrent income during period 1. Here it is assumed that the redistributionof current incomeleaves the primary distributionof income in future periods unaffected. As depictcd here, the dynamiccost of current transfers is quite large as B2 is well below A2 .' (Becausecapital transfers involve investmentswith a long gestation lag, they were generallynot amongthe strategies consideredduring the 1980s, when the key issue was to determine whether protectingthe current income of the poor was feasible in light of the overriding concern with

maintaining-- or establishing-- efficiencyduring the crisis.8) The main objectiveof the OECD project was to determineto what extent the process of adjustment,which in itself included non-neutral policias, affectedthe distributionof income. In terms of figure 1, the issue is how much did A2 deviatefrom ON, that is, how much did strategiesthat includeda transfer redistributioncomponent (B2) deviatefrom ON and what was the cost of redistributivepolicies (how much beloX LA2 is LB2). In applying the frameworkin figure 1 to individualcountries, initial conditionsare important. For example, as the discussionbelow indicates,Indonesianand Malaysianauthoritiesplaced a relatively high weight on equity issues in their policy choices. In terms of figure 1, this impliesthat the intertemporalbase run scenariosin the modelingexercises would be less steep than the distribution-neutralpath A,N in figure 1. Also, to take a specific example, in the Indonesiancase, dift._ientcombinationsof government current and capital expendituresyield significantlydifferent short- and long-run effects on growth and incomedistribution.

(b) Adjustmentand distribu'sn in the short-run: initial conditions,structure, and rigidities

Since adjustmentmeant that developingcountrieshad to transfer more resources abroad to service their debt, a loss of purchasingpower in terms of tradables was inevitableand an increase in poverty was possible. Terms of trade losses also imply a loss in purchasingpower of income in terms of cradablesfor the economyas a whole. Therefore, to observe that some measureof the aggregateeconomywidereal wage fell during adjustmentshouid be no surprise. This is not to say, however, that all incomeswere affectedproportionatelyby adjustment. Indeed, certain distributional shifts are to be expectedduring adjusument. To illustratethe tradeoffsbetweenequity and efficiency,it was convenientto assume that primary incomeswere unaffectedby current transfers and, by implication,that adjustmentpolicies did not affect the distributionof income. While such a simplificationis useful for discussingpovertyreduction strategiesin economiesin long-run equilibrium,it is certainlynot appropriatefor studying how incomedistributionwas affected in the short run by the adjustmentpolicies that were designedto cope with the macroeconomiccrises of the 1980s.9 Figure 2 i!lustrat s the interactionbetweenstabilizationand structural adjustmentpolicies for two identical economies(sametastes, endowments,and technology)but with different initial conditions. The economies,labelledF for flexible and R for rigid, are decomposedinto a tradables sector T and a nontradablessector N. EconomyF is at a full-employmentefficient (undistorted) productionequilibriumP while economyR is insideits productionpossibilityfrontier at PRbecause of distortionsand unemployedresources. Both economiesreceive a net transfer (or renewablecapital inflow) from abroad and are therefore able to consumeat Cp and C, which are beyond their earned incomeby the amount of the (identical)transfer."0 Consider now an extemnalshock resulting in a

Figure 2 Expenditure switching, expenditure reduction, a;ndstructural adJusment T Tradables C \

0

/'

C 0 (eo) ,,.. -

~~I cc

income-consumption curves

( el)

Non tradables

N

10

higher interest payment on the foreign debt such that both economiescan no longer consumebeyond their productionpossibilityfrontier. Consider first economyF. If the relative price of nontradablesremainedunchanged, consumerswould wish to consumeat C' which is on the same income-consumptioncurve correspondingto the initial relative price of tradablespO. However, at this set of relative prices, there is excess supply of nontradables(or demand for tradables). If relative prices are flexible, the economywill adjust and move to E. With full price flexibility,there is only a primary cost of adjvstmentwhich involvesbringing real expe litures in line with real income. However, if there is some relative price rigidity in the system and the real exchange rate cannot adjust, then there will be a secondaryadjustmentcost. The line EF traces the locus of intersectionpoints between incomebudget lines and income-consumptioncurves (not drawn), each correspondingto a differentvalue of the real exchangerate. Thus, dependingon the extent of relative price rigidity, the economywill adjust somewherealongEF. In the extreme case of no relative price flexibility (as would occur, for example, in a fully indexed economywhere a devaluationof the exchangerate would be shortly followedby a proportionalincrease in prices and wages), the economywould end up at F, where the externalconstraint is met."t Consider now economyR, with initiallyunemployedresources. If economyR can pursue structural adjustmentpolicies at the same time as it brings expendituresin line with income, then it can achievethe same equilibriumE, since by assumptionboth economiesare identicalin endowments and tastes. For economyR, however, becauseof reforms that improve the allocationof resources, the cost of adjustmentis less than in F and, in the extremecase, adjustmentmight even be welfare increasing. Of course, in the more likely event of price rigidity, economyR will end up further inside its production-possibilityfrontier. In the case of no relative price flexibility, the economywill end up at point R and will therefore, also experiencea secondarycost of adjustment.

11

It is clear from this discussion, and from Meller's case study of Chile (presented in this issue), that adjustmentis likely to have involvedsecondaryadjustmentcosts beyond the income loss and povertyincrease resultingfrom the need to service a higher externaldebt. It is precisely in this contextthat there is a role for governmentpolicy to attemptto minimizethe secondarycosts of adjustmentby undertakingappropriatestructural adjustmentpolicies to minimizethe social costs of adjustmentby appropriatecomplementarymeasures. We have not yet discussedhow shiftingresourcesfrom nontradablesto tradables is likely to affectthe distributionof income. Supposethat the poor get all their incomefrom the supply of labor serviceswhile the rich get all their income from the return to capital. For the time being, we shall assume that both groups have identicalpreferences. We assumefurther that tradables are labor intensive and that capital is sector specific in the short run. Then, we know that the real incomesof the owners of the tradable-sectorcapital stock will improve, that of the owners of the capital stock in the nontradablesector will decline in absoluteterms, and that owners of labor serviceswill fall in between.'2 In the long-run, we know from the Stolper-Samuelsontheorem that, under the usual competitiveassumptions,an adjustmentthat involvesan increase in the relative price of tradableswill shift the distributionof income in favor of labor becauseof the rise in the wage-rentalratio that accompaniesthe resource shift toward tradables. But the extent of the income redistributiondepends on the economy's structure. Figure 3 contrasts the same two economiesthat have to adjust to a reductionof capital inflows (which is equivalentto an increase in transfers abroad). Consideragain the long-run case, where all factors are mobile across sectors. In the right-handpanel, the curves labelled NN show the locus of points where there is internalbalance, that is equilibriumin the nontradablegoods market. Startingfrom a positionof equilibrium,where income Y is equal to absorptionA (both shown on the horizontalaxis), an increase in expenditureswill require an increase in the relative price of nontradablegoods - a fall

Figure 3 Adjustment and the Factoral Distribution of Income rT N eP /P

FR WR

WF

FR

NR

FF

..

.

..

.

. . .

. .

.

. .

. .

. .

.

.... .

.

.

....

.

.

\NR F WFWO WF~~~~~~~~w = YW/YK W?

*

Yl A

R~~~~~~~~~~~ 4

w

Y\w'YK

4

_

_____

_____

___

_

_____

_____

_____

_____

____

Y, A

13 in the value of the real exchangerate - to eliminateexcess demandfor nontradablegoods. Similarly, the FF schedulesshow the equilibriumin the tradable goodsmarket. Again, starting from an equilibriumposition, an increase in absorptionwil! create excess demand for tradable goods which will be eliminatedby an increasein the relative price of tradable goods. The left-handpanel depicts the relationshipbetweenthe wage-rentalratio and the relativeprice of tradable goodsunder the 3 A move away from the origin implies a assumptionthat tradables are labor intensive."

redistributionof incometoward labor.14 Return now to the comparisonof the two economies. Both have to adjust to the same reduction in capital inflows (i.e. the same leftwardshift in the FF curve measuredat the initial real exchangerate). Economy F is flexiblecomparedto economyR; small shifts in relative prices will lead consumptionpatterns to shift betweentradables and nontradables. Likewise,small relative price shifts will elicit a supply responsebecause the technology is flexible. By contrast, economyR is rigid. All imports are essential intermediatesand capital goods that cannotbe produced domestically, so there is little scope for substitutionin consumptiontoward domesticallyproduced goods even with large relative price shifts. Likewisetiere are rigidities in the productionstructure that make it difficultto shift resources from nontradablesto tradables."5Consequently,the NN and FF schedules for economyF are relativelyflat while those for economyR are relativelysteep. Also, the factor price frontier for the flexibleeconomyis steeper than for the rigid economysince the cost structures 6 are different."

Consider now what happens to the distributionof incomewhen both economieshave to adjust to the same reduction in capital flows. In the flexibleeconomy,the required real depreciationis smal!and the correspondingchangein the distributionof income (W. to W,) is also small. Economy R, because of rigidities in consumptionpatternsand productionstructures,has to depreciate its real

14 exchange rate much more to achievethe sameadjustment. This sharper real depreciationmeans, in turn, a larger shift in the distributionof the income (W. to W'). So economicstructure matters. The same apparatuscould also be adaptedto show how another form of rigidity - factor price inflexibility- affects the short-run factoraldistributionof income resulting from adjustment. It has often been observed that certaingroups fiercelyresist any adjustmentthat requires a cut in their living standards. For example,labor in the industrialsector of many Latin American economiesis organized in powerful labor unions that resist any drop in real wages. Owners of capital also often resist any fall in profit margins. Where trade and industrialpolicies shieldthem sufficientlyfrom competitivepressures abroad and at home, they rely on markuppricing, which allows them to pass any cost increasesthrough to consumers.'" So far, we have assumedthat both groups have identicaltastes. Now supposethat we are looking at adjustmentin a net food-importingcountry (like Morocco) and that workershave a strong preference for tradables. As figure 4 shows, adjustmentthat entails a depreciationof the real exchangerate from eoto el will lower the real income of workers and raise that of capitalists. Real incomeshifts like this that are due to differencesin tastes are likely to be quite importantfor some low-incomegroups, whose expendituresare mostly on essentialgoods like food. For example, adjustmentin Chile had a significantredistributiveeffect because it changedthe purchasingpower of 8 real incomes.'

We have seenthat how adjustmentpoliciesaffectthe short-runfactoraldistributionof income of initialconditionsreflectedin the stateof from the supplysidedependson the combination in the economy,the extentof rigidityin productionanddemandstructures,andthe disequilibrium of wagerates andprices. of the economyreflectedin the determination institutionalcharacteristics groupsimplythat changesin Wehave alsoseenthat differencesin preferencesacrosssocioeconomic relativepricesresultingfromstabilizationandstructuralreformpolicieswillhave a differentialeffect

Figure 4 Real Exchangepate Depreciation, Tastes, and Income Distribution lradables WO

(worker

s)

0

K 1 (capitalists)

-\ X'K

-0.

OKo K0 0

taabK Non tradables

16 on real incomes. For these reasons, an economywidemodel with a sufficientlevel of disaggregation across sectors, factors of production,and householdswas developedand applied in the country studies of the OECD project. The simulationmodels, explainedin greater detail in the companionpaper by Bourguignon, de Melo, and Suwa (1991) in this symposium,are of the open-economyvariety described graphically above. The modelshave several sectors, distinguishedby their degree of tradablity, and incorporate input-outputlinkagesto better account for the net factoral intensityacross sectors. Several types of labor with partial mobility across sectors and different wage-determinationrules accommodatereal wage rigidity and labor market segmentation. Factor incomesare mappedinto households,which receive their incomesfrom several sources. Householdshave different expenditurepatterns to account for the effects of adjustmenton the purchasingpower of real incomes. The models also include a governmentsector with current and capital expendituresand a financialsector that integrates portfolio choiceby householdsand firms. These extensionsof the basic frameworkpresentedhere enablethe modelsto trace the short-to-mediumrun effects of adjustmentpolicies on income distributionacross households."'

3. THE CASE STUDIES: A DIVERSITYOF INITIALCONDITIONSAND SHOCKS

The country studies reveal a remarkablediversity in the patternsof adjustment. While all six countrieshad to retrench and adjust in the sense of bringingexpenditurein line with income, two of them - Indonesiaand Malaysia- continuedto make progresson the distributionaland poverty fronts, albeit at a slower pace than previously. And for two other countries in the group - CMted'Ivoire and Ecuador - the verdict on adjustmentwas not yet in at the end of the period covered in the studies. Beforepresenting a comparativeevaluationof the adjustmentexperienceacross this heterogeneous

17 group, we briefly describethe sample (see Table 1) in terms of the origins of the crisis, countryspecific characteristics,main adjustmentpolicies, and trends in poverty indicatorsduring adjustment (which is discussedin detail below in section4). Of course, the changes in the poverty and distributionalindicatorsdescribed in Table 1 are not necessarilyattributableto the adjustmentpolicies pursued. This will be alludedto in section4 and dealt with in greater detail in section 5, which reports on the results of counterfactualsimulations. Although externalshockswere by no meansthe only cause of the crisis, all countriesf

.

1

into the dependent-economyframeworkused in section 2. In our sample, two economies,Ecuador and Indonesia, are primarilynatural-resource-basedeconomies,with oil exports accountingfor well over two-thirds of foreign exchangerevenuesat the time of the crisis. And three economiesEcuador, Cote d'Ivoire, and to a lesser extent, Morocco- can be characterizedas having relatively rigid demand and supply structures.' While falling terms of trade and rising interest rates were the main externalcauses of crisis in the countries in our sample, there were other interdependentcauses as well, as the country studies show. Except in Chile and Indonesia,fiscal deficits exceeded5 percent of GDP, in part becauseof the high interest paymentson the externaldebt. Perhaps the leading proximatecause of the crisis was the optimisticoutlookof all the countriesin our sample followingthe short-livedboom in commodity prices in the second half of the seventies(phosphatesfor Morocco in 1974; coffee and cocoa for COte d'Ivoire in 1976; rubber and tin for Malaysiain 1979; oil for Ecuador, Indonesia, and to a lesser extent, Malaysia in 1979; and a foreign capital inflow boom for Chile in 1980-81).21 At the same time as the commodityboom, the recyclingof petrodollarsin financial markets provided these countries with accessto private sources of finance at negativereal borrowing costs. Governments' spendingrose, made easier by the fact that the main source of foreign exchange earningswas usually controlledby public sector companiesor, as in the case of C6te d'Ivoire, by a

Tabl I COUy dlom CgMgry A4.u ImSdp

Yea Ofcri. ian acMain

CA dlvi

IPI

*

(391046)

Morocco IPI

0

(1910-87)

*

1043 Unsustainable publicinvetnt pm fowig abo ivd cofee ondcocoa boom(1975-77) Drught (1983)

19803 * Declinein le oftadc (phoshaO Uartalnbl publicexpendiu

Shoh, a_uOez,

Particuhr

*

0

0

Beo to CFA c zoe (convattbblecurey sn accessetoFrench t_ary for Iecal fwund) Cannotdcar mm mited fieedomin monetarypolicy (lod rinfall during adjusuncu (1989546) Incr iansttscca from abroad

04

0

*

* S

0

* * 1911#2 hicy-inducedea sxcbage rate arciation contributingto debt cri Deline in term of trade(copper)

0 0 0

IM43 OECDrecesaion Expansionaryfiscalpoficy(public ihnacra) Deverionatigtens of trade (rubber, tin, oil)

*

*

*

191142 Dcli in termsof trade (oil) Debt cri Furher crids in 1986with oil Prke fall Ecfcake in 1917

1ndonesa IFPI * (1979-81) *

19124-3 Decie in termsof trade (oil) OECDrecesion

*

* *

Maay su (XI 197347)

* * *

Ecudor IF! (197547)

0

0 *

Povety indicatortes duing afjudamers

djustmentpolicies

*

ChIle PC} (1979-88)

sd disturlbu nlindcaton

0

* 0

No ficaldeficit Distrtion-free conomy Copperpace incrascein 1937

0

No recour to MP. Rediibutko-with-growth strategydudrg 1970 Usd acces to foreign borrowingtill 1984 Prient rge fiscl deficit

*

Impodt-ubstitution led gowth in 1970s

*

* *

0

0

No rcoure to IMF Banced budget cottio y requied Co*nwied accessto externd fonds

Failedaucpa in 1978ad 1981 Adjumeat programin 1983 Monetry an fical expediture rmaraint Taxreform(introductionof VAT) Tradellrlization (QRand tarir reducion) Fancial sectorliberalization Devaluation

0 S

0

0 0 0

0

* * *

lbormakrc, dabaized

Socil pcndingcut Decema in urbtanual ncomega Povenydownin ral aes, up in urban ama

* * * *

Povety and ubsn phenometo Worncng incomedistribution Succeful rcgetins whilcsocial qcndis wa faling Reductionin iant moality to one of lowcst velain Latin Amcrica

Expenditurewitching runingin 1984 Cut in fical expecnditues

*

Ditrbutiol sift towardMalaya,whichstared under New EconomicPhanof 1975 Urb-a povety up slightly,ral povey down Duing adjustnt peiod (19 S17), ycedytcreass in expenditurcain educaion(5.7%)andhet (3.7%)

*

0

Strongi qualy (M of 0.52). egctd am region inequality Urbanworkerson ; agicui tamrs Rual wbsn inome ga narowcd

Devalurtioa(19S2) Talporary increaucin proeceion(now 15%tariff Copper stabilizationfund

0

0

0

Adjustmentprogramin 19SI Fiscalcuts: wags, subsidics;cait expendituxs Funher adjusmas progrm in 19445 (witbeffectsof devaluaionmimickedby cpe wubdi plus tari)

Adjustmentstaningio 19S4: cbical nco-liberalprogram Large devuation, mnovalof price contl, tade libernlization

Denvaution,withmoveto fill convetlibiity Inet rate liberalization Trade reform(OoweFing of tadffs, removl of most QRs) Tax rtform (ntroduin of VATand Incometax) Fiscalrestrint (cut in capital expenditures)

maauF:etingnl follaGPg couny nameo snrer toaesqofb eifleotnin used in theconyron reponRed iDtuble2: iXI Counr aufctuning in GDP 2 1311;fF1 Counitricawithbshdare of furelexportsin totalexports 2 50X; IPI Residualgrouping.

wt a re

* *

0

* *

Fruf

IAome ditrbuion worned in agdrcultuebut iopwovdin u0bn sector(shup fall in incomeof urbanric) During1980-47.incom per capta fell by 30% for urban and ural poor. by 38% for urWneducatedclan

Povertyfell from 337%in 1984to 20% in 1917 Povenyfel in both urbanand rual ; wiain maor empymp grous, inome distributionbecam more equal Cut in socidalnvmenls (schools,clinirs) ng in 1986

n exporam totl expotsu 20X

r a ae of

19 commoditystabilizationfund that reapedthe windfallprofits from high coffee and cocoa prices in world markets. Except in Indonesiaand Malaysia,the large-scalepublic investmentprojects undertakenduring this period were often foundto have low rates of return once the crisis raised substantiallythe opportunitycost of funds borrowedon commercialterms. The projects also proved difficultto scale down. Another commonpattern in the countriesin our samplewas a large appreciationof the real exchangerate in the years immediatelyprecedingthe crisis. (his is, of course, predictedby the model presented in section 2.) In Chile and Malaysia,where this appreciationwas particularly acute, the sense of euphoria was strong and led to a boom in private sector purchasesof imported consumer durables. Other countries also experiencedsome appreciationof the real exchangerate, as a portion of the increasedgovernmentexpendituresfell on nontradables. Despite these commonpatterns, however, the timing and severity of the crisis varied across the sample, and the need to adjust was determinedby several different factors. For example, the need to adjust came later for the oil-exportingeconomiesin our sample, as the price of oil tumbled in 1986. Chile benefittedfrom a sharp increase in the price of copper toward the end of its adjustment period. Another importantfactor during adjustmentwas the degree of accessto foreign financing. Adverse movementsin the terms of trade were not the only factor behind the crisis for the two Latin Americaneconomiesin our sample. For Ecuador, perhaps the most significantfactor during the adjustmentperiod was its loss of accessto foreign borrowingjust when it was needed most --

a decade of import-substitution-ledgrowthhad resulted in a relatively rigid economy. In Chile,

faulty macroeconomicmanagementwas largely to blame for the crisis. The combinationof a fixed exchangerate, wage indexationin the formal sector, and an open capitalaccount (and a ratwof inflationthree times that of its major trading partners)led to capital inflows and a sharp loss of competitivenesson external markets. Whenthe crisis came with the halt in private foreign capital

20 inflows, Chile, like Ecuador, had little accessto foreign credit to smooth the adjustment. Unlike Ecuador, however, Chile was in fiscal surpluswhen the crisis began in the fall of 1982. In the two Asian economies,the fall in the prices of oil and other primary commodities brought on the crisis. In both Indonesiaand Malaysia,expenditureswould eventuallyhave to be brought in line with incomes, and Malaysianeeded to prune a huge public sector deficit (over 20 percent of GDP). But both countriesadjustedfairly early, and both continuedto have access to private foreign credit and so did not require recourse to the IMF. These relativelyfavorable circumstancesare at least partly accountablefor the dramaticallybetter performanceof the two Asian economieson the distributionalfront during adjustment. In Morocco, as the price of phosphatesslumpedin 1976, the governmentwas unable to trim its investmentexpenditures,which had tripled during the phosphateboom that began in 1974, and stabilizationattemptsfailed in 1978and 1981. A halt to expansionistpolicies was finally brought about with the comprehensiveadjustmentprogramof 1983. Morocco, like Chile, Ecuador, and Cote d'Ivoire had a high externaldebt burden when adjustmentstarted. Turning next to the stabilizationand structural adjustmentpackages, we find that the sample countries introducedsimilar measuresto reduce aggregatedemand: cuts in public expendituresand devaluation(accompaniedby a more restrictivemonetary policy). Structural adjustmentinvolved deregulationof internalmarfretsand foreign trade liberalization. The cuts in public expendituresfell disproportionatelyon capital expendituresbecause, as explainedin the discussionof figure 1, current expenditurescan provide immediatepoverty relief whereas capital expendituresprovideonly future benefits. Adjustmentthrough cuts in capital expenditureswas also politicallyeasier since construction workershave less political power than bureaucrats,who are likely to resist to a cut in their salaries. Not surprisingly, public capital expendituresfell drastically- by over 50 percent in COted'Ivoire,

21

Ecuador, and Morocco. Indonesiaand Malaysia,with less need to adjust, more flexible economies, and better access to foreign financing, only had to cut public capital expendituresby about 10 percent. The other major cutback in public expenditureswas in subsidies,often on commoditiesthat were importantconsumptionitems for the poor. Chile, Morocco, and COted'Ivoire raised tariff rates on public sector services (e.g., energy and transport) in an effort to reduce the losses of public enterprises. But unless efficiencywas improvedat the same time, as was apparentlythe case in Chile, the distributiveeffect of these measuresis likely to have been regressive. The distributive effect of the reductionsin food subsidiesis more complexto evaluate,since it depends on supply and demand elasticitiesand on subsidy leakagesto the non-poor. For examnple,in Morocco only 16 percent of food subsidyexpendituresreachedthe poorest 30 percent of the population(Mateus, 1983). A reduction in food subsidiescould actuallyreduce inequalityin some cases when accompaniedby decontrol of food and other prices. By improvingthe incentivesfor food production, such policies can boost rural incomes, which are generallylower than urban incomes,thereby reducing inequality.' The centerpieceof the structural adjustmentprograms was a sharp devaluationof the exchangerate, which was intendedto reduce the macroeconomicimbalanceresulting from the large current account deficits and to move resourcestoward tradables. With the exceptionof CBted'lvoire, the countries in our sample managedto effect a real devaluationof the exchangerate and thereby to changeproductionincentivestoward tradables (see Table 2 below).' The other main componentsof the adjustmentpackage in countriesthat received IMF-World Bank support were structural reforms that aimed at deregulatinginternalmarketsand foreign trade. Price controls were removedto a large extent in Ecuador, Morocco, and COted'Ivoire. Trade liberalizationmeasures, includingthe removal of export taxes on traditionalexports, reduction or outright abolishmentof quantitativeimport restrictions, and a reduction in the level of protection,

22 were a particularly importantcomponentof adjustmentin the economieswith the most distorted system of incentives: Cote d'Ivoire, Ecuador, Morocco, and Indonesia. No structural reform measureswere needed in Chile, which had carried out structural adjustmentreforms during the second half of the 1970s,or in Malaysia,which was already open to foreign trade. Rationalizationof the fiscal system also figured prominentlyin the Moroccanand Indonesianreforms. Besides introducinga value addedtax, fiscal reform was directed at broadeningthe tax base by reducingthe numberof exemptionsand narrowingthe tax rates to reduce incentivesto avoid taxes or to arbitrage across tax categories.2A To get a sense of the representativenessof our sample, we comparedthe evolutionof key macro indicators by subperiodfor each countrywith correspondingaveragesfor appropriate subsamplesfrom a large sample of over eighty other developing countries(Table 2): primary exporters(CCted'Ivoire and Morocco), manufacturingexporters(Chile and Malaysia), and fuel exporters(Ecuador, Indonesia).' The indicatorsas GDP growth; the investmentshare in GDP, a measureof the sustainabilityof adjustment;the real exchangerate, here a measure of competitiveness; and external debt indicators. Period averagesare supposedto be representativeof the immediateprecrisis (1978-81)and post-crisis (1982-85)periods and of the end of the adjustmentperiod (1986-88), except for fuel exporterswhich had a second shock in 1986. While the period averages do not always correspondto the significantepisodesfor the six countries in our sample, the period averages do correspond broadlyto the periods when the main external environmentchangesof the 1980s occurred. The stylized facts that emerge from the period averagesfor the country groupingsare broadly consistentwith the analyticaldiscussionin section2. Three facts stand out. One is that only manufacturingexportershave resumedgrowth at pre-crisis levels (mostlythe East Asian countries). Althoughthe debt-serviceburden of this group is high (partly becauseof a few Latin American

23 countries in the grouping), it has stabilized. Growthamongthe fuel exportershas deteriorated throughoutthe three periods. Primary exportershave recuperatedmost of their loss in growth, but they have not arrested the deteriorationin their externaldebt indicators. These differencesin growth patterns are consistentwith the timing and size of shocks and with the greater ease of adjustmentfor manufacturingexporters becauseof their more flexibleeconomicstructures. The second significantfact is the universal, and pronounced,decline in the investmentshare in GDP. For the non-fuel-exportinggroups, the share has fallen by about 20 percent; for fuel exporters, the decline was an even sharper 30 percent. While it is true that overambitiousinvestment programs neededto be scaled down, thesedeclines are high enough,to cause concernabout the prospects for sustainedrecovery. They also suggestthe difficultyof cutting current expenditure during a crisis. The third significantfinding is the sharp real depreciationof the exchange rate. Six years into the crisis, the real exchange had depreciatedby close to 40 percent for all three country classifications. Without it, the required shift toward tradableactivities neededto increase the net transfer from debtor to creditor would not have materialized. Comparedto the averagesfor their correspondinggroups, the countries in this study were doing better than average in the pre-crisis periodbut -- except for Morocco - were harder hit by the crisis, experiencingrelativelylarger declines in growth rates during 1982-85. Recovery, however, was also stronger in our small sample of countries. Except for CBted'Ivoire and Morocco, which have high end-of-perioddebt indicators,the countries in our sample were showingstronger than average signsof sustained recovery, althoughadjustmentis far from over for two in our sampleof six countries.

Table2. Moeconnomic inditors dirningadjustment A comparisonwithother countries, period averagtsfor 197841, 198W5, and 198648

CountiylGNP per capita

GDP gowth

Inveatment/GDP

197841

1982-85

1986-88

1978-81

1982-85

Cte d'lvoire(S1,180t

3.9

0.2

0.5

28.0

Morocco (S930)

2.3

3.9

3.9

Primary expouten (45 countries)

2.8

1.4

2.4

1986-88

1978-8

1982-85

198688

1978I

16.8

12.9

1.05

1.23

1.10

23.7

22.2

20.3

1.00

1.15

21.4

18.4

17.2

1.03

1.13

_

Chile ($2,100) Malaysia($1,690) Manufacturingexporters (23 countris)

7.5

-1.5

7.6 4.8 MsnuGletunagcxpolten

6.3 4.6

Debt/GDP (debt-serviceraio, %)

Rcal exchange ae

19.8 30.3

12.1 34.1

1982-85

1986-88

0.47 (24)

1.01 (40)

1.14 (44)

1.25

0.50 (35)

0.93 (38)

1.10 (36)

1.39

0.35 (15)

0.53

0.61

(20)

(29)

0.37 (46)

0.83 (51)

0.93 (29)

0.23

0.49

0.65

_

16.2 25.7

1.01 0.95

1.12 0.86

1.61 1.01

4

~~~~~~~~~~~~~~~~~~~~(8) (15) (18)

4.8

2.4

4.9

26.2

23.5

21.8

1.03

1.12

1.39

0.21

0.35

0.40

Ecuador($1,260)

5.2

1.7

2.0

25.8

19.6

21.8

0.96

0 Jl

i.43

0.38 (35)

0.57 (40)

0.81 (33)

Indonesia($470)

7.3

2.9

4.0

26.1

27.2

24.1

0.92

1.00

1.52

0.26 (18)

0.30 (20)

0.58 (36)

Fuel expozters(15 countics)

6.6

2.0

0.9

27.9

24.9

19.4

1.0

0.95

1.35

0.34

0.44

0.63

(18)

(25)

(39)

_________

._______

Note:

AUldata are unweightedperiod averages. For definitionof country groupingssee tabk 1. Sampleof 83 countrieswith populationcxceeding I milion in 1980.

a. b.

Thc real xchage rate index is the ratio (expressedin common currency units) of a weightedsum of trading patemns' wholse index (1980 = 1.00). An increase in the value of the index indicatesa deprction (increasedcompetiiveness). In 1980 U.S. dol"an.

Sjou:

For comparatorrigures, Faini and de Melo (1990).

price indexesovcr the domesticconsumerprice

25 4. DISTRIBUTIONALOUTCOMESOF ADJUSTMENT: A SYNTHESIS

What were the distributionalconsequencesof adjustmentin the six countriesin our sample? We have already seen that, whilethe countriesshared commontargets during adjustment- closing the expenditure-incomegap that had developedbefore and during the crisis - initial conditionsand characteristicswere different. There are many ways in which we could compare results for our sample countries. We could compare the experienceof Indonesiaand Malaysia,the only two countriesthat adjustedwithoutrecourseto IMF and World Bank stabilizationand structural adjustmentloans. Or we could synthesizethe distributionalexperienceof our sampleby examining successivelyhow each countryperformed in terms of distributionalindicators: wages, primary incomes,poverty, and incomedistribution. What we have chosen to do, however, is to rely on three pairwisecomparisons: Chile and Malaysia, classifiedas manufacturingexportersbecause of the relativelylarger share of their industrial sectors and the role of manufacturedexports and the two most open economiesat the time of the crisis; C6te d'lvoire and Morocco, primary product exporters already on the edge of crisis even before the debt crisis (Moroccohad several failed stabilization experiencesbefore embarkingon adjustmentprograms);and Ecuador and Indonesia,the two oil exporters. Chile and Malaysia. The two economiesare clearly the most outward oriented in the sample and are certainlyamongthe developingcountrieswith the least governmentinterventionin goods and factor markets. They experiencedthe least prolongedshocks, and muchof their need to adjust was self-generated. As describedby Meller and by Demery and Demery in their countrystudies, both countries followedpolicies ihat led to a sharp real appreciationof the exchangerate. In while, the appreciationoccurred becausethe governmentused the exchangerate to bring down inflationwhile the capital market was open. Inflowsof short-termprivate capital fueled a private expendituresurge

26 that was accompaniedby appreciationof the real exchangerate. In Malaysia, a classic primary commodityboom in 1979 was naturallyfollowedby real exchangerate appreciation. But when the terms of trade movedagainst them, with a resulting 7 percent loss in the purchasingpower of income in 1981and another 4 percent loss in 1982, the exchangerate was not allowed to depreciateas was required. What both economiesneededto do was to bring expenditurein line with income: in Chile, becauseforeign capital inflowswere no longer availableto finance private sector consumption expenditures,and in Malaysia,because fiscal expansionwas no longer consistentwith the new terms of trade. Unlike while the other countriesin our sample, Chile and Malaysiadid not require structural adjustmentpolicies. Of course, matters were somewhatmore complicatedthan that. Chileanshad incurred large dollar-denominateddebts when they were cheap comparedto peso-denominatedloans. Then, as Meller points out, when the recessionand devaluationthrew the financialsystem into chaos (with bad loans accountingfor 350 percent of commercialbank equity), the governmentsubsidizedholders of dollar liabilitiesthrough debt-dedollarizationand preferential exchangerates. As Meller eloquently puts it: 'While 600,000 unemployedworkers were receiving 1.5 percent of GDP as unemployment subsidy, fewer than 2,000 dollar debtors were receivingsubsidiestotaling3 percent of GDP." Althoughslightly different from the often-mentioned"Latin Americancapital flight" syndromesince the increased indebtednesswas the result of a consumptionboom (mostlyof imported consumer durables)- the Chilean crisis shows clearlythat macroeconomicdisequilibriumhas the potentialto adverselyaffect the distributionof income. The role of flexibility, which was emphasizedin the discussionof adjustmentin Figure 2, is also highlightedby the contrastingexperienceof Chile and Malaysia. In Chile, adjustmentoccurred through the labor market, with a sharp increase in unemploymentexacerbatedby widespreadwage indexation- effectiveunemploymentreached 31 percent at the peak of the crisis.3' Thus, whereas

27 Malaysiaadjustedby stayingon, or close to, its productionpossibilityfrontier, Chile adjustedby moving inside it (see figure 2). During the first of the two post-crisisperiods (1982-5)shown in table 2, Chile is the only economywith negativegrowth. What were the distributionalconsequencesof this type of adjustment? Householdsurveys show that it was essentiallythe heads of householdsof the lowest income quintilethat bore the brunt of unemployment: in that group, 25 percent were unemployed,compared to 9 percent in the lower-middle-incomequintileand 2 percent in the highest-incomequintile.Also, unemploymenthurt the lower-incomegroupsthe most becausethey had fewer incomeearners per family. While householdsurveys were not availablefor the other counties, so we cannot draw comparisons,it is unlikely that this pattern of adjustmentwould have been observed in other countries where the crisis mostly led to underemploymentamongthe poor in the informal sector. Another interestingpiece of evidencefrom the Chilean adjustmentepisode is the impactof the real exchangerate depreciationon the purchasingpower of real incomes. As pointed out in figure 4, if householdshave consumptionpatterns that differ in their tradable goods intensity, a depreciationof the real exchangerate will have a redistributiveeffect through the consumptionside. The evidence (see Meller, table 6) suggeststhat this effect was quite significantbecauseof the large changein the value of the real exchangerate during adjustment: the middle 40 percent income group lost about 4 percent less real income through the real exchangerate devaluationthan the lower 40 percent income group. In contrast, Malaysiaadjustedwith very little secondaryadjustmentcosts during 1984-87private consumptionfell by 8.4 percent in real terms and investmentby about one third. But there were two fundamentaldifferencesin initial conditionscomparedto Chile: Malaysia's investmentrate was much higher and its initial indebtednesswas less (see Table 2). Both factors undoubtedlyhelped the Malaysiangovernmentto secure externalfunding at a time when other developingcountries were

28 being denied accessto foreign financing. Malaysiaadjustedmore through expenditureswitchingthan through expenditurereduction. Adjustmentwas not accompaniedby any significantdistributionalshifts in Malaysia. In fact, even though progress in reducing inequality- a hallmark of Malaysianpolicy since it launched its "growthwith redistribution"strategy in 1971- stoppedtemporarily, social expendituresgrew in real terms during adjustment. In sum, because Malaysiaentered into adjustmentfrom an economythat had been growingrapidly, it was relatively easy to reduce expendituresand even though investment had to be cut, high investmentlevels by internationalstandardswere maintainedduring adjustment. Another interestingcomparisonbetweenthe two countries relates to issues of political economyand the sustainabilityof adjustment. Chile was able to continuemuch of its difficult expenditurereductionthrough to the end in large part becauseof the political situation resultingfrom the military dictatorship. Malaysia, was able to carryout a sustainableadjustmentprogram under a democracywithout socialunrest, probablybecause equity considerationshad been a prominentsocial welfare concern for over a decade. This said, it should also be notedthat Chile, despite a worsening income distribution,manageda very successfultargeted programof assistanceduring adjustmentthat undeniablyhelped the very poorest. Also, infantmortality continuedto fall during adjustment, reaching one of the lowest levels (8 per 1,000) amongmiddle-incomecountries. CBted'Ivoire and Morocco. During the 1970s,CMted'Ivoire and Morocco, like manyother exportersof primary commodities,experienceda short-livedcommodityboom that they treated as though it were permanent. In CBted'Ivoire, the receipts of the commoditystabilizationfund went up tenfold with the coffee and cocoaboom. These funds, which were supposedto be put aside for coffee and cocoa producersduring periods of low prices, were spent on infrastructureconstruction projects - projectedwith long gestation lags and no prospects for earning foreign exchangeearnings. In Morocco, they used the windfallrevenue gains from the tripling of the phosphateprice in 1974to

29 finance a large public investmentprogram, to raise governmentsalaries, and to provide food subsidies. When commodityprices slumpedin 1976 and 1977, both countriesresorted to foreign borrowing. Not surprisingly, both countrieshad higher than average debt burden indicatorsin the period before the crisis (see 1977-81period averagesin Table 2) and a level of indebtednessthat left them with little room to maneuverwhen the debt crisis occurred in 1981-82. External crisis struck both countriesearlier than the other countries in our sample. Morocco also postponed adjustmentlongerthan any other country. Specialcircumstancesaccountedfor this postponement. Thanks to continuedaccess to externalborrowing, especiallyfrom Saudi Arabia, Morocco was able to postponeadjustment. COted'Ivoire, as a metnber of the CFA zone, had accessto the French treasury for extemal funds. Indeed, the relative indebtedness(vis-a-visgroup averages)of both countries increasedthroughoutthe adjustmentperiod.' In contrast to Chile and Malaysia,COted'Ivoire and Moroccohad relatively interventionist policies - price controls, high and dispersedtariff structurescomplementedby quantitative restrictions- and relatively weak fiscal administrations. In terms of Figure 2, both economieswere insidetheir respectiveproductionpossibilityfrontiersand in need of microeconomicstructural adjustmentmeasures. CBted'lvoire implementedsuch measuresin the adjustmentprograms of 198183 and 1984-86(see the article in this issue by Lambert, Schneider, and Suwa, Table 2 for a description);in Morocco, in 1983-85(see Morrisson, Table 1 for a description). Measures in both countrieswere aimed at closingthe fiscal expenditure-receiptgap and improvingthe efficiencyof resource allocationby rationalizingdomesticand foreigntrade tax structuresand by removingprice controls and quantitativerestrictionson imports. Morocco also relaxed interest rate controlsand tried to inducethe repatriationof workers' remittancesfrom abroadby providingpreferential interest rates. Morocco is the only countryin the sample whosegrowth rate rose during adjustment,while CBted'Ivoire is the only one that showedno sign of improvementuntil the end. Two factors account

30 for this difference in performancedespiterelativelysimilar adjustmentpackages. First, the weather favored Morocco (good rainfalls in 1985and 1986)and plaguedCBted'Ivoire (drought in 1983). How importantwere weather conditions? For Morocco, the agriculturalsector benefittedgreatly from the exceptionallygood rainfallsof 1985-86,althoughliberalizationof agriculturalmarkets also helped and the devaluationstimulateda growth in export crops. During those two years, agriculture is estimatedto have contributedabout half of the increase in economicgrowth. The growth in agricultural export crops raised the income of farmers since the supply increaseswere not generally accompaniedby a fall in real prices. Morrisson(see article in this issue) concludesthat primary incomes in agriculturefared well, and that this helped reduce income inequalitybetween rmraland urban incomes. For C6te d'Ivoire, simulationssuggestthat the drought reduced GDP growth by more than 8 percent after 1983, at the same time increasingthe agriculturalterms of trade by more than 10 percent. Second, Cote d'Ivoire had less room to maneuver becauseit could not explicitlyuse the exchangerate as a switchingdevice. Indeed, by the end of the period, the amount of expenditure switchingthrough real exchangerate devaluationwas lower in Cote d'Ivoire than in any other country in the sample (see Table 2 and comparatorfigures). Furthermore, becauseof its weak administrative capability,Cote d'Ivoire failed in its efforts to reduce the deficitvia tax increasesand to compensate for an inabilityto devalue its currencythrough an equivalentcommercialpolicy of uniform tariffs and export subsidies. In both Morocco and Cote d'lvoire, the brunt of adjustmentwas borne by city dwellers. The combinationof reforms and tavorableweatherin Moroccohelped reduce rural unemployment,but in the towns, employmentdid not grow fast enough to absorb new entrants in the labor force, and urban unemploymentrose 3 percentagepoints, reaching 15 percent, despite a respectable3.8 percent a year growth in employment. At the same time, becausethe supplyof skilled labor was growing much

31 faster than demand (due largely to educationpolicies of the previous decade), there was a sharp drop in the real wage of skilled labor. These factors,together with the removal of quantitativerestrictions whose quota rents had gone to wealthierMoroccans,helped to reduce income inequality. Much the sane pattern developedin Cote d'Ivoire with the brunt of adjustmentfalling on urban workers who lost their jobs in the private and parastatalsectors. If one assumes that the unemployedurban workers receivedsome support from those that kept their jobs, then the per capita incomesor urlian workers fell the most. On the other hand, governmentemployees(the better-off socioeconomicgroup) managedto retain their jobs althoughtheir real wages fell. In rural areas, the little evidenceavailablefrom householdsurveyssuggestsan increase in per capita expendituresfor non-food items. Lambert, Schneider,and Suwa also suggest that the inicreasein food prices relative to nonagriculturalproducts must have helped sustain rural income levels. T'husboth countries managedto reduce the rural-urbanincome gap during adjustment,largely by improvingagriculturalincentives. Adjustmentmeasurescontributedin two ways: devaluationof the real exchangerate, which stimulatedagriculturalexports, a particularly significanteffect in Morocco, and liberalizationof agriculturalmarkets - reductionof price controlson food, reduction of taxationof agricultural exports, the abolishmentof public monopoliesin fertilizers and agricultural exports in Morocco, and the commercializationof rice in Cote d'Ivoire.9 Unfortunately,these positive effectsof adjustmenton supply and on the rural-urban primary incomegap were not matchedby trends in social expenditures,especiallyin COted'Ivoire, which had to rely heavily on cuts in public expenditures. In C6te d'Ivoire, per capita expenditureson education fell, and primary school enrollmentgrew slower than primary school population. Not only social expenditures,but transfers to the poor were cut as well, includingsubsidieson items that constituteda significantproportionof the consumptionof the poor, such as rice, electricity, water. Moroccoalso

32 cut educationexpenditures,but the effects were alleviatedby rising school enrollments. And Morocco maintainedits food subsidies. Ecuador and Indonesia. Ecuador and Indonesiaare more a study in contraststhan in similaritiessince they have little in commonin their adjustmentexperienceexcept the need to adjust to a fall in oil revenues. Indonesiacarried out its adjustmentprogram withoutasking for support from internationalinstitutions,suggestingstrong internalpolitical support for adjustment.Y Like Malaysia,Indonesiaalso benefittedfrom a relativelylight debt burden when the crisis occurred (see Table 2), due largely to the constitutionalrequirementfor a balancedbudget. Its continued accessto foreign financinghelped cushionthe adjustment. Investmentin Indonesiawas al;o considerably stronger than the average for fuel exporters. By contrast, Ecuador had recourseto support from the intemationalorganizationsand in a broad sense did not "own" its adjustmentprogram, and so had difficultygetting the support neededto carry out its structural reforms. Nor did Ecuadofrhave access to externalfinancing when it was most needed. These problemsnotwithstanding,both Ecuador and Indonesiadid muchbetter than other fuel exporters(Table 2), resuminggrowth in the last period and cutting back less on investmentexpenditures. Indonesianot only achievedstabilizationbut also carried out structural reforms that helpedto increase credibilityof its adjustmentprogram. In particular, financialsector reforms leading to convertibilityof the rupiah inspiredconfidencein the expenditure-switchingand-reducingmeasures that were being implementedconcurrently. Much like Malaysia,Indonesiabegan its adjus,mentfrom a positionof high growth and high investment. What is remarkableis that while Indonesiaw;3 undergoingstructural adjustmentsthat reduced its dependenceon exports of nonrenewableresources for foreign exchange earningsand governmentrevenuesand improvedthe efficiencyof resource allocation and investments(see the article by Thorbecke in this issue), the country was also reducing poverty and undernutrition.Data from householdsurveyssuggest a much lower incidenceof poverty

33 in 1987 than in 1984 in both rural and urban areas, with the proportionof the populationbelow the povertyline falling from 33 percent in 1984to slightly over 20 percent in 1987. But the impactof favorable initial conditionsmust not be overlooked. Thorbeckepoints out that several measures already underway in the pre-adjustmentperiod helped Indonesiamaintainthe momentumof its povertyalleviationpoliciesduring the adjustmentperiod - investmentin rural infrastructure,the fertilizer subsidy, and other measuresthat contributedto the rice boom and the process of agricultural adjustment. It also appears that these improvementsdid not come at the expenseof capital expendituresin social servicesin the first years of the adjustmentprogram, althoughbeginning in 1986, capital expenditures(constructionof schools, hospitals, clinics and dispensaries)were cut drastically. In Ecuador, implementationof the stabilizationand adjustmentprogram was hampered by the strongly divergent interests of agroexporterson the coast and governmentbureaucrats in the capital.3 These conflictsled to half-heartedimplementationof difficultmeasuresto reduce the fiscal imbalanceand weakenedthe credibilityof the adjustmentprogram. Despite initial successesin 1984 and 1985, the austerity measurescould not be maintainedin 1986, when the price of oil fell again. To sustain the falteringalliance betweeneconomicelites and lower-classurban groups, the governmentturned to populismand regionalism,compromisingits austerity measureswith a large public works program. An earthquakehit the country in March 1987, damagingan oil pipelineand reducingoil output by about one-third. With no externalfunding available, it should come as no surprise that there was little internalsupport for the adjustmentmeasuresand that capital flight was extensive. Althoughnot much informationis availableon distributionaltrends in Ecuador, there is little doubt that poverty increased substantially;it is likely as well that income distributionworsened towards the end of the adjustmentperiod since capital flight benefittedthe upper classes. During

34 1981-85,as part of its fiscal austerity measures,the governmenthad to sharply reduce educationand health benefits. The share of value-addedaccruingto labor fell by over 8 percent per year during 1980-86. The decline in living standardswas particularlyacute in urban areas: unemployment(and underemployment)for Quito was estimatedat over 20 percent for 0987, and the real wage was 53 percent lower in 1989 than in 1980. Industrialemploymenthad to fall once adjustmentbecame inevitablebecauseprevious rapid employmentgrowth in the sector had been fueledby high levels of protection." Since data on householdincomeswere unavailablefor Ecuador, de Janvry, Sadouletand Fargeix estimatedhousehold-specificincomeequationsthat allowed them to predict changes in real per capita incomes. They concludethat the boom period had a progressiveeffect on income distributionin agricultureand a neutral effect in the urban sector and that the crisis had a regressive effect in both sectors. They also concludedthat the educatedurban populationlost the most in per capita terms during adjustment(about two-thirdsduring 1980-87). These estimates,however, do not take into accountthe potentialgains from asset protectionthrough capital flight.

5. LESSONSFROM THE COUNTERFACTUALSIMULATIONS

In this section, we draw lessonsfrom the model-basedexercisesin three areas: completing the picture where informationis lacking, especiallyon the magnitudeof real income loss in the informalsector; estimatingthe relative effects of alternativeadjustmentpolicies on poverty; and assessingthe sustainabilityof various adjustmentpackages.

35 (a) Completingthe picture

Except for Indonesiaand Malaysia, for which we have detailed householdsurveys that cover the period of adjustment,we have no systematicevidenceof the likely trends in poverty and income distributionduring the adjustmentperiod. A main advantageof the simulationmodels is that they providea comprehensiveperiod-by-periodaccountof the evolutionnf householdincomes that allows for inferencesabout the likely evolutionof incomedistributionand poverty indicators during 3 3 In addition, the simulationsare adjustment. useful for examiningthe importanceof shocks and of

the timing of adjustmentmeasures. We highlighta few results. In C6te d'Ivoire, the 1983drought occurred in the middle of adjustment.How importantwas the resulting shortfall in agricultural supply, and by how much did it increasepoverty? Lambert, Schneider, and Suwa simulatedthe economywideeffectsof the drought and found that it expandedthe numberof people in povertyby more than 15 percent. They also examinedthe benefitsof migration back to rural areas during the crisis for those in the informal sector. Governmentstend to postpone adjustmentuntil it can no longer be avoided. What are the effects of such delays? For Malaysia- the countrywith the most successfuladjustmentprogram Demery and Demeryshow that an earlier adjustmentwould have smeothedthe intertemporal distributionof income, which would have increasedwelfare if householdswere risk-averse. They also find that earlier adjustmentwouldhave led to a slight decrease in the numberof people in poverty, althoughthe intensityof povertywould have been slightlyhigher. By contrast, Morrisson finds that in Morocco, substantialgains wouldhave occurred from a policy of early adjustment, which would have reducedthe size of the transfers neededto stabilizethe incomeof the poor.3

36 (b) Packagedesign and poverty

How well did the adjustmentpackagesachievetheir goals of efficiencyand equity? To addressthis issue, most authors simulatedthe effects of alternativepolicies for reducingthe fiscal income-revenuegap (governmentwage freeze, tax increase, or a different mix of reductionsin current and capital expenditures),to see whetherthese alternativepolicies wouldhave significantly altered the outcome in terms of growth, incomedistribution, and poverty. We summarizethe main findings. In Morocco and Cote d'Ivoire, the need to cut the fiscal deficitwas imperative,and they tried to do so primarilyby reducingreal governmentsalaries and the growth of public sector employment. Cote d'Ivoire tried to reduce nominal salaries in 1990, but soon abandonedthe attempt. So the questionfor Morocco is whetherthis measurewas better than alternativemeasuresto reduce global demand. Morrisson shows that the worst sclution would have been to lay off public employees,since that creates unemployment,reduces growth, and increasespoverty. It turns out that a wage cut is the only measurethat reducesinequalitybecausepublic sector employeesare in the middle- and upperincomedeciles. The same outcome is also found for CBted'Ivoire. However, Lambert, Schneider and Suwa are careful to point out that the model-generatedresults do not recognizespillover effects that would dampenthe magnitudeof estimatedpoverty reduction (civil servants probably send a part of their earningsto relativesliving in rural areas, who are likely to be poor). While Morrissonshows that no other alternativeexpenditure-reductionpackage consideredis more favorableon distributional grounds than the reductionin civil servant wages, he finds that cutting operating expenditureshas a slightly better effect on poverty. Unlike a reduction in public sector wages, a cut in operating expendituresdoes not lead to a fall in demandfor informal sector goods and so, indirectly, to a fall in

37 the incomesof those engaged in the informalsector. Morrissontherefore concludesthat the Moroccan government chose the right policies.3 5

An examinationof the outcomesof a real exchangerate devaluationto switch expenditures and generate a supply-augmentingreallocationof resources also provides some interestinglessons. Interpretingthe results from this simulationis tricky, however. In most simulations, a devaluationof the exchangerate usually alters the country's net external indebtedness,which implies that one needs to take into account the effect of a devaluationon the economy's externaldebt position. For all countries, a devaluationprovides short-termbenefits by stimulatingexports and by avoidingthe more recessionaryimpactof the alternativepolicy of cutting public expenditu:es.' In the long run however, results differ across countries. lI Ecuador, the higher inflationgeneratedby the devaluationis assumedto directlyreduce investmentbecause of the increaseduncertainty. Growthis lower as well, and other policieshave a better effect on income distributionand poverty in the long run. In Indonesia,accelerateddevaluationresults in an accelerationof inflation and in capitalflight, but no loss of growth and mixed effects on income distributionsince all income groups, except large and medium-sizefarmers, lose. In Moroccodevaluationturns out to be the preferred instrumemin terms of social criteria, in the short run, in part because the drop in informal sector incomesresulting from the fall in real wages in the modern sector is compensatedfor by increased spendingon informal sector goods by farmers and agriculturalworkers, whose real incomesrise becauseof the devaluation. Morrissonpoints out, however, that a policy of successivedevaluations would quicklymeet with resistanceby modernsector wage earners. Furthermore, devaluation induces a fall in investmentbecauseof higher real interest rates. Devaluationalso loses its appeal in Morocco in the longer run because of lower growth. COted'Ivoire is the only countryin our sample that would benefit from a devaluationboth in the short run and in the long run. Ironically, it is also the only countrythat cannot use this option

38 becauseit belongsto the CFAfranczone. Thisfavorableoutcomeis duemostlyto the devaluation's growth-inducing effectas unemployment falls withhigherexportgrowthand externaldebt is reduced becauseof the improvementin the tradebalance. The treasuryalso benefitsfrom the boostin revenuesfromthe exportstabilization fund(evenwiththe levy rate remainingconstant).37Poverty fallsbecauseunemployment is lower,and incomedistributionimproves.Incomeinequalityis reduced becauseof the increasein the costof livingfor urbanhouseholds,for whomimportsconstitutea large shareof consumption,andthe higherdemandfor the outputof the informalsector. Not surprisingly,moststudiesfoundthat devaluationalone,withouta cut in expenditure, equilibrium,evenwhenit wasfoundto have wouldnot havebeen sufficientto restoremacroeconomic short-runexpansionaryeffects. The sameconclusionwaspointedoutby the analyticaldiscussionin section2, whichsuggestedthe needfor joint muietary andfiscalpoliciesto containor reduce aggregatedemand. Indeed,whilemostauthorsconsideredthe effectsof devaluationin isolationin their simulations,whenthc-yjudgedthe adequacyof measuresadopted,they consideredpolicy packagesin whichdevaluationwasonly oneof severalcomponents.It is nonethelessinterestingto contrastthe diversityof resultsfroma policyof devaluati,. In general,a relativelyrobust conclusionfrom thesecomparisonsis thatdevaluationtendsto reducepovertyin rural areasif small farmersproduceexportcropsandto reduceinequalitybecausethe real incomesof the rural poor and the urbanpoor in the informalsectorfare betterthanthoseof the modernsectorworkers. (rhe incomesof the poor in the urban informalsectorare relativelylessaffectedby the devaluation becauseinformalsectordemanddoesnot fall muchas a resultof a devaluation.A) In IndonesiaandMalaysiawhereadjustmentwas by and largesuccessful,the issueof policy designconcernedmainlywhetherthe authoritiesusedsufficientfiscalrestraintor the propermixof expenditurereductionandwhetherincreasingtaxationwouldhave beenpreferable. For both countries,the authorsfind that the alternativepackageswouldnothave performedbetter thanthe

39 measures actuallyadopted. A larger cut in expendituresis deflationary,which is not desirable even though it improvesthe country's externaldebt position. For Indonesia,where highlydisaggregated data on govermnentexpendituresis available,Thorbecke concludesthat a shift in the mix of governmentspending toward public investmentprojects might have been marginallybetter since the rural and urban poor would benefit while the rural and urban rich would lose marginallyfrom such a shift. 9 In Malaysia, a more austere fiscal adjustmentpackage is also found to be less desirablethan the course actuallypursued. Demery and Demery also find that a policy of reducingthe deficit by raising commodityand corporatetaxes wouldbe regressive, since it leads to a drop in household incomesand consumptionand a rise in the incidenceof poverty, a result Morrisson also finds for increasedcommoditytaxes in Morocco. Morrisson finds for Morocco that a more efficientadjustmentoutcome couldhave been achievedthrough a public works program that created employmentat low wages for the young and unskilled, financedthrough a cut in governmentwages. While the logic and benefitsof such a program are clear and substantial,such a policy would probablyhave been difficultto implement because of resistancefrom public servants. A final issue examinedis the effect of stabilizationpolicies on the distributionof income through portfolio shifts and asset revaluation,a point also stressed in the paper in this issue by Bourguignon,de Melo, and Suwa on modelingthe effects of adjustment. In the Ecuador study, macroeconomicdisequilibriumthat leads to inflationand lower investmentalso leads to capital flight, a form of asset protectionavailable only to the rich. With capital flight comesthe need for a larger real exchangerate devaluation,which meansa lower real wage and more poverty. These undesirable distributionalconsequences,a reflectionof a lack of credibility in the adjustmentpackage, have been emphasizedin several discussionsof the adjustmentexperiencein Latin America (see, for example,

40 Diaz-Alejandro1985). Thus counterfactualsimulationsare helpful in illustratingthe distributional consequenceof poorly designedstabilizationpolicies.

(c) Sustainablepackages

All too often, it is forgottenthat stabilizationand adjustmentmeasuresfail because they do not take into accountthe resistance of those whose standardof living will fall as a result of the measures. Among our sample countries, lack of support for stabilizationand structural adjustment policiesthat would benefit agroexporterswas particularlyevident in Ecuador, especiallyafter the 1988presidentialelections,when the incumbent,representingthe interests of agroexporters,was ousted by the czndidaterepresentinga coalitionof urban interests. Drawingon the simulations reported by de lanvry, Sadouletand Fargeix, we illustratehow counterfactualsimulationscan help in selectingwhich policy amongavailable alternativesis likely to meet the least political resistance. Two characteristicsof the Ecuadorianexperienceare typical of the problemsfacing many countriesduring adjustment. First, efficientadjustmentimplieda resource shift toward agriculture, which evidenceon adjustmentin several countriesin our samplecorroborates(Morocco, Indonesia, Chile). Prima facie evidencesuggeststhat such an adjustmentpattern is favorable to the interestsof agriculturallandownersat the expenseof urban interests - unless there are compensatingmeasures such as food subsidies. Second, adjustmenttook place under a fledglingdemocraticregime with weak institutionsand tenuous political support, a combinationthat implies heighteneduncertainty about the outcome of difficult measuresand an interplayof political interestsand pressure groups. In this context, then, the issue is which adjustmentpackagesthat are desirableon economicefficiency groundsare also attractivein terms of their effects on the distributionof income. In other words,

41

what adjustmentpackagesare sustainablepolitically,that is, which packageswill not engender coalitionsthat can bring a halt to adjustmentor a reversal in policies? The simulationswith the Ecuador model show the most efficientpackage in the long run to be one of fiscal austerity achievedthrough reductionsin current expenditures. Packagesthat do not involvefiscal restraint generateinflation, which eventuallycrowds out private investment. However, the problem with this adjustmentpackage, which reduces rural poverty, is that it also reduces the package that welfare of all urban groups and is therefore not politicallyfeasible. An adjustmnent protects the poor and involvesno loss in real incomeduring the adjustmentrequires a transfer equal to 3.5 perccnt of GDP (for Morocco, the correspondingestimate is 1.5 percent of GDP). Not surprisingly, de Janvry, Sadoulet,and Fargeix show that the only politicallyfeasible option is an adjustmentpackage with a foreign aid componentbecause the next best alternative- taxing the rich would require too large a tax intake from the non-poorto be politicallyfeasible.'4 In the absenceof foreign aid, what are the supportingpolicies that can make a fiscal austerity program sustainablepolitically? Public campaignsof informationand persuasioncan help. Also, structural reforms that increasethe elasticityof supply (by raisingthe elasticityof substitution between capital and labor) both increase growth while reducingthe distributionalshift required to effect the resource transfer also improvepolitical sustainability(see Table 7 de Janvry, Sadoulet, and Fargeix in this issue). This is, of course, the aim of microeconomicreforms in structural adjustment packages: to raise the efficiencyof resource allocationwhile minimizingdistributionalshifts (recall the comparisonin Figure 3 between rigid and flexibleeconomies). Policiesthat shift public investmenttoward the agroexportsector are also beneficialbecausesuch investmentboosts growth in that sector, thereby reducing rural poverty, and indirectlyimproveconditionsof the urban poor, who have access to cheaper importssince less real exchangerate depreciationis needed. Interestingly,de Janvry, Sadoulet, and Fargeix point out that the Ecuadoriangovernmentthat represented the interests

42 of agroexportersfailed to pursue such a course - the share of governmentexpendituresdirected to agriculturecontinuedto fall after the electionof Cordero. The simulationsfor Ecuador, however, also point out how little room governmentshave to maneuverwhen they must undertake wide-rangingadjustmentpackages, a fact that is known intuitivelyby those who have to implementthe measuresbut often overlookedin academicdiscussions of the social costs of adjustment. What is lacking is an economywidepolitical-economyframework that clearly highlightsthe point that distributionalshifts are an unavoidablecomponentof adjustment packages and that a reduction in fiscal expenditures(when there is no accessto foreign funds to cushion the impact)is likely to be unfeasibleif economicpolicies are open to interest-grouplobbying. The other countrystudies did not addressthe issue of sustainabilityas explicitlyas the Ecuador study, althoughthe authors usuallyalluded to the political consequencesof alternative adjustmentpackages. For example, in comparingthe outcomesof the various counterfactual scenarioswith the adjustmentpackage adoptedby the Indonesiangovernment(the base run), Thorbeckeconcludedthat the adjustmentpackage selectedby the govermnentwas consistentwith most objectivesOaidout by the government). In particular, it sheltered the incomesof the civil servants in both the short and long run more than each alternativecounterfactualsimulationexcept one that called for an even greater level of governmentcurrent expendituresthan in the base run scenario. A scenario that emphasizespublic investmentmore than in the base run yields the most favorable policy outcomesin the long run in terms of higher GDP growth, lower inflation,and higher incomesfor most agriculturalhouseholdgroups, but these advantageshave to be weighedagainst significantlylower standards of living for urban and rural higher-incomegroups. Given the political power of civil servants, Thorbecke concludesthat this cost could not be borne by the government.

43 6. CONCLUSIONS

How the poor fared during the structuraladjustmentprograms of the 1980shas been - and continuesto be - the subjectof debate. Some argue that the poor suffered a great deal and that their fate dependedon adjustment. Others argue that the declinein living standards was not so severeor pervasive, and that the fate of the poor would have been much worse had adjustmentmeasures not been taken. The purpose of the OECD project was to bring evidenceto the debate by addressingtwo key issues: What happenedto the poor, and what might have happenedto them with adjustment packagesother than those adopted? We started with an analyticaldiscussionstressingthat adjustmentinvolved,among other things, the need to cope with an adverse externalshock involvinga permanentloss in the income of the poor. Under those circumstances,the marginalvalue of incometo the poor rises, justifying an income transfer to the poor. However, becausethe initial conditionscharacterizingmany economies on the eve of adjustmentimpliedthat they were operating well withintheir productionfrontier, the possibilityexisted of improvingthe situation of the poor by successfullyundertakinggrowthaugmentingstructural reforms. We concludedthat, ultimately,the issue remains an empiricalmatter that can be resolvedonly by recourse to careful analysisof countriesthat carried out adjustment programs. Becauseof the enormous difficultyof identifyingthe poor and of attributingchangesin their well-beingto policies or statesof nature, the OECD project relied on case studies, complementedby counterfactualsimulationanalysis. An interestingfeature of the six countrystudies in this symposiumis their diversity. ';wo of the countries, Indonesiaand Malaysia,managedto adjust without any apparent adverse impacton the poor, despite some cuts in social expendituresin Indonesia. This superior performanceon the distributionalfront was attributedto a numberof factors, includingrelativelyfavorable initial

44

conditions, early adjustment,good and crediblepolicies, and continuedaccess to externalfinance to smooth adjustment. Not surprisingly,the explorationof alternativeadjustmentpackages for these two countriesrevealedonly potentiallynegligibleimprovementson the distributionalfront. In Chile and Ecuador, unsustainablemacro policiesprior to adjustment(Chile) or during adjustment(Ecuador)contributedto a worseningdistributionof income despite otherwisesound structural adjustmentpolicies. The experienceof these two countriesunderscoresthe importanceof adopting crediblepolicies, since reversalsbenefit some segmentsof society at the expenseof others. In Ecuador, capital flight was distributionallyregressiveand in Chile, subsidiesto holders of dollar debt resulted in a significantredistributionof income from the rest of society to the beneficiariesof the preferential exchangerate for dollar liabilities. Finally, for Cote d'Ivoire and Morocco, urban poverty increasedduring adjustmentwhile improvementseems to have occurred in the distributionof incomeas the rural-urban incomegap was reduced either mostlyas a result of measuresadopted (Morocco)or mostly as a result of exogenous events (Cote d'Ivoire). Simulationsshowedthat, for these two countries, alternativeadjustment packageswould likely have yielded superior outcomes,particularlytrue for C8te d'Ivoire, where large imbalancesremained. The centerpieceof virtuallyall adjustmentprograms was a sharp devaluationin real terms. This devaluation,which was part of the expenditure-switchingpoliciesthat had to be carried out to reduce the externaldeficit, was also part of the stru.lral adjustmentpolicies aimed at improving resource allocationand moving the economiescloser to their production frontier. The evidencefrom the various simulationexercisessuggeststhat this policy instrumentwas beneficialsince in the short run it avoided the recessionaryimpactof adjustmentthrough fiscal expenditurecutting. The simulationsalso indicatedthat this policy usuallyhad favorableshort-run effects on the distributionof income. When the longer-runeffects of devaluationincludinginflationand higher real interest rates,

45

were taken into account however, the results were more diverse. For some cases, devaluation remained superior to the alternativeof contractionarymonetary and fiscal policies, while for others, the contrary was true. Interestingly,the analysisof alternativeadjustmentinstrumentsfor Cote d'Ivoire pointed out to the superiorityof devaluation,an optionnot availableto that country. These conflictingresults point to the need for caution in interpretingthe results of individualstudies, since authors were often forced to rely on mechanismsoperating during adjustmentwhosevalidity cannot be directly tested, or transferred to another setting. The comparativestudies also show that distributionalconflicts can arise during adjustment. Amongthose, the most prevalentis the conflictbetween agricultureand labor-intensivesectors in manufacturingon the one hand, and the bureaucratic and import-substitutingsectors on the other. In general, agriculturewas found to be shelteredduring adjustmentand to have benefittedfrom the liberalizationmeasuresand real exchangerate devaluationthat accompaniedadjustmentprograms. Becauseof this conflictof interests, adjustmentprograms must be carefullymanaged, especiallyin economieswith large bureaucraciesand import-substitutingsectors. This suggeststhat efficiency criteria may have to accommodatedistributionalconcernsto avoid excessivedistributionalconflicts. All too often, it is forgotten that packagesfail becausethey do not take into account the resistance of politicallypowerful groups to measuresthat reduce their standardof living. In Ecuador, this resistance was at the base of a change in administrationduring adjustment. Simulationssuggest that packages that are most efficient in the long run are not feasible politicallybecausethey involve too great a deterioration in the welfare of t ,e political elite in urban areas. In those cases, timely foreign aid can make the difference at relativelylow cost to the donor becauseof synergisticeffects through the inducedgrowth effects of foreignaid. At the same time, insofar as structural reforms do indeed improvethe flexibilityof the economy, foreign aid can make a difference by making adjustmentpackages that are desirable on economicgroundsmore feasible politically.

46 The disciplineimposedby general equilibriummodelingalso shows the narrow room for maneuveringwhen the three criteria of efficiency,welfare, and politicalfeasibility are taken into account. In this setting, all actionshave opportunitycosts beyond those that would be revealedby a narrowly focused economiccost-benefitcalculus. This exposesthe potentially fatal flaws of narrowly designed adjustmentprograms, be they efficiency-focusedor welfare-focused. By the sametoken, becausethe margin for maneuveris so small, unanticipatedadverse shockscan be devastatingfor an ongoing adjustmentprogram. Under those circumstances,foreign assistanceloans are crucial for managingthis tenuous balance, with foreigngrants necessarywhen the size of the shock is very large. Nonetheless,despite these difficulties,the case studies in this symposiumshow that adjustment programs can be successful. In the same vein, the researchpoints out to a great diversityof impacts of adjustment programs. Sharply different distributionaloutcomescan occur with identicalacjustmentpackages when institutionalcharacteristicsdiffer widely. Sharply different distributionaloutcomescan also emerge as a result of changes in the mix between current and capital expenditurecuts. This diversity suggeststhe need for careful package design - 'passe partout" adjustmentprograms will not do. Tailoringadjustmentprograms to take into accountthe economicand political environmentis essential for equity and for the sustainabilityof the program itself. Moreover, the case studies in this symposiumshow that there is no inherent conflictbetweenfiscal retrenchment,which implies a smaller state, and the balancingof the objectivesof efficiency, welfare, and political feasibility, which implies an active state. The debate on the impactof adjustmenton the poor is far from being settled. Yet it is likely that, as our analysisof incomedistributionand poverty during adjustmentexpands and as reliable household-leveldata become available, our understandingwill increase. We hope that the studies in

47 this symposiumshed new light on the debate and point at least a thin beam of light towards some of the areas where our ignorance is greatest.

48 NOTES

1. Until the crisis of the 1980s,the economywideapproachto incomedistribution analysisrelied heavily on the long-run objectivestreated in the influentialwork of Chenery, Robinson, and Syrquin (1987). 2. Cornia, Jolly, and Stewart(1987) advocatea combinationof expansionarymacro policies and targeted micro policies designedto increase equityand efficiency. In an IMF study, Heller et al. (1988)suggest a descriptiveapproachthat relies on a classificationof the poor by meaningful socioeconomicgroups so as to be able to speculateon how the poor fared during adjustment. 3. The countrystudies in this symposiumare synthesesof longer, single-volumecountrystudies publishedby the OECD. For a fuller discussionof each country's experience, the reader is referred to the volumes referenced in the foreword to this symposium. 4.

This sectiondraws on Bourguignon(1991).

5. As discussedbelow, for economiesinsidetheir productionpossibilityfrontier, adjustmentmay not conflictwith poverty alleviation,if adjustmentalso entails movingcloser to the frontier, as would occur, for example, with effectivestructural reforms that improveresource allocation. 6. This analysisneglectsthe political-economyreasonsthat make redistributivepolicies difficult to implement(e.g., interest group coalitionsthat oppose the policies). This issue is discussedin further detail in section5 below. 7.

One could, of course, redistributecurrent incomeat B2. This would lead to positionB above (not shown in the figure). The evidencesuggeststhat in economieswith well-functioningtax and admimistrativesystems (e.g. the Nordic Europeancountries), substantialcurrent income transfers can be sustainedat low distortionarycosts. In developingcountries,by and large, static and dynamic costs of current income transfers are likely to be large. B2

8. Evaluatingthe outcome of a capital transfer strategyis likely to be even more complex as it involvescarrying out a cost-benefitanalysisof investmentin human capital of the poor. Sincethe poor are not assumedto receive any current incomeunder this strategy, both the poor and non-poor lose from this strategy in period 1, and the economymoves to Cl. As depictedhere, the asset transfer strategy which leads to C2 , dominatesthe current transfer strategy, but is shown to be less productivethan investmentin the rest of the economybecauseCl and C2 lie on ISO-linesbelow Al and Az. However, once all considerationsare taken into accovant,includingthe difficultyof measuringthe economicefficiencybenefitsresultingfrom less social conflict, it is likely that if the capitaltransfer strategy is well implemented,the economycouldmove to C;, and the equity-efficiency conflictsuggestedby a capital transfer would then disappear. 9. For example, the frameworkoutlined in figure 1 would be useful for analyzingthe tradeoffs involvedin Malaysia's New EconomicPolicy (NEP) of 1975.

49 To simplifythe graphicalexposition,we have assumedthat both economiesface the same 10. initial real exchangerate, el. This is unlikely to be so in practice, though it couldoccur as a result of a combinationof product and factor maiket distortions. As drawn here, we have also assumedthat nontradablesare a luxury good. This, of course, need not be the case. To measure the primary and secondarycosts of adjustmentwould require drawing budget 11. lines with slope e. through F and CF. Measured in terms of tradables, the primary (secondary)costs would be measuredby the distance on the vertical axis betweenthe intersectionof the relevant budget lines with the vertical axis and the budgetline through pF. For more discussion, see Corden (1988). Labor real income will fall (rise) relative to traded (nontraded)goods prices. This 12. distributionalresult was first developedin Jones (1971). For an empirical supporton the Stolper-Samuelsonrelationshipbetweenfactor intensitiesand 13. the factoral distributionof incomefor a group of developingcountries, see Bourguignonand Morrisson (1989, chapter 2). In particular, they find statisticalsupport for the hypothesisthat opennessto foreign trade shifts the factoralterms of trade toward labor. This is the case illustratedin figures 3 and 4. For a more completediscussionof the propertiesof this model, known as the dependent14. economymodel, see for example,Dornbusch(1980). In terms of the modelsused in the country studies, the elasticityof substitutionbetween 15. domestic-and foreign-producedgoods is low, so there is little scope for import substitutionin response to a real exchangerate depreciation. 16.

The easiest way to visualizethis is to considerthe model in its dual from, as in Mussa (1978).

Resistanceto cuts in real wages would tend to rotate the factor-pricefrontier 17. counterclockwisewhile resistanceto cuts in profits would rotate it clockwise. However, one cannot ascertain a priori how the factor price frontier will behavebecause it would dependon the exact origin of the price rigidity. Furthermore, in practice, it remainsan open questionwhich group will have the upper-handwhen both engage ir resistance to cuts in living standards. The structuralist literature addresses this issue at length (see, e.g., Taylor 1987). 18.

See Table 6 in Meller's discussionin section4 below.

For a full descriptionof the model, see the article by Bourguignon,de Melo, and Suwa in this 19. issue. Given Indonesia's strongperformance during adjustment(see table 2), it is unlikely that it had 20. rigid demand and supplyistructures. Or, if that was the case prior to adjustment,structural reforms were successful in eliminatingbottlenecks. Chile would also fall in that category if it were not for the major reforms of the late 1970s,which transformedif into an economywith no institutionalrigidities except for wage indexationin the formal sector, which was abandonedin 1982 in the depth of the crisis. Chile, in spite of falling copperprices, also had a period of euphoria when it benefitedfrom 21. large capital inflowsin 198-81.

so

In Cote d'Ivoire and Morocco, the amountof food marketedby the poor is quite small, so 22. this effect is not likely to be very significant. Becauseof the fixed exchangerate, increasedcompetitivenesswas achievedby the involuntary 23. appreciationof the dollar and by the voluntarydepreciationof the French franc againstother EMS currencies in 1981, 1982 and 1983, and, later in the adjustment,by the combinationof a tariff surchargeand export subsidies. In Cote d'Ivoire, little increase in competitivenesswas achieved indirectly. For an assessmentof recent tax reforms, includingthose for Indonesiaand Morocco, see 24. Thirsk (1990). Definitionsof country groupingsare given in table 1. A fairly similar pattern of results 25. obtainswith groupingsdefined by income levels. Note, however, that all the data in table 2 are from the World Bank. While internallyconsistentto the extent possible, the period averages in table 2 might deviate slightlyfrom those that wouldbe obtained from the data in the country studies. See the article by Meller in this issue (table 4) for a decompositionof the primary and 26. secondarycosts of adjustmentin terms of tradableand nontradablesectors. It should be noted that Chile had widespreadwage indexationin the formal labor market which contributedto inflexibility. In fact, Morocco tried to stabilize attemptsin 1978ard 1980,but govermmentattemptsto cut 27. food subsidies, led to riots and to the abandonmentof the measures. See the article by Morrisson in this issue for further discussion. 28.

Both countriesare amonga group of 18 highly indebtedcountries.

In terms of figure 2, these supply-augmentingstructural reforms moved both countries closer 29. to their respectiveproductionpossibilityfrontiers. loans from the World Bank (starting in 1987, but EventuallyIndonesiareceived adjustmnent 30. by then, most adjustmentmeasureshad already been adopted. With the electionof Cordero in 1984, political power shiftedback toward the traditionalelite, 31. agroexportersand bankers Industrialemploymentfell at an average annual rate of 2 percent during 1981-85. See the 32. article by de Janvry, Sadoulet,and Fargeixin this issue (table 2). They note that fiscal also austerity reduced employmentopportunitiesfor skilled and unskilledlabor. It is particularly encouragingthat the base run of the Malaysiamodel tracked very closely the 33. observed changesin the distributionof incomeduring adjustment. See the article by Demery and Demery in this issue (section6). Morrisson simulatesearlier adjustmentwith a package that includesa freeze on the minim'zm 34. wage. While such a policy is likely to be more acceptablewhen applied early on, it would still probably meet with resistance.

51 35. Consideringthat a similar policy had to be abandonedwhen it was applied in C6te d'Ivoire in 1990, and in light of the discussionbelow on the politicalsustainabilityof adjustmentpackages, one cannotbut help but wonder whetherthis policycould be implementedin a less authoritarianregime than Morocco's. 36. We do not report on the devaluationexperimentsfor Malaysiasince devaluationis also accompaniedby less fiscal restraint. It turns out that this package has a slightly negativeeffect on the balanceof payments, probablybecausedevaluationis accompaniedby a less restrictive fiscal policy. The package also results in slightly highergrowth than in the base run. 37. It is interestmgto contrastthis outcomewith the more direct alternativeof raising export taxes. Lambert, Schneiderand Suwa show that raising export taxes generates significantindirect effects, which dampen the initial fiscal improvement. 38. The controversyon the effects of devaluationon output (and even more so on income distribution) is far from settled. For a recent survey of the theoreticaldebate see Lizondo and Montiel(1989). For empiricalresults, see Edwards (1989). It should therefore not be surprisingthat countrystudies find different results. 39. In the Ecuadoriancase, adjustmentthrough a cut in current expendituresis also found to be preferable to that through a cut in capital expenditureson distributionaland efficiencygrounds. This is becausea cut in capital expendituresreducesthe real incomesof unskilledworkers whereas a cut in current expendituresresults in a cut in living standardsfor skilled workers. See the article in this issue by de Janvry, Sadoulet, and Fargeix (figure 1). 40. Interestingly,the simulationsshow that th3 amnountof foreign aid required is only about twothirds of the required transfer becauseof the growth-inducingeffects of foreign aid (see de Janvry, Sadoulet,and Fargeix (sectionlSb). This result, which impliesthat timely aid can make the difference between a sustainableadjustmentpackage and one that fails, is often overlooked. In the Ecuadorian case, timelydebt relief did not follow the earthquakein March 1987.

52 REFERENCES

Bourguignon,F., "Optimalpovertyreduction, adjustmentand growth" World Bank Economic Revie, Vol. 5, No. 2 (May 1991),pp. 339-366. Bourguignon,F., J. de Melo, and A. Suwa, "Modellingthe effemsof adjustmenton income distribution," World Development(this issue). Bourguignon,F., and C. Morrisson, ExternalTrade and IncomeD:istribution(Paris: OECD DevelopmentCentre, 1989). Chenery, H., M. Ahluwalia,C. Bell, and R. Jolly, Redistributionwith Growth, (London: Oxford UniversityPress, 1974). Corden, W. M., "Macroeconomicadjustmentin developingcountries," World Bank Research Qbserve,Vol. 4, (January 1989),pp. 51-64. Cornia, A., R. Jolly, and F. Stewart, Adjustmentwith a Human Face: Protecting the Vulnerableand PromotingGrowth (New York: OxfordUniversityPress, 1987). Demery, L., and D. Demery, "Poverty and macroeconomicpolicy in Malaysia, 1979-87,"Wold1 Develonment(this issue). Diaz-Alejandro,C., 'Good-bye financialrepression, hello financial crash," Journal of Development Economics, Vol. 19 (September/October1985),pp. 1-24. Dornbusch,R., Open EconomyMacroeconomics,(New York: Basic Books, 1980). Edwards, S., Real ExchangeRates. Devaluationand Adjustment: ExchangeRate Policy in Dlevelping Countries(Cambridge: MIT Press, 1989). Faini, R. and J. de Melo, 'Adjustment, investment,and the real exchangerate in developing countries," EconomicPolic, Vol. 11 (1990), pp. 492-519. Heller, P., A. Bovenberg,T. Catsambas,K. Chu, and P. Shome, "The Implicationsof FundSupportedAdjustmentPrograms for Poverty: Experiencesin SelectedCountries," Occasional Paper No. 58 (Washington,D.C.: IMF, 1988). Janvry, A. de, E. Sadoulet,and A. Fargeix, "Politicallyfeasible and equitableadjustment: some alternatives," World DeveloIment(this issue). Jones, R., "A three-factormodel in theory, trade and history," in J. N. Bhagwatiet al. (Eds.), Trade, Balanceof PEayents and Growth: Essays in Honor of C. P. Kindleb&j;r (Amsterdam: North-Holland, 1971).

53 Lambert, S., H. Schneider, and A. Suwa, "Adjustmentand equity in CBted'Ivoire: 1980-86," World DeveloDment(this issue). Lizondo, S. and P. Montiel, "Contractionarydevaluation: An overview," IMIFStaff Pap, (1989), pp. 182-227.

Vol. 36,

Mateus, A., "Targettingfood subsidiesto the needy: The use of cost-benefitanalysis and institutional design," WBSP No. 617 (Washington,DC: World Bank, 1983). Morrisson, C., "Adjustment,incomesand poverty in Morocco," World Development(this issue). Mussa, M., "The two-sectormodel in terms of its dual: A geometricexposition,"Journal of InternationalEconomics, Vol. 9(4) (1979), pp. 513-26. Robinson, S., "Macroeconomics,financialvariables, and computablegeneral equilibriummodels," World Development(this issue). Taylor, L., Varieties of StabilizationExperiences(Cambridge: CambridgeUniversityPress, 1987). Thirsk, W., "Recent experiencewith tax reform in developingcountries,"Ricerche Economiche,Vol. 44 (1990), pp. 321-48. Trhorbecke,E., "Adjustment,growth and income distributionin Indonesia," World Development(this issue).

PolicyResearchWorkingPaperSeries Contact

forpaper

Author CherylW. Gray WPS800 The Legal Frameworkfor Private SectorDevelopmentin a Transitional RebeccaJ. Hanson MichaelA. Heller Economy: The Caseof Poland Peter lanachokov YoussefDjehane

November1991

CECSE 37188

WPS801 Unravelingthe Mysteriesof China's Arvind Panagariya ForeignTradeRegime:A Viewfrom Jiangsu Province

November1991

D. Ballantyne 37947

WPS802 Strengtheningthe Bank's Population StevenW. Sinding Work in the Nineties

November1991

0. Nadora 31091

WPS803 FinancialRegulation:Changingthe Rulesof the Game

MillardLong Dimitri Vittas

November1991

W. Pitayatonakarn 37666

WPS804 GlobalTrendsin RawMaterials Consumption

Boum-JongChoe

November1991

S. Lipscomb 33718

WPS805 Privatizationin the SovietUnion: The Beginningsof a Transition

Sergei Shatalov

November1991

CECSE 37188

WPS806 MeasuringCommercialBank Efficiency:Useand Misuse of Bank OperatingRatios

Dimitri Vittas

November1991

W. Pitayatonakarn 37666

WPS807 ModerateInflation

RudigerDornbusch StanleyFischer

November1991

S. Moussa 33490

WPS808 The NewTradeProtection:Price AnnHarrison Effectsof Antidumpingand CountervailingMeasuresin the United States

November1991

D. Ballantyne 37947

WPS809 Opennessand Growth:A lime Series,Cross-CountryAnalysisfor DevelopingCountries

Ann Harrison

November1991

WDROffice 31393

WPS810 Povertyand IncomeDistribution during Adjustment:Issuesand Evidencefrom the OECDProject

FrancoisBourguignon Jaimede Melo ChristianMorrisson

November1991

D. Ballantyne 37947

PolicyResearchWorkingPaperSeries

AuLthor

Da

31047

the DebtCrisis? WPS786 FiscalPolicyfor Managing Indonesia'sEnvironment

Contact for paper

SadiqAhmed

WPS787 PrivateInvestmentUnderMacroeco- KlausSchmidt-Hebbel TobiasMuller nomicAdjustmentin Morocco

October 1991

B. Prasertwaree 82477

October 1991

S. Jonnakuty 39074

WPS788 HowExpectationsAffect Reform Dynamicsin DevelopingCountries

FrancescoDaveri

October 1991

S. Jonnakuty 39074

WPS789 IntrahouseholdInequalityand the Theoryof Targeting

LawrencePaddad Ravi Kanbur

October 1991

J. Sweeney 31021

WPS790 Reformingand PrivatizingHungary's Esra Bennathan Jeffrey Gutman RoadHaulage LouisThompson

October 1991

B. Gregory 33744

Lant Pritchett WPS791 MeasuringRealExchangeRate Instabilityin DevelopingCountries: EmpiricalEvidenceand Implications

October 1991

K. Cabana 37947

Jan Svejnar KatherineTe;rell

October1991

B. Gregory 33744

Karen M. Brooks WPS793 Decollectivizationand the AgriculturalTranshtionin Easternand CentralEurope

October1991

C. Spooner 30464

WilliamEasterly RobertKing RossLevine SergioRebelo

October 1991

R. Martin 39065

WPS795 EconomicStagnation,Fixed Factors, WilliamEasterly and Policy Thresholds

October 1991

R. Martin 39065

October 1991

W. Pitayatonakarn 37666

October 1991

0. Nadora 31091

WPS792 ReducingLaborRedundancyin State-OwnedEnterprises

WPS794 How Do NationalPoliciesAffect Long-RunGrowth?A Research Agenda

WPS796 ExcessLiquidityand Monetary Overhangs

GerardCaprio,Jr. Patrick Honohan

RichardHeaver WPS797 Using Field Visitsto Improvethe Qualityof FamilyPlanning,Health, and NutritionPrograms:A Supervisor's Manual WPS798 Agriculture'sDeclinein Indonesia: Supplyor DemandDetermined?

Will Martin PeterG. Warr

October 1991

M. Sanchez 33731

WPS799 Growthin OpenEconomies

SergioRebelo

November1991

R. Martin 39065