("p ip 71POLICY RESEARCH WORKING PAPER 3 092

The Integrated Macroeconomic Modelfor Poverty Analysis

A Quantitative Macroeconomic Frameworkfor the Analysis of Poverty Reduction Strategies

Pierre-Richard Agenor

Alejandro Izquierdo

Hippolyte Fofack

The World Bank

World Bank Institute

Poverty Reduction and Economic Management Division

July 2003

Pub

lic D

iscl

osur

e A

utho

rized

Pub

lic D

iscl

osur

e A

utho

rized

Pub

lic D

iscl

osur

e A

utho

rized

Pub

lic D

iscl

osur

e A

utho

rized

I POLICY RESEARCH WORKING PAPER 3U09

Abstract

Agenor, Izquierdo, and Fofack present a dynamic, of public expenditure on supply and demand, and creditqu"-_.-.;.'_ --±oaI onomLUt WILIKCL 11LIpCLLItIIS. aNucrcn.ic ai siiniouiations ror a

analyzing the impact of adjustment policies and prototype low-income country highlight the importanceexogenous shocks on poverty and income distribution. of accounting for the various channels through whichThey emphasize the role of labor market segmentation, poverty alleviation programs and debt relief mayurban informal activities, the impact of the composition ultimately affect the poor.

This paper-a nrncduct nf the Pnverrv Reduction and Economic Management Division, World Bank Institute-is part ofa larger effort in the institute to understand the impact of adjustment policies on the poor. Copies of the paper are availablefree fromrI.a the o:rld Bank, 18i1 8 H StreetW, Washng ton, DIAA203. fleaseco,ntact MIaria Gos iengiao rooma----TA'

-, Aa,r -U I I fl O3IL5

tJ1 t VS ~ AJTJ L I tJJi1~IU, JiF?-OU

telephone 202-473-3363, fax 202-676-9810, email address mgosiengfiao@worldbank.org. Policy Research WorkingPapers are also posted on the Web at http:/fecon.woridbanK.org. Tne autnors may be contacred at pagenor@woridbank.orgor hfofack@worldbank.org. July 2003. (126 pages)

A JC]IOOL&O3~V*.. ..... J / KC PV W'lic; R:/ toU UUUHdevelopment issues. An objective of the series is to get the findings out quickly, even ifthe presentations are less than fully polished. Thepapers carry the names of the authors and should be cited accordingly. The findings. interpretatins; and conclusions PxnrassPd in thispaper are entirely those of the authors. They do not necessarily represent the view of the World Bank, its Executive Directors, or thecountries they represent.

Produced by Partnerships, Capacity Building, and Outreach

The Integrated Macroeconomic Modelfor Poverty Analysis

A Quantitative Macroeconomic Framework for theAnalysis of Poverty Reduction Strategies

Pierre-Richard Agenor, Alejandro Izquierdo,and Hippolyte Fofack*

The World BankWashington DC 20433

JEL Classification Numbers: C68, D58, 011

*We are grateful to Ishac Diwan, Bill Easterly, Christian Emini, Eduaxdo Haddad,Michael Grimm, Henning Jensen, Jeff Lewis, Luiz Pereira da Silva, participants at variousseminars, and most importantly Peter Montiel and Dominique van der Mensbrugghe, forhelpful comments and discussions. Nihal Bayraktar, Derek Chen, and Lodovico Pizzatiprovided excellent research support.

Contents

1 Introduction 4

2 Analytical Background 6

3 Structure of IMMPA 113.1 Production ............. ............... 12

3.1.1 Rural Production ....... . . . . . . . . . . . . . . 123.1.2 Urban Informal Production ..... . . . . . . . . . . 143.1.3 Production of public goods and services ..... . . . 143.1.4 Urban Formal Private Production ..... . . . . . . . 15

3.2 Wages, Employment, Migration, and Skills Acquisition . . . . 163.2.1 Rural Wages and Employment . . . . . . . . . . . . . . 173.2.2 Urban Unskilled Wages and Employment . . . . . . . . 213.2.3 Urban Skilled Wages and Employment ..... . . . . 25

3.3 Supply and Demand . . . . . . . . . . . . . . . . . . . . . . . 283.4 External lhade ........ . ............ .. . . . 303.5 Prices .......... . .................. . . 303.6 Profits and Income ....... .. ........ . . . . . . . 343.7 Savings, consumption, and Investment . . . . . . . . . . . . . 363.8 Financial Sector ........ . ........... . . . .. . 40

3.8.1 Households . . . . . . . . . . . . . . . . . . . . . . . . 403.8.2 Firms ......... .. . .. . .. . .. . .. . .. . 423.8.3 Commercial Banks ....... . . . . . . . . . . . . . 42

3.9 Public Sector ......... .. . .. . .. . .. . .. . .. . 453.9.1 Central Bank and the Balance of Payments .... . . . 453.9.2 Government ........ . . . .. . . . . . . .. . . . 46

4 Poverty and Income Distribution Indicators 49

5 Calibration 585.1 Initial Values ......... . . .. . .. . . .. . . .. . . . 59

5.1.1 Endogenous Variables ...... . . . . . . . . . . . . 595.1.2 Exogenous Variables ...... . . . . . . . . . . . . . 60

5.2 Parameter Values ........ . ........... . .. . . 61

2

6 Some Illustrative Experiments 626.1 Terms-of-Trade Shock ........ .. . . . ......... . 626.2 Cut in Domestic Credit to Government ..... . . . . . . . . 666.3 Debt Reduction and Expenditure Reallocation ..... . . . . 68

6.3.1 lTansfers to Households ...... . . . . . . . . . . . 696.3.2 Investment in Infrastructure ...... . . . . . . . . . 716.3.3 Investment in Education . . . . . . . . . . . . . . . . . 72

7 Conclusions 74

References 78

Appendix A: List of Equations 85

Appendix B: Variable Names and Definitions 93

Appendix C: Parameter Values 100

3

1 Introduction

The current international debate on debt relief for highly-indebted, low-income countries has led policymakers, international development organiza-tions and academic economists alike to reaffirm the goal of sustained povertyreduction as one of the key objectives of adjustment programs.' At the sametime, renewed efforts have been made to amend existing macroeconomic pol-icy tools and develop new ones in order to better understand the channelsthrough which adjustment policies affect the poor and the possible trade-offs that poverty-reduction strategies may entail regarding the sequencingof policy reforms-particularly between short-term stabilization policies andstructural measures. This paper presents an integrated quantitative macro-economic framework developed recently at the World Bank for the specificpurpose of analyzing the impact of policy and exogenous shocks on incomedistribution, employment and poverty in low-income, highly-indebted coun-tries as well as middle-income developing economies.2 This new framework(baptized IMMPA, for Integrated Macroeconomic Model for Poverty Analy-sis) dwells on the extensive analytical and applied research conducted in aca-demic and policy circles over the past two decades on macroeconomic andstructural adjustment issues in developing economies. It captures a varietyof channels through which stabilization and structural adjustment policiesaffect growth, income inequality, and poverty in both the short and the longrun.

Our analysis, as discussed in detail below, differs from existing approachesin several dimensions. It emphasizes, most notably, the role of labor marketsegmentation (induced either by government legislation or firms' wage-settingdecisions), the role of informal employment in the transmission of policyand exogenous shocks to the poor, and (in the case of low-income, highly-indebted countries) the adverse effect of external debt on private incentivesto invest. It accounts at the same time for the impact of different compo-

'On December 22, 1999, the International Monetary Fund (IMF) and the World Bankendorsed the elaboration of a Poverty Reduction Strategy Paper (PRSP) as the centralmechanism for providing concessional lending to low-income countries. The declared ob-jective of the PRSP is to provide a medium-term framework to reduce poverty and gen-erate more rapid economic growth, with assistance from bilateral donors and multilateralfinancial institutions.

2We focus in this paper on a presentation of the prototype IMMPA model for low-income economies. As discussed in the concluding section, most of the building blocks ofthe model are also relevant for policy analysis in middle-income economies.

4

nents of government expenditure (on infrastructure, education, and health)on the production process and the accumulation of physical and human cap-ital by the private sector. It also captures credit market imperfections andlinkages between the financial system and the real side of the economy, byrelating firms' borrowing needs (for both working capital and physical capitalformation) to bank lending. Because of its integrated treatment of the realand financial sides, the model allows policy analysts to study not only theimpact of structural reforms (such as changes in tariffs or the compositionof public expenditure) on relative prices and output, but also the effect ofshort-term stabilization policies (such as a cut in domestic credit or a risein policy interest rates) as well as other financial shocks (such as an outflowof private capital or a rise in world interest rates). Our detailed treatmentof the labor market is particularly important to assess the poverty reductioneffects of adjustment programs, because the poor often generate their mainsource of income from wages, and the latter depend on available employmentopportunities. By distinguishing between rural and urban sectors (and ac-counting for migration d'ynamics driven by relative wages), the model allowsus to study separately the evolution of poverty in urban and rural areas andits relation with output and employment fluctuations across sectors.

The remainder of the paper is organized as follows. Section II presentsan overview of the main issues that our analysis focuses on and highlightsareas in which our framework differs from some of the past and more recentresearch in the field. Section III describes in detail the structure of theprototype IMMPA model for low-income, highly-indebted countries, con-sidering in turn production and employment, demand and external trade,prices, income and expenditure, the financial system, the public sector, thepoverty and income distribution indicators that the model generates, andthe link between consumption and income predictions and household expen-diture surveys. Section IV discusses issues associated with the calibrationand solution of the model and the extent to which our parameters and initialvalues can be deemed "representative" of a poor, highly-indebted economy.Section V presents the simulation results of various shocks (both exogenousand policy-induced) and discusses their real, financial, income distributionand poverty effects. We analyze, in turn, the short- and long-run impactof a temporary terms-of-trade shock, a permanent cut in domestic credit tothe government, and a poverty-reduction program consisting of partial ex-ternal debt forgiveness coupled with a reallocation of savings on debt servicepayments to three alternative forms of government expenditure: lump-sum

5

transfers to households, spending on infrastructure and outlays on education.This last experiment is of particular interest in light of the current interna-tional debate (alluded to earlier) on the need for debt relief for low-incomecountries-conditional on a productive use of associated savings. The lastsection summarizes our main results and suggests various extensions of ouranalysis.

2 Analytical Background

Development macroeconomists and policy analysts in general use a varietyof policy tools (ranging from single equations to full-blown econometric mod-els) to provide quantitative policy assessments. Within that range of tools,our framework is closer in spirit to dynamic, financial computable generalequilibrium (FCGE) models. However, our approach differs from the exist-ing literature in important ways. The point of departure of our analysis isthe need to construct a framework that allows analysts to provide appropri-ate answers to a variety of policy issues that developing countries (includinghighly-indebted, low-income ones) are currently facing in designing pro-poordevelopment strategies. For instance, how does resource reallocation inducedby structural policy shocks affect income distribution and poverty rates? Tomaximize the impact of debt relief on the poor, should governments increaselump-sum transfers to the poor, or invest in health, education, or infrastruc-ture capital? Are there trade-offs between stabilization and structural adjust-ment policies with respect to their impact on employment levels and poverty,and if so within what time frame? What are the implications of these poten-tial trade-offs for the sequencing of policy reforms? Our fundamental premiseis that these questions can be addressed in a meaningful manner only if thecomplexity of the labor market structure is properly represented, the link-ages between the financial and the real sides that condition the transmissionprocess of macroeconomic and structural policy shocks are well accountedfor, and if the channels through which government expenditure affects theeconomy (and eventually the poor) are adequately captured.

The first important distinguishing feature of our model is its detailedspecification of the labor market. Although it has long been recognized thatthe structure of the labor market can have a major impact on the transmis-sion of macroeconomic shocks and adjustment policies to economic activity,employment, and relative prices (see Ag6nor (1996, 2000, 2001), Bodart and

6

Le Dem (1996), and the World Bank (1995)), the treatment of this market inapplied policy models has focused on only a narrow set of its well-documentedfeatures-such as an economy-wide rigid minimum wage. Except for a fewexceptions (see for instance Maechler and Roland-Host (1995, 1997)), insuf-ficient attention has been paid to the macroeconomic implications of alter-native sources of labor market segmentation, differences in wage formationacross various labor categories, inter-sectoral wage rigidity (as opposed to ag-gregate wage rigidity), and feedback effects between relative prices and wagedecisions by price-setting firms. All of these features may have importantimplications for understanding the impact of policy and exogenous shocks onpoverty. Labor market segmentation, in particular, tends to restrict labormobility and can be associated with persistent wage differentials; these, inturn, may prevent the reallocation of resources necessary to cope with exter-nal and policy-induced shocks. Because the poor in many countries generatea significant fraction of their income from labor services, modeling these fea-tures of the labor market is crucial for understanding the impact of prettymuch any type of shocks on poverty in the short, medium, and long run.

Our analysis captures some of the important institutional characteristicsof the urban labor market observed in low- and middle-income developingcountries.3 In these countries it is important to distinguish, in analyzing thefunctioning of the urban labor market, between formal employment (suchas employment in large private enterprises and the public sector) and infor-mal employment. In many cases (particularly in sub-Saharan Africa) publicsector employment represents a very large share of formal sector employ-ment. Formal employment in many countries (in both Latin America andsub-Saharan Africa) has increased only slowly in recent decades, whereasmigration to urban areas has been extensive. Informal urban employmenthas thus increased dramatically in size.4 In Kenya, for instance, the shareof the informal sector in employment outside agriculture is currently about60 percent, whereas in Cameroon in 1993 it was estimated that 57 percentof the employed labor force worked in the informal sector (Fortin, Marceau

3See Agenor (1996) for a detailed overview of the literature, and Bigsten and Horton(1998) for a survey of labor markets in sub-Saharan Africa. See also the World Bank(1995).

4The informal sector can be defined in various ways; one common definition is that itincludes self-employed workers (except for professionals) unpaid family workers, workersemployed in small firms (less than, say, 5 or 6 workers), and those working in the tradeand services sector without a legally-binding contract.

7

and Savard (1997)). In Bangladesh, 90 percent of the labor force is in theinformal (nontraded) sector, mainly in rural areas (Devarajan, Ghanem, andThierfelder (1999)). In Ghana, between 1980 and 1990, employment in theformal sector declined significantly, despite a substantial increase (by almost50 percent) of the non-agricultural labor force. Available estimates suggestthat much of the expansion in the labor force was absorbed by the informalsector, whose size increased from 36 to 45 percent of the total (agriculturaland non-agricultural) labor force. The urban poor are also disproportionatelyemployed in the informal sector.

At the same time, wage formation and the composition of the labor forcetend to differ substantially between the formal and the informal segmentsof the labor market. Whereas workers in the informal economy tend tohave relatively low skill levels and face flexible wages, both low- and high-skilled workers are found in the formal sector. Particularly in the privateformal urban economy, the use of relatively capital-intensive technologiesimplies that skilled labor and physical capital tend to be (net) complements,whereas they both tend to be substitutes for unskilled labor. As a result ofboth the relative scarcity of highly educated labor and its complementarywith physical capital, wages for skilled workers tend to be high, relative toaverage wages in the economy. 5 In addition, the available evidence suggeststhat firms may pay higher wages to skilled workers for efficiency reasons-to reduce turnover costs (particularly if hiring, firing and training costs arehigh) to enhance productivity, attract better workers, and maintain loyalty

and morale. A large number of studies focusing on sub-Saharan Africa (seee.g. Bigsten and Horton (1998)) and Latin America have indeed found thatin the manufacturing industry in these countries there is often a significanteffect of firm size on wages, as implied by various efficiency wage theories.The phenomenon persists even after controlling for (observable) labor qualitydifferences and working conditions. This suggests that firms may pay moreeither to reduce costly labor turnover (larger firms are more capital intensiveand require more educated labor) or enhance the level of effort (which is lesseasily observable and monitorable in larger firms).

Our framework captures all of these features in a stylized manner. Wedistinguish between the formal and the informal components of the urbanlabor market. We take wages to be fully flexible in the informal economy

51n some countries, however, there is excess supply of skilled labor and a high rate of

open unemployment for that category of workers.

8

and assume that labor is heterogeneous in the formal sector, using a basicdistinction between skilled and unskilled labor. We account for both privateand public sector production and employment; and whereas wages of workersemployed in the public sector and those working as unskilled labor in theprivate sector are taken to be fixed by government fiat, we assume thatwages of skilled workers in the private sector are set on the basis of efficiencywage considerations. Although for tractability we use a specific efficiencywage formulation (based on the wage-productivity link), our formulation isobservationally equivalent to various other specifications.

The second important feature of our framework is the treatment of thefinancial system. In the "archetype" poor economy that we consider in thispaper, savers have access to only a limited range of financial assets (essen-tially money and bank deposits, held both at home and abroad) and commer-cial banks play a predominant role in the financial intermediation process.6

Specifically, we assume that firms are unable to issue tradable claims on theirassets or future output. We also dwell on some of the existing literature onFCGE models and the various channels that they have embedded: a portfoliostructure that accounts for the impact of interest rates on asset allocation de-cisions and capital flows (see for instance Robinson (1991), Adam and Bevan(1998), Rosensweig and Taylor (1990), Thissen (1999, 2000), and Thissen andLensink (2001)), real balance effects on expenditure (as in Easterly (1990)),and linkages between bank credit and the supply side through working cap-ital needs-as in Taylor (1983), Bourguignon, de Melo, and Suwa (1991),Bourguignon, Branson, and de Melo (1992), Lewis (1992, 1994), and Fargeixand Sadoulet (1994). For instance, we explicitly incorporate the bank lend-ing rate into the effective price of labor faced by firms that must finance theirworking capital requirements prior to the sale of output. This link turns outto be a crucial channel through which the real and financial sectors interact.But we depart from the specification adopted in many existing studies in twoimportant ways. First, we provide a full stock treatment of portfolio deci-sions (as opposed to the flow approach used in several studies) within thecontext of a Tobin-type asset demand system. Second, we capture the roleof balance sheet (or net worth) effects in the determination of bank lendingrates, in addition to funding costs, by explicitly modeling the "finance pre-mium" along the lines of recent research on credit market imperfections and

6 The IMMPA application to Brazil described by Agenor, Fernandes, Haddad, and vander Mensbrugghe (2003) provides a richer description of the financial system.

9

the importance of collateral, as discussed by Ag6nor and Aizenman (1998,1999b), Bernanke, Gertler, and Gilchrist (2000), Kiyotaki and Moore (1997),and Izquierdo (2000).

A third distinguishing feature of our approach is its emphasis on the neg-ative effect of external debt on private investment. Such a relationship canresult from various channels, including disincentive effects associated with alarge external debt. Indeed, several studies, evolving from Krugman (1988)and Sachs (1989), have argued that an excessive amount of debt raises ex-pectations of higher future (implicit or explicit) taxation and confiscationrisk to satisfy debt service payments. This, in turn, creates disincentives toinvest-most notably by lowering the expected after-tax rate of return oncapital (Hernandez-Cata (2000))-and may generate capital flight and lowernet private capital inflows.7 Studies by Elbadawi, Ndulu, and Ndungu (1997),Iyoha (2000), and Pattillo, Poirson, and Ricci (2002), provide some recentempirical evidence of a negative relationship between external debt and pri-vate capital formation for sub-Saharan Africa. lyoha (2000) for instance, ina study covering the period 1970-94, found that the ratio of external debt toGDP had a significant and negative effect on gross domestic investment. Pat-tillo, Poirson, and Ricci (2002), in a study covering 93 developing countriesover 1969-98, found that on average external debt begins to have a negativeimpact on growth when it reaches about 160-170 percent of exports or 35-40 percent of GDP. In the present model, we capture this adverse incentiveeffect by introducing a nonlinear effect of the ratio of debt service to taxeson private investment. As a result, the higher the initial level of debt, thestronger the negative marginal effect of a further increase in foreign borrow-ing on the propensity to invest. Put differently, the debt overhang in oursetting provides a justification for debt reduction on efficiency grounds.

Finally, our approach differs from existing studies by accounting explic-itly for the channels through which various types of public investment out-lays affect the economy. Economists and policymakers have long knownthat different forms of public investment can have different effects on out-put and employment, but the different channels through which alternativeforms of public spending operate have seldom been incorporated explicitlyin applied macroeconomic models used for development policy analysis. Inour framework, investment in infrastructure (or, rather, the stock of public

7As pointed out by Husain (1997), the taxation ability of the debtor government may beimportant in determining the magnitude of the debt overhang effect on private investment.

10

capital in infrastructure) affects directly the level of production in the pri-vate sector-and thus the marginal productivity of primary factors employedin that sector-whereas public investment in education has a direct impacton unskilled workers' decision to acquire skills (as in Jung and Thorbecke(2001)). This effect operates in addition, of course, to movements in relativewages across skill categories and the initial level of individual wealth, whichacts as a constraining factor in the presence of restrictions in accessing thecredit market. Our view is that this distinction is essential to understandinghow the savings created by debt relief should be allocated by governments (ordonor agencies) concerned with maximizing their poverty-reduction effects.

3 Structure of IMMPA

Given the type of policy issues that we wish to address, and the complexityof some of the channels highlighted above (regarding notably the structureof the labor market and the linkages between the real and financial sidesof the economy), our strategy has been to specify our framework in a rel-atively parsimonious way by restricting its various building blocks to whatwe believe to be essential components. In this section we review these var-ious building blocks and consider in turn the production side, employment,the demand side, external trade, sectoral and aggregate prices, income for-mation, the financial sector and asset allocation decisions, and the publicsector. We specify behavioral functions for each of the six broad categoriesof agents included in IMMPA: households (disaggregated by levels, of skillsand location into various types of rural and urban households), firms, thegovernment, the central bank, commercial banks, and the rest of the world.We also discuss the poverty and income distribution indicators that we use asa basis for analyzing the effects of our various simulation experiments on thepoor, as well as the links between the model's predictions and household ex-penditure surveys. Throughout the discussion, we often use "generic" formsto specify functional relationships; explicit functional forms (as well as vari-able names and definitions) are provided in Appendices A and B and theirexact specification in the computer simulation program are provided in Chenet al. (2001)). Finally, as indicated earlier, we focus in this paper on theprototype IMMPA model that we have designed with low-income countriesin mind, leaving for a companion paper (see Ag6nor, Fernandes, Haddad,and van der Mensbrugghe (2003)) the modified version that we have built

11

with some specific characteristics of middle-income developing countries.

3.1 Production

In many low-income countries, the majority of the poor lives in rural areas;it is therefore essential for a framework whose aim is to help policymakersdesign poverty-reduction programs to be able to trace differences in economicperformance in the rural and urban sectors. We therefore begin with a dis-tinction on the production side between the rural and urban sectors. Therural economy is itself divided between a tradable sector, which consists solelyof a homogeneous good sold abroad, and a nontraded goods sector, whichproduces a (composite) good sold only domestically (see Figure 1). The ra-tionale that underlies this distinction can be found in the available evidenceon the structure of the rural labor market, which suggests the existence ofan often large wage differential between workers employed in the productionof predominantly exported agricultural goods and workers producing. mainlyfor the domestic market-many of whom are involved in subsistence agri-culture (see for instance Bigsten and Horton (1998)). Because the povertyimplications of such differentials can be large, our strategy is to model themseparately.

Although income in the urban economy tends on average to be higherthan in the rural sector, the incidence of poverty has increased significantly(and sometimes dramatically) in urban areas in many developing countries.Concomitantly, the informal economy has grown in size in these countries,in part as a result of the lack of employment opportunities in the formalsector and the pervasiveness of labor market segmentation. We accountfor these characteristics by including both formal and informal componentswhen specifying urban production. Furthermore, we separate the formalurban economy into production of private goods (which are sold abroad anddomestically) and a public good, as discussed below.

3.1.1 Rural Production

The rural sector produces two goods; one traded, one nontraded. Landavailable for each of these activities is assumed to be in fixed supply andthere is no market to trade property claims on it. Gross output of nontradedgoods, XAN, and exported agricultural goods, XAT, are given by the sum of

12

value added (VAN and VAT, respectively) and intermediate consumption:

XAN = VAN + XAN aiAN, for i = AN, AT, I, P, G (1)

XAT = VAT + XAT aiAT, for i = AN, AT,I , P, G (2)

where the aij are conventionally-defined input-output coefficients (sales fromsector i to sector j) and AN, AT, I, P, G are used in what follows to refer,respectively, to the nontraded agricultural sector, the traded agriculturalsector, the informal sector, the private urban sector, and the public sector.

Value added in each sector is assumed to be produced with a Cobb-Douglas (CD) function of land, LAN, and a composite factor, defined asa constant elasticity of substitution (CES) function that depends on thenumber of unskilled rural workers employed (UAN in the nontraded-goodsector and UAT in the traded-good sector) and the economy-wide stock ofpublic physical capital (KG, which is defined below):

*VAN = CD[LANAN, CES(UAN, KG)], (3)

VAT = CD[LANAT, CES(UAT, KG)]. (4)

As noted above, we assume that land is a fixed input in each sector; andfor simplicity, we normalize the area of land allocated to production to unityin what follows. Given the CD specification, agricultural production exhibitsdecreasing returns to scale in the remaining (composite) input.

The introduction of KG in the production functions (3) and (4) is basedon the view that (cumulative) public investment in the economy improvesthe productivity of private firms and other production units in agriculture,because it facilitates not only trade and domestic commerce but also theproduction process itself. Thus, our concept of public capital includes notonly roads and public transportation that may increase access to markets,but also power plants and similar public goods that may contribute to anincrease in productivity.

As discussed below, the private, formal urban sector produces a singletradable good that may be either sold domestically or abroad. The struc-ture of the rural economy by contrast is rather different: we assume thatproduction of the rural traded good (XAT) is exclusively allocated to ex-ports (see below). In contrast, the nontraded agricultural good is exclusively

13

produced for the domestic market (XAN = DAN). The reason for choosingthis approach is that in many poor countries, there is often a sharp contrastbetween the "external" component of the agricultural sector, where produc-tion is mostly targeted to exports, and the "domestic" component, whereproduction is targeted mainly toward domestic consumers. The productionprocess, access to bank credit, and wage formation mechanisms often differsignificantly across these components, making the assumption of a smoothproduction possibility frontier (with a conventional concave shape) inappro-priate. These modeling choices are obviously important for the interpretationof some of the simulation results discussed below.

3.1.2 Urban Informal Production

Gross production in the urban informal sector, XI, is given as the sum ofvalue added, VI, and intermediate consumption, with value added given as afunction of the number of unskilled workers employed in the informal econ-omy, Ul, with decreasing returns to scale:

XI = VI + XI Eai, for i = AN, AT, I, P, G (5)

whereVI = aex, Upl, cxj > °, ° < ,l3x < 1. (6)

Moreover, given what the available evidence suggests regarding the degreeof labor intensity of production in the informal sector, we take 3x, to berelatively high in our policy experiments.

3.1.3 Production of public goods and services

Gross production of public goods and services (or public good, for short),denoted XG, is given by the sum of value added, VG, and intermediate con-sumption. Production of value added requires both types of labor (skilledand unskilled) and is given by a two-level CES function. At the lower levelunskilled workers, UG, and skilled workers, SG, combine to produce "effec-tive" employment in the public sector; and at the second level, effective laborand public capital, KG combine to produce net output:

XG = VC + XG aiG, for i = AN, AT, I, P, G, (7)

14

whereVG = CES[CES(SG,UG),KG]. (8)

Employment levels of both categories of workers are treated as predeter-mined policy variables. This specification of the production process of thepublic sector good makes it possible to analyze the effects of changes in gov-ernment employment. Public sector layoffs, for instance, will influence therest of the economy both through quantity effects on the labor market aswell as a reduction in public sector output.

Note also that, as indicated below, public sector wages for skilled workersare set equal to the efficiency wage paid to that category of workers in theprivate formal sector. Wages for unskilled workers in the public sector are,by contrast, set equal to the minimum wage; unskilled workers are therefore"off" their supply of labor. We will assume that "excess" employment (ordisguised unemployment) of unskilled labor prevails in the public sector (sothat output is never constrained), as documented in some studies of the labormarket in developing countries.8

3.1.4 Urban Formal Private Production

As with the production of the public good, private formal production alsoemploys skilled and unskilled labor. In addition, physical capital is alsoincluded as an input in the production process. An important step in spec-ifying the production technology consists of defining the degree of substi-tutability among inputs. We assume-as suggested by some of the empiricalevidence-that skilled labor and private physical capital have a higher degreeof complementarity (lower degree of substitution) than physical capital andunskilled workers. In order to account explicitly for these differences in thedegree of substitutability among inputs, we adopt a nested CES productionstructure. Specifically, gross production of the private formal-urban sector,Xp, is taken to be given by the sum of value added, Vp, and intermediateconsumption:

Xp =Vp+XPEaip, for i = AN, AT, I, P, G, (9)

whereVP = CES{CES[CES(ef.Sp, Kp), Up], KG}, (10)

8See Agenor (1996), who discusses the role of political factors in the determination ofpublic sector employment.

15

and ef, as discussed below, denotes the level of on-the-job effort providedby skilled workers. At the lowest level of equation (10), the effective supplyof skilled labor, ef.Sp, and private capital, KP, are used in the productionof the composite input T2 (assuming a low elasticity of substitution betweenthem). At the second level, this composite input is used, together with un-skilled labor, Up, to produce the composite input T1. With this specificationit is possible to choose a higher elasticity of substitution between T2, andunskilled workers, Up. In other words, whereas skilled labor and privatephysical capital are essentially complements, either one of these inputs is asubstitute to unskilled labor in the production process. The final layer ofthis nested CES has T1 and KG (the stock of government capital) as inputsin the production of private formal urban output.

Given our assumption that both the public and informal goods are nottraded (that is, they are sold only on the domestic market), the private formalurban good is the only component of urban production that can be exportedabroad. In line with conventional CGE models (see, for instance, Dervis, DeMelo, and Robinson (1982), Devarajan et al. (1997), and Robinson et al.(1999)), we assume that firms choose to allocate their output to exports orto the domestic market according to a production possibility frontier, definedby a Constant Elasticity of Transformation (CET) function:

XP = CET(Ep,DP) (11)

Equation (11) states that output, Xp, produced according to equation(9), is allocated either to exports, EP, or the domestic market, Dp. Theactual levels of each component are shown later to depend on the relativeprices of exported and domestic goods, in standard fashion.

3.2 Wages, Employment, Migration, and Skills Acqui-sition

As mentioned earlier, there are two types of workers in this model: skilledand unskilled. In practice, of course, there is a continuum of skills, but onemay want to think of our distinction as a broad one that suffices for analyticalpurposes. Unskilled workers may be employed either in the rural economy,UR, or in the urban economy, Uu, whereas skilled workers are employed only

16

in the urban economy (see Figure 2).9 We also assume that skilled workers arenot employed in the informal economy either-perhaps as a result of signalingconsiderations, as discussed later. In practice, there may obviously be manycases in which (semi-) skilled workers take part in the informal labor force.However, their productivity should not differ much from that of unskilledworkers, at least in part because monitoring is relatively easier than in theformal economy (this being in turn related to the smaller size of productionunits in the informal sector). As a result, modeling them differently wouldnot substantially alter a broad range of simulation results.

3.2.1 Rural Wages and Employment

The available empirical evidence on wages in the rural economy of many low-income developing countries suggests that there is a significant discrepancybetween wages paid in the export sector (cash crops and other agriculturalproducts), and wages paid in the nontraded sector. Because of higher wagesin the export sector, all workers in rural areas will opt to seek employmentthere first. In general, nominal wages in the export sector, WAT, can betaken to be indexed either on the value added price or gross output price inthat sector, or on the economy-wide consumer price, so that

WAT = WAT(PINDAT)iAT, (12)

where PINDAT = PVAT, PXAT, or PLEV, and 0 < indAT < 1, with PVAT(PXAT) being the price of value added (gross output) in the export sector,and PLEV the consumer price index, which are defined later.

In the simulations reported below, we take the degree of indexation to beperfect with respect to the price of value added (as a result, say, of strongbargaining power of workers in that sector), so that PINDAT = PVAT, andindAT = 1. Thus, the product wage, WAT, is taken to be fixed in real termsand firms are assumed to hire workers up to the optimal level given by theirlabor demand curve. The profit-maximizing demand function for labor inthe export sector, UAT, can be shown to be given by

1

d _ +lPXAAJT -XAT /XAT 1+PXAT WAT

wTT\ PXAT'WAUAT= VAT 1 +ILA_1)iWAT aeXXAT hrAT / PVA

(13)9This is of course a simplification, as there are some firms, located in the rural area,

which may require skilled labor.

17

Equation (13) indicates that labor demand in the rural export sector ispositively related to the level of net output, VAT, and negatively related tothe effective product wage, (1 + IL_,)WAT. Labor demand does not dependdirectly on the price of the exported agricultural good, PXAT (which is pro-portional to the value added price, as shown below in equation (47)), becausethe nominal wage rate paid to workers, WAT, varies proportionally to PVAT

in order to keep the real wage rate fixed at WAT. Again, the assumptionof perfect real (product) wage rigidity is not necessarily appropriate for allcountries; there are cases where partial indexation of nominal wages mightbe more appropriate. The assumption of full indexation can be easily relaxedin the IMMPA software for practical applications.

Note also that the product wage rate is multiplied by (1 + IL-1 ), whereIL-, is the one-period lagged bank lending rate, to account for the fact thatfirms in this sector rely on working capital to pay wages in advance of thesale of output. As a result, firms consider the effective price of labor, whichincludes the cost of borrowing, when making labor hiring decisions.'0 Weassume that the cost of credit specified in loan contracts negotiated for thecurrent period is based on the interest rate prevailing in the previous period.

The supply of labor to the nontraded agricultural sector is determinedresidually, as the number of workers who are unable to find a job in thebetter-paying sector seek employment in the nontraded sector:

UAN = UR-UUAT. (14)

The wage rate in the nontraded agricultural sector is flexible and deter-mined so as to equate labor demand (derived from standard profit maximiza-tion conditions), UAdN, and labor supply, given by (14). The market-clearingequilibrium product wage, WAN, is thus given by

WA 3 A(l - 7A) (_V+1PxAN 8AN

WANV- oXAN1XAN VXAANV +PXANN where WAN = PV (15)aXAN WUAN) /IA

The size of the labor force in the rural sector, UR, is predetermined atany given point in time. Over time, UR grows at the exogenous population

'0 There is a large literature emphasizing the role of interest rates in a credit-in-advanceeconomy on the supply side; New Structuralist economists (see e.g. Taylor (1983)) havebeen particularly strong advocates of the necessity to take this effect into account whenspecifying macroeconomic models. See also Izquierdo (2000) for a more recent discussion.

18

growth rate, net of worker migration to urban areas, MIGR:

UR = UR, 1(1 + 9R) - MIGR. (16)

In line with the traditional analysis of Harris and Todaro (1970), theincentives to migrate are taken to depend negatively on the ratio of theaverage expected consumption wage in rural areas to that prevailing in urbanareas. We assume that costs associated with migration or other frictions maydelay the migration process, introducing persistence in migration flows.

Unskilled workers in the urban economy may be employed either in theformal sector, in which case they are paid a minimum wage, WM, or theycan enter the informal economy and receive the market-determined wage inthat sector, Wl. When rural workers make the decision to migrate to urbanareas, they are uncertain as to which type of job they will be able to get, andtherefore weigh wages in each sector by the probability of finding a job inthat sector. They also take into account expected wages in the rural sectorwhen making migration decisions. Suppose, for simplicity, that the rural la-bor market operates as a sequential "auction" market-all rural workers arelaid-off at the end of each production period and re-hired randomly at thebeginning of the next. In such conditions, the probability of employment ineach sector subject to entry restrictions (that is, sectors where employmentis demand determined) can be approximated by prevailing employment ra-tios. Finally, potential migrants also consider what their expected purchasingpower in rural and urban areas will be, depending on whether they stay in therural sector and consume the "typical" basket of goods of rural households,or migrate and consume the "typical" urban basket of goods.

Combining these hypotheses implies that the average expected urban con-sumption wage is a weighted average of the minimum wage in the formalsector and the going wage in the informal sector, deflated by an urban con-sumption price index, PUU (defined below)."1 The weights are given by Ouand 1 - Ou, where Ou is the probability of finding a job in the urban formalsector. The expected, unskilled urban real wage for the current period, Ewu,is thus

Ewu = OUWM,p1 + (1 - Ou)W1 ,-1 (17)PUU,-1

"1 The weights of each type of good in the price index represent consumption patternsfor urban unskilled workers in the base period (see the discussion below).

19

where Ou is measured by the initial proportion of unskilled workers in theprivate formal sector, relative to the total number of unskilled urban workersnet of government employment prevailing in the previous period (that is,Ou = (Up,o/(Uu,o - UG,o)).' 2 A similar reasoning is used to calculate theexpected rural consumption real wage, EWA:

EWA = ORWAT,-1 + (1 - OR)WAN,-1

FR,-1

where PR is the consumption index for the rural sector (defined below) andOR is approximated by the initial proportion of the rural unskilled labor forceemployed in the export sector (that is, OR = UAT,o/UR,o).' 3 The migrationfunction can therefore be specified as

MIGR = UR,-lAm amIn ( Ewu)] + (1-Am)UR'' MIGR-1, (18)[1 (EWA AmUR, 2 MIR, (8

where 0 < Am < 1 measures the speed of adjustment and aM > 0 measuresthe elasticity of migration flows with respect to expected wages.

It should be noted that we have abstracted in the above discussion fromthe role of risk aversion in individual migration decisions, by focusing on "ex-pected income" differentials rather than "expected utility" differentials. Fora risk-averse worker, for instance, the probability of employment is a moreimportant determinant of migration than the wage rate. More generally, wehave not captured in our specification some of the other factors that mayaffect the decision to migrate, as emphasized in the recent analytical andempirical literature (see for instance Stark (1991) and Ghatak, Levine andPrice (1996)). Wage uncertainty in agriculture or the urban sector (resulting,in the former case, from output variability, and in the latter, from imperfectinformation about labor market conditions), income inequality and relativedeprivation in the rural sector, the joint nature of the decision process inhouseholds, and the lack of access to credit markets, may all affect migration

12Note that this expression for Ou assumes that there is no "auctioning" of publicsector jobs, that is, the government does not turn over its employees over every period asoccurs in the private sector. This assumption captures the view that public jobs are not"randomly" offered but result rather from various non-economic considerations, such aspolitical patronage. If there is turnover in the public sector, however, the probability offinding a job would be (Up,o + UG,O)/UU,O, instead of (Up,o(Uu,o -UG,O).

'3In principle, both Ou and OR vary over time. In the simulations reported below,however, we treat them as constant.

20

decisions and lead potential migrants to alter their decision to move."4 Forinstance, imperfect (and costly) information about urban labor market condi-tions may lead potential migrants to attach a greater weight to the availableinformation regarding the rural sector and may lead them to postpone theirdecision to move-despite the existence of a large (expected) wage differen-tial."5 However, some of these factors may be country-specific and we haveopted in building our prototype framework to focus on a more parsimoniousspecification that highlights the role of relative income opportunities.

3.2.2 Urban Unskilled Wages and Employment

The public sector is assumed to hire an exogenous level of unskilled workers,UG. For simplicity, the wage rate paid by the government to unskilled workersis set at the same level as the wage rate paid to that category of labor in theprivate formal sector. Furthermore, we assume that there is a legally-bindingminimum wage in place, which is indexed to either the consumer price indexor an alternative price:

WM = WM(PINDM) ndM ,

where PINDM = PLEV for instance, and 0 < indM < 1. In the simulationexercises reported later, we assume that the minimum wage is fully indexed(so that indM = 1) on the price of the composite factor T 1.'6

Labor demand in the formal private sector, Ud, is determined by firms'profit maximization. Given that formal private sector firms also borrow tofinance their wage bill prior to the sale of output, the effective price of laborincludes again the bank lending rate; thus

Ud = T 3XP1 ap mUp = T ((1 + ILL)wm PXPI , where WM =- (19)

\~~(i + axPIj PT,

As was the case for the rural sector (and in order to avoid corner solu-tions), we assume that the wage rate paid to unskilled labor in the formal

14Stark (1991) also emphasized that individual migration can be the outcome of a familydecision, often as a response to uninsurable risks.

15 We also abstract from the negative externalities associated with rural to urban migra-tion, such as congestion costs or pollution in shanty towns.

"6An alternative would be to assume that the minimum wage is fixed in nominal terms(indM = 0). As is well known, simulation results of particular shocks may differ signifi-cantly under the two assumptions (see for instance Devarajan, Ghanem, and Thierfelder(1999)).

21

urban sector is systematically greater than the real wage rate paid in theinformal sector. Consequently, unskilled urban workers will first seek em-ployment in the private formal sector. The actual level of employment inthat sector is determined according to equation (19). The remainder of theurban unskilled labor force, Uu, will thus seek employment in the informaleconomy, where there are no barriers to entry:

Uj = UU-UP-UG. (20)

From (6), the demand for labor in the informal sector is given by UI =

3px(Vj1/wI), where w, is the product wage defined as W,/PV1. Becausethe informal labor market clears continuously (Ud = U,), the equilibriumproduct wage is given by

WI = ,XI VI where w W (21)

The urban unskilled labor supply, Uu, grows as a result of "natural" urbanpopulation growth and migration of unskilled labor from the rural economy,as discussed earlier. Moreover, a fraction of the urban unskilled populationacquires skills (SKL) and leaves the unskilled labor force to increase thesupply of skilled labor in the economy. We make the additional assumptionthat individuals are born unskilled, and therefore natural urban populationgrowth. (not resulting from migration or skills acquisition factors) is repre-sented by urban unskilled population growth only, at the rate gu. Thus, thesize of the urban unskilled labor supply evolves according to

Uu = Uu,-(l + gu) + MIGR - SKL. (22)

We treat the growth rate of the urban unskilled population endogenouslyin our framework. We do so by dwelling on the several studies have docu-mented the existence of a negative association between population growthrates and income levels; fertility rates tend to fall as the level of income in-creases (see for instance Barro and Becker (1989), and Becker, Murphy andTamura (1990) for the growth implications of this relationship). Specifically,we assume that the growth rate of the urban unskilled population is relatedto a distributed lag of current and past values of the ratio of average incomein the urban economy for skilled and unskilled workers. Assuming that theselags follow a declining geometric pattern and using the Koyck transformation,

22

we get

gu = Aga&gu[ [U ( + ()UP + UG)WM] /UU-U > (23)

where 7gu > 0 measures the elasticity of the growth rate of the urban un-skilled population with respect to the wage ratio.

The acquisition of skills by unskilled workers is assumed to depend onthree factors: a) relative expected consumption wages of skilled to unskilledurban workers (as a proxy for the future stream of earnings associated withhigher levels of education); b) the government stock of education capital, KE,

which affects the ability to invest in skills; and c) the average level of wealthheld by each unskilled worker, which plays an important role in the presenceof liquidity constraints.

Consider first the effect of wages. In case they acquire skills, currentunskilled workers expect to earn wage Ws if they are employed (with proWability Os) and nothing if they are unemployed. The purchasing power ofthis wage is obtained by deflating it by the skilled consumption price index,Pus,-, (defined below):

EwS = 0 WS I,Pus,-1,

where Os is, as before, approximated by the initial ratio of the number ofskilled workers employed in the private sector, over the total number of skilledworkers that are not employed in the public sector seeking a job in the privatesector (that is, Os = Sp,O/(SO - SG,-O)).

In contrast, if they remain unskilled, they expect to get the average un-skilled wage, which is a weighted average of the minimum wage WM andthe informal wage rate. Assuming, again, that there is no job turnover inthe public sector, the average expected real wage is given by (17), which isrepeated here for convenience:

OUWM',- + ( 1-OU)W,,_Ewu - Puu,-1

with Ou as defined above.Consider now the effect of initial wealth. As can be inferred, for instance,

from various studies in the endogenous growth literature (see for instance DeGregorio (1996)), the existence of constraints in obtaining credit to fund the

23

cost of acquiring skills may completely outweigh the impact of relative wagesand the prevailing stock of government capital in infrastructure on workers'decisions. Indeed, in the framework that we consider here, only firms haveaccess to bank credit (see the discussion below); no matter how high thewage differential is or is expected to be, workers simply cannot borrow tofinance human capital accumulation."7 To capture the existence of thesecredit constraints we assume that the decision to acquire skills is a functionof the (lagged) ratio of the level of wealth of urban formal and informalunskilled households divided by the number of unskilled workers in the urbansector, (WTUI + WTUF)/UU (with WT defined below). Admittedly, this isa rather simplistic way of modeling these frictions; an alternative (but lesstractable) approach would be to assume that access to the credit market issubject to a threshold effect, with financial wealth playing a "signaling" rolebecause of workers' ability to pledge a fraction of it as collateral.

Given these three effects, the flow increase in the supply of skilled laborcan be written as:

SKL A5 [(WTUI_1 + WTUF,l) Ced. K (EWs) w(KE)UEI(24)

+(1-As)SKL_1,

where 0 < As < 1, Ke is a shift parameter, and /edu > 0. For simplicity, wetreat as constant the cost of acquiring skills (as measured by the number ofyears of schooling multiplied by the average cost of education per year).

Public investment in education, IE (which is treated as exogenous), de-termines the rate at which the stock of public capital in education grows overtime:

KE = KE,-1(1 - 6 E) + 1E, (25)PQp,,'

where 0 < BE < 1 is a depreciation rate. We assume that only the privategood is used to invest in education and therefore deflate nominal investmentby the demand price PQp (defined below).

1 7 This assumption is well supported by the evidence on the composition of bank credit inlow-income countries, which suggests that only a small fraction of bank loans is allocatedto households.

24

3.2.3 Urban Skilled Wages and Employment

As indicated earlier, we assume that wage-setting behavior for skilled laboris based on efficiency considerations. Specifically, we assume that in order toprovide incentives for employees to work and avoid shirking, firms must seta sufficiently high (product) wage. Along the lines of the specific functionalform derived by Agenor and Aizenman (1999a), we assume that the level ofeffort skilled workers provide depends negatively on the ratio of their realopportunity cost, Q2w, to their consumption wage, that is, the nominal wageWs deflated by the skilled consumption price index, Pus:

ef =1-efm ne where 'Yef > 0, (26)[Ws/Pus] f(6

and where 0 < efm < 1 denotes the "minimum" level of effort. The oppor-tunity cost of effort is taken to be constant in what follows.

Firms determine the levels of skilled and unskilled employment, as wellas the product wage for skilled labor, ws, so as to maximize profits, takingas given the real minimum wage paid and the production technology (10). Itcan be established that the demand for skilled labor is given by

r~~,dm(1~ /3 XP2 17XP2 W

p= ( 1 IL-,)WsaPXP2 , where Ws = PT2 (27)

whereas the optimal wage-setting equation is given by

13 XP2'Yef(1 - ef) ( T2 ) +PXP2 PUS (28)a=PXP2 tef -Sp) PT2

Note that in this equation the ratio of the consumption price index forskilled workers, Pus, over the price of the composite input, PT2 , appears be-cause firms set the product wage whereas effort depends on the consumptionwage. This creates an important channel (seldom accounted for in empiricalmodels) through which changes in relative prices affect wage formation.

Given that firms set wages and are on their labor demand curve, openskilled unemployment may emerge. The rate of skilled unemployment, UNEMPs,is given by the ratio of skilled workers who are not employed either by theprivate or the public sector, divided by the total population of skilled work-

25

ers: 18

UNEMPs = S-SG- SdP (29)

Note that here we assume that skilled workers who are unable to find ajob in the formal economy do not enter the informal economy, in contrastto unskilled workers. In principle, any worker who is not hired in the for-mal economy could get a job in the informal economy at the going wage.However, there are several reasons why they may choose not to do so. Animportant consideration may be that skilled workers are concerned aboutpossible adverse signaling effects associated with employment in the infor-mal economy and may prefer instead to remain openly unemployed (Gottfriesand McCormick (1995)). Indeed, the existence of "luxury" unemployment isa well-documented feature of the labor market in developing countries.

The skilled labor force evolves over time according to:

S = S-i + SKL. (30)

Because we assumed that workers are born unskilled, the skilled workforcedoes not grow at the natural growth rate of the urban population, but ratherat the rate at which unskilled workers choose to acquire skills. Note also that,for simplicity, we assume no "de-skilling" (or obsolescence) effect, althoughthis could be easily captured.

It is important, to conclude this discussion of wage formation mechanisms,to return to our assumption that skilled workers' wage is determined on thebasis of an efficiency effect of compensation on productivity. In practice,it is notoriously difficult to discriminate among various forms of efficiencywage theories; these theories tend to be "observationally equivalent." At thesame time, however, this makes our choice here of an explicit formulationbased on the wage-productivity link less restrictive than it appears at firstglance; in fact, a wage-setting equation fundamentally similar to (28) canalso be derived by considering, say, turnover costs. A bargaining frameworkbetween firms and a centralized trade union could also lead to a similarwage-setting specification. The important point is to assume in each casethat whereas firms are concerned with the product wage, workers (or theunion that represents them) are concerned with the consumption wage. This

18As noted earlier, skilled employment in the public sector is set exogenously by thegovernment.

26

creates a wedge (as derived above) through which relative prices affect wage-setting decisions.

Although our treatment above provides a tractable specification of thewage-productivity link, a more general specification of wage formation forskilled labor may also be warranted in some applications. Such a specificationis also available in IMMPA and consists of replacing (28) by the "generic"formulation

WS = WS(PINDs )inds(UNEMPs)-OuQ1 (p ) )2 (1 + IL_1) 3, (31)

where PINDs = PLEV or PT2 (depending on whether the nominal wageis indexed to the overall level of prices or the composite "product" price),0 < inds < 1, and Qw, •u, Xl, 02, )3 > 0.

If this specification is used, ws in equation (27) must be defined as ws =WS/PT2, with Ws given in (31), and the production technology, equation(10), must be replaced by

Vp = CES{CES[CES(Sp, Kp), Up], KG}. (32)

Specification (31) is quite flexible; for instance, full indexation on theconsumer price index only requires setting PINDs = PLEV, inds = 1, andOU = 01 = 02 = 03 = 0. To assume that the product wage depends onlyon the ratio PUs/PT2 (as above) requires setting PINDs = PT2 , inds = 1,02 > 0, and •u = 4l = 03 = 0. Note also that in (31), as long as 4 u > 0,the level of skilled unemployment will affect (negatively) the level of nominalwages, instead of the rate of growth of wages (as would be the case with aPhillips curve formulation). This level effect is consistent with various formsof efficiency wage theories in which unemployment acts as a "worker disciplinedevice," such as the effort elicitation model of Shapiro and Stiglitz (1984);evidence supporting it has been provided for instance by Hoddinot (1996) forCote d'Ivoire. Finally, note that the gross lending rate, 1 + IL_1 , enters inthis specification because, say, an increase in the cost of borrowing raises theeffective price of labor, which firms try to offset, at least in part, by reducingthe nominal wage. If O3 < 1 (03> 1) the net effect is an increase (reduction)in the effective cost of skilled labor, and the demand for skilled labor will fall(increase). The wage equation (31) is used by Ag6nor, Fernandes, Haddad,and van der Mensbrugghe (2003) in their IMMPA framework for Brazil.

Note that the inverse relationship between the level of skilled unemploy-ment and the skilled workers' wage can also be introduced in either (26) or

27

(28), by assuming that the reservation wage, Qw, is inversely related to theunemployment rate:

Qw = Qw(UNEMPs)-u,

where qu > 0. Put differently, unemployment acts as a "disciplining device"by reducing the perceived opportunity cost of working and/or in reducingwage demands. In the simulation experiments reported below, however, wefocus on the case where qu = 0.

3.3 Supply and Demand

As noted earlier, we assume that both the informal and public sector goodsare nontraded. This therefore implies that, in each sector, total supply isequal to gross production, that is,

XI = Q,, XG= QG (33)

Agricultural and private formal urban goods, by contrast, compete withimported goods. The supply of the composite good for each of these sectorsconsists of a CES combination of imports and domestically-produced goods:

QAN = CES(MAN,DAN), (34)

Q= CES(Mp, Dp), (35)

where DAN = XAN-

Aggregate demand for each of these sectors consists of intermediate andfinal consumption, government spending, and investment demand:

QAN = CAN + INTAN, (36)

I = C, + INTI, (37)

Q = CG + ZG + INTG, (38)

QP = Cp + Gp + Zp + INTp, (39)

where INTj (for j = AN, I, G, P) is defined as total demand (by all produc-tions sectors) for intermediate consumption of good j:

INTj = E ajiXi for i = AN, AT, I, P, G. (40)i

28

Government expenditure on good j, Gj, is expressed in real terms as:

Gj = G for j=p, AN, G, (41)-ggj j5Q

where G represents total nominal government expenditure, PQh is the marketprice of goods purchased by the government, and 99AN+ 99G +99P = 1. Notethat the government is assumed not to spend on informal sector goods.

For the nontraded agricultural and informal sectors, aggregate demand(QAN and QI) consists of intermediate consumption and demand for finalconsumption (CAN and Cl), whereas aggregate demand for the public good,QG, consists of intermediate consumption as well as demand for final con-sumption, CG, and investment demand, ZG. Aggregate demand of the privateformal good, Qp, is taken to consist of intermediate consumption as well asdemand for final private consumption, Cp, final government consumption,Gp, and private investment, Zp.

Final consumption for each production sector i, Ci, is the summationacross all categories of households of nominal consumption of good i, deflatedby the demand price of good i:

Ci Cih = EhCh cc-h CONh where 0 < ccih <1, E i 1h ~~~PQ, CCi = 1(2

where Cih is consumption of good i by household h and PQi is the com-posite sales price of good i (defined below). Coefficients ccih indicate howtotal nominal consumption by household h, CONh, is allocated to each typeof good. Equations (42) can be derived by maximization of a Stone-Gearyutility function (see for instance Deaton and Muellbauer (1980)). They rep-resent a linear expenditure system in which, for simplicity, the subsistencelevel of consumption is set to zero.

Finally, aggregate investment made by firms, Z, consists of purchases ofboth public and private goods and services (ZG and Zp respectively):

Z*PKZ, = ZZ, PQ , where ZZG + ZZP.

Coefficients zzi measure the allocation of total investment demand topublic and private goods.

29

3.4 External lrade

As indicated earlier, firms in the private formal sector allocate their outputto exports or the domestic market according to the production possibilityfrontier (PPF) specified in equation (11) and the relative price of exports(PEp) vis-a-vis domestic goods (PDp). Efficiency conditions require thatfirms equate this relative price to the opportunity cost in production. Thisyields:

( PEp __-__T

EP = DP (PDP. 1 -T ) (43)The agricultural traded good is fully exported, as also indicated earlier;

given that this sector is the only one using only its own good as intermediateconsumption, we have

EAT = VAT = (1 -aAT,AT)XAT- (44)

A similar reasoning applies to the determination of the demand for im-ports. We assume that imports compete with domestic goods in the agricul-tural nontraded sector as well as in the private formal sector. Making use ofArmington functions for the demand for imported vs. domestic goods andrelative prices, import demand for both sectors (MA and MP) can be writtenas:

MAN =DAN (PDAN /3 QA O-QAMAN = DAN (PMAN 1 -f3QA) (45)

MP = DP (PDp /QP ) QPMp p (PMP 1 - /3 QP )(46)

These equations indicate that the ratio of imports to domestic supplyof both categories of domestic goods depends on the relative prices of thesegoods and the elasticity of substitution, UQA and oQp, between these goods.

3.5 Prices

As we have seen previously production requires both factor inputs and in-termediate inputs; we therefore define the net or value added price of outputas:

PVi == -PXj(1-indtaxi)- ajiPQj Xi, where i, j = AN, AT, I, P, G

(47)

30

where indtaxi is the rate of indirect taxation of output in sector i (withindtax, = 0 because there is no indirect taxation of informal sector output).This equation relates the value added price of output of sector i to the price ofgross output, PXi, net of indirect taxes, less the cost of intermediate inputs(purchased at composite prices).

We are considering a small economy and therefore assume that the worldprices of imported and exported goods are exogenously given. The domesticcurrency price of these goods is obtained by adjusting the world price by theexchange rate, the import tariff rate, tm, or the export subsidy rate, te:

PE, = wpei(l + tei)ER, for i = AT, P, (48)

PMi = wpmi(1 + tmi)ER, for i = AN, P. (49)

Because the transformation function between exports and domestic salesof the urban private good is linear homogeneous, the sales price, PXp, isderived from the expenditure identity:' 9

PXPXP = PDPDP + PEPEP,

that is,20

= PDPDP + PEPEP (50)XP

For the informal and public sectors (both of which produce goods thatare not exported and do not compete with imports), the composite salesprice is identical to the price of domestic sales, which in turn is equal tothe price of gross output. In the agricultural sector, the sales price of thetraded agricultural good, PXAT, is simply the domestic-currency price of

19An alternative approach to price formation in the private formal sector is to assumethat prices are set monopolistically, as markups over input costs (as for instance in Yeldan(1997), among others). With a fixed markup, equilibrium would then require the level ofoutput to be demand determined, implying that the production function would be droppedfrom the system.

201n solving the model, we use equation (46) to solve for PDp, and, because QSp = Qdwe use (39) to solve for the equilibrium value of Qp. We then invert the composite goodCES equation (35) to solve for Mp and invert the CET function (11) to solve for Dp.This procedure ensures that the composite price (and thus indirectly the price of domesticsales) adjusts to equilibrate supply and demand.

31

agricultural exports, PEAT, whereas the sales price of the nontraded good,PXAN, is equal to the domestic price of agricultural goods, PDAN.

For the nontraded agricultural sector and private urban production, thesubstitution function between imports and domestic goods is also linearlyhomogeneous, and the market price is determined accordingly by the expen-diture identity:

PQiQi = PDiDi + PMiMi, for i = AN, P,

that isPDiDi + PM1 MiPDQDi = P , for i = AN, P. (51)

Qi

For those sectors that do not compete with imports (informal and publicsector goods), the domestic price, PDi, is simply equal to the market price,PXi:

PDi = PXi, for i = I, G. (52)

The nested CES production function of private formal urban goods is alsolinearly homogeneous; prices of the composite inputs are therefore derived insimilar fashion:

T1 PT1 = T 2PT2 + (1 + IL-1)WMUP, (53)

T 2 PT2 PROFP + (1 + IL- 1 )WsSp, (54)

where PROFp, as defined below, denotes profits of private firms in the urbanformal sector.

The price of capital is constructed as using the investment expenditureidentity, which involves those goods for which there is investment demand,namely, the public good and private-formal urban good (see equations (38)and (39)):

PK = Ei PQiZi PQGZG + PQPZP (55)z z

Markets for informal goods and government services clear continuously;equilibrium conditions are thus given by

QS = Qd QS = Qd

32

In solving the model, we use equations (33) to determine the equilibriumquantities Q, and QG, that is, equations (5) and (7). We also use the demandequations (37) and (38) to solve residually for C, and CG, that is:

XI-G,-INT, = Cl, (56)

XG-GG-ZG-INTG = CG. (57)

Equation (42) for i = I, G, is then solved for PQ, = PX, and PQGPXG, respectively. This yields:

PX Eh CCihCONh i =I, G. (58)ci

The aggregate price level, PLEV, or consumer price index (CPI), is aweighted average of individual goods market prices, PQi:

PLEV = CPI = E wtiPQi, (59)

where 0 < wti < 1 denotes the relative weight of good i in the index, andPQ1 = PD1 and PQG = PDG. These weights are fixed according to theshare of each of these goods in aggregate consumption in the base period.Inflation is defined as the percentage change in the price level:

PLEV - PLEV-1 (60)PLEV-1

Finally, the consumption price index for the rural sector is given by

PR = wriPQi, (61)

whereas the consumption price indexes for urban unskilled and skilled work-ers are given by

Puu = EwuiPQi, (62)

Pus = wsiPQi, (63)

where the wri, wui and wsi are relative weights with Ei wr, Ei wui andEi wsi summing to unity. The deflator of GDP at factor cost (used below),is given by

33

PGDPFC = EiviPVi, v1 _ PViXi/E3PV7X, (64)

where, again, Eivi 1.

3.6 Profits and Income

Firms' profits are defined as revenue minus total labor costs. In the case ofagricultural nontraded sector firms and urban informal sector firms, profitsare simply given by

PROF, = PViV - WA, for i = AN, I. (65)

Firms producing in the traded agricultural sector must include workingcapital costs as well in measuring their production costs, that is, interestpayments on their wage bill; their profits are therefore given by

PROFAT = PVATVAT - (1 + IL_1)WATUAT. (66)

Finally, profits of private-urban sector firms account for both workingcapital costs and salaries paid to both categories of workers:

PROFP = PVpVp - (1 + IL-)UpWM - (1 + IL- 1 )SpWs, (67)

where, as noted earlier, the nominal wage paid to unskilled workers is thelegally-imposed minimum wage, WM.

Firms' income is simply equal to profits minus interest payments on loansfor investment purposes. Firms' income and profits are defined separately,because not all sectors are assumed to borrow on the credit market to financeinvestment. Specifically, we assume that only firms in the formal urbaneconomy accumulate capital.2 1 Firms' income can thus be defined as:

YF, = PROFi, for i = AN, AT, I, (68)

YFP = PROFP - IL 1 DLP,-1 - IF * FLP,_1 ER, (69)

where IF is the interest rate paid on foreign loans, and DLP and FLP are thelevels borrowed domestically and abroad by private urban firms for physicalcapital accumulation.

21f course, this assumption can be relaxed in specific country applications.

34

Commercial banks' profits must also be taken into account. They aredefined as the difference between revenues from loans to firms (be it for work-ing capital or investment needs) and interest payments on both households'deposits, Eh DDh, and foreign loans received from international creditors,FLB:

YFPB = IL- 1 [DLp,-1 + DLG, 1 + UATWAT + WMUP + WSSP] (70)

-ID E DDh,.l - IF. ER. FLB,_1 ,h

where ID is the interest rate on bank deposits, assumed to be set by thecentral bank.

Household income is based on salaried labor, distributed profits, trans-fers, and net interest receipts on holdings of financial assets. Householdsare defined according to both labor categories and their sector of location.There are two types of rural households: one comprising workers employedin the traded sector, and the other workers in the nontraded sector. Inthe urban sector there are two types of unskilled households, those workingin the informal sector and those employed in the formal sector. The fifthtype of households consists of skilled workers employed in the formal urbaneconomy (in both the private and public sectors). Finally, there are "capital-ist" households (including rentiers) whose income comes from firms' earningsin the formal private sector, the agricultural traded sector and commercialbanks. We further assume that households in both the nontraded agriculturalsector and in the informal urban economy own the firms in which they areemployed-an assumption that captures the fact that firms in these sectorstend indeed to be small, family-owned enterprises.

Income of agricultural nontraded and urban informal groups is given by

YHi = iTRH + WA + ID DDj,_ + IF FDi,-ER + YFi, for i = AN, I,(71)

where yj is the portion of total government transfers (TRH) each group re-ceives, Wii denote wage earnings, DDi domestic bank deposits, FDi foreignbank deposits (taken to be a fairly small share of assets in the simulationsreported below), and YFi firms' income in these sectors.

Income of the agricultural traded sector household, as well as that ofthe urban formal unskilled and skilled households, depends on governmenttransfers, salaries and interests on deposits; firms provide no source of income,

35

because these groups do not own the production units in which they areemployed:

YH, = -yiTRH+WiU 1+ID.DDi,1±+IF*FDi,_1 ER, for i = AT, UF, (72)

YHs = ySTRH + Ws(Sp + SG) + ID- DDs,-1 + IF FDS,_1 ER, (73)

where UUF = UP + UG.Firms' income in the traded agricultural and private urban sectors are as-

sumed to go to capitalist households, along with commercial bank's income,and interest on deposits. Because there is no capital accumulation in thetraded agricultural sector to be financed, the entire amount of firms' profitsfrom that sector are transferred to capitalist households. By contrast, firmsin the private urban sector retain a portion of their earnings, re, for invest-ment financing purposes, and transfer the remainder to capitalist households.Capitalist households' income is thus:

YHKAP = ID . DDKAP,-1 + IF FDKAP,-1ER + YFAT (74)

+(1- re)YFp + YFPB + ^yKApTRH.

3.7 Savings, consumption, and Investment

Each category of household h saves a fraction, 0 < savrateh < 1, of itsdisposable income:

SAVh = savratehYHh(l - inctaXh), (75)

where 0 < inctaxh < 1 is the income tax rate applicable to household h.The savings rate is a positive function of the real interest rate on deposits:

I1±ID USh (6savrateh = 8 0,h ( + PINF) (76)

Note that in practical applications the propensity to save can be made afunction of not only the real deposit rate, which implies that inflation andthe savings rate are inversely related, but also of the inflation rate itself, asa result of a precautionary motive (with higher inflation acting as a signalof greater uncertainty about future real income). The evidence on the lattereffect is significant (see Agenor (2000) and Loayza, Schmidt-Hebbel, andServen (1999)) and may be highly relevant for some countries.

36

The portion of disposable income that is not saved is allocated to con-sumption:2 2

CONh = (1- inctaXh)YHh - SAVh.

Finally, the total flow of savings of each household is channeled into theaccumulation of financial wealth, WTh, which also accounts for valuation ef-fects on the stock of foreign-currency deposits, FDh, associated with changesin the nominal exchange rate, ER:

WTh = WTh,_1 + SAVh + AER- FDh,1l.

Capital accumulation occurs only in the private urban sector. The de-cision to invest is assumed to depend on several factors. First, there is apositive effect of the after-tax rate of return to capital relative to the costof funds. Second, there is an accelerator effect, which aims to capture theimpact of the desired capital stock on current investment. Third, there isa negative effect of the (lagged) inflation rate, which may be viewed as ameasure of macroeconomic instability. Fourth, there is a positive effect ofthe public capital stock in infrastructure-cumulated investment in railroads,paved roads, water systems, telecommunications, and power-a relationshipfor which there is also robust evidence (see Agenor and Montiel (1999)).Finally, there is a negative effect of the economy's level of debt, which mayresult from several possible factors: a) the risk of confiscation associated witha debt overhang (as discussed earlier); b) the diversion of foreign exchange toservice foreign debt and consequently insufficient amounts of foreign currencyto import capital goods; and c) the possibility that a high external debt mayforce a reallocation of public expenditure away from productive uses (mainte-nance and infrastructure investment, most notably) and toward debt service.In the second case the inclusion of foreign debt acts as a "proxy" for foreignexchange availability, whereas in the third it is a proxy for the compositionof public spending.2 3

22Note that we do not account for any real balances effect (or wealth effect) on con-sumption, as in Easterly (1990) and others. This effect can be, however, easily added ifwarranted by the empirical evidence-for instance by making the savings rate a function ofwealth as well. It could prove important in assessing the effects of exchange-rate inducedvaluation changes on domestic expenditure.

23jt should be noted that the disincentive effects of external debt may be related not onlyto private investment but also to international capital flows. As discussed by Khan andHaque (1985), high public external debt can lead to capital flight if domestic investment is

37

Specifically, we model the investment function by firms in the privateurban sector as

Z (KINF ) {(1 + AURGDPFC )ACC (77)

Kp,_, KINF,-1 RGDPFC,-l

z _ _ ((1 + IK) (1 - inctaXKAP) IK

(1 + PINF_,)-p 1 + IL

KIFG * ER.* FLG,-l0 IFG * ER * FLG,-l0 2-OD (F )VREV ) |PDD (IFG R2

The analytical form adopted for the investment function incorporatessome key features for the analysis of debt reduction strategies. Equation(77) indicates that the ratio of investment, Z, to the (lagged) stock of pri-vate capital, KP, is positively related to the return to capital, IK, net ofthe income tax rate that capitalists are subject to. This term introducesan adverse effect of higher taxes on investment. The negative effect of thelending rate represents the cost of borrowing to finance capital accumula-tion. High inflation also has a negative effect on investment decisions; thisvariable captures the impact of increased uncertainty about relative pricesunder higher inflation, which makes investment decisions riskier. Severalrecent studies-see, for instance, Serven (1997, 1998) and Zeufack (1997)--have indeed shown that macroeconomic instability may have a significantdetrimental effect on the decision to invest, particularly when capital outlaysare irreversible. There is also substantial empirical evidence that public andprivate sector capital in infrastructure tend to be complements (see Agenorand Montiel (1999)). This is accounted for by the addition of the growthrate of the public capital stock in infrastructure (KINF) in determining thegrowth rate of private capital.

Second, equation (77) integrates the "flexible" accelerator effect on pri-vate investment. The ability of the firm to respond to changes in its desiredcapital stock is reflected in the positive effect on investment of the growth invalue added, which is measured by changes in real GDP evaluated at factorcost, RGDPFC, defined as

subject to "expropriation risk." Conversely, debt reduction may not only stimulate privatedomestic capital formation but also net capital inflows; see for instance, Bhattacharya,Montiel and Sharma (1997) for sub-Saharan Africa. This may be an important source ofpositive externality associated with debt relief, which suggests that the benefits of thisshock, as discussed below, may be under-estimated.

38

RGDPFC = EiPViXiIPGDPFc, (78)

where PGDPFC, the deflator of GDP at factor cost, is defined in (64).There is considerable evidence supporting this effect, particularly for sub-

Saharan Africa, as discussed for instance by Ag6nor (2000, Chapter 1). Fi-nally, the last two terms in equation (77) indicate that investment is inverselyrelated to the ratio of interest payments on public sector debt to tax rev-enues. This ratio may capture various factors (as noted above), includingthe existence of confiscation risk. For instance, when government revenuesfall, investors may infer that there is a higher probability that private sectorcapital may be either taxed or confiscated to finance existing debt service.

This particular form of the investment function introduces a key rolefor fiscal policy. For instance, increasing income tax rates will increase taxcollection and therefore result in a positive effect by reducing confiscationrisk; but at the same time it will also reduce the net return to physicalcapital. A debt reduction program combined with suitable fiscal policies mayboth reduce interest payments and improve tax collection, thereby reducingconfiscation risk and boosting private investment. The specific formulationthat we have adopted here (which includes both linear and quadratic termsin the debt service-to-taxes ratio) implies that the marginal effect on privateinvestment of a reduction in the debt ratio is magnified if the initial levelof that ratio is high. We could also have assumed that the relationshipbetween investment and external debt has a concave form, as suggested bythe econometric results of Elbadawi, Ndulu, and Ndungu (1997); in that case,the coefficient OD should be negative, implying that external debt has at firsta positive impact and a negative one only when the it becomes relativelylarge.

The rate of return on capital is defined as the ratio of profits to the stockof capital:

IK= PROFp (79)PK~ (79)

Capital accumulation depends on the flow level of investment, Z, and thedepreciation rate of capital from the previous period, Sp:

Kp = Kp,- (1 - p) + Z-1 , (80)

where 0 < 6p < 1.

39

3.8 Financial Sector

The financial balance sheets of each group of agents are presented in summaryform in Table 1. In what follows we consider in turn the determination ofthe portfolio structure of households, the demand for credit by firms, andthe behavior of commercial banks.

3.8.1 Households

In contrast to various FCGE models that assume that existing stocks ofassets cannot be traded, and only additional flows from savings can be al-located to existing assets, we assume here that agents can freely alter thedesired composition of their stock of financial wealth-subject to the overallconstraint that initial or beginning-of-period wealth is predetermined at anygiven moment in time. Each category of households allocates instantaneouslyits stock of wealth to either money (in the form of cash holdings that bear nointerest), Hh, domestic bank deposits, DDh, or foreign bank deposits, FDh(see Figure 3):

WTh = Hh + ER FDh + DDh. (81)

Note that in our definition of private wealth we have excluded land andother types of real assets (such as livestock), which can be important forhouseholds located in the rural sector. Some of these assets are often held in"unproductive" form and no market per se exists to measure their relativeprice. Thus, in practice, accounting for real assets is likely to raise someinsurmountable measurement problems, involving both values and quantities.Nevertheless, it should be kept in mind that in a setting in which such assetsare accounted for, a reallocation of wealth away from (say) real to financialassets would have significant real effects, by affecting interest income oninterest-bearing assets (such as bank deposits), disposable income, and thusexpenditure.

Real money demand functions for each household category are taken todepend positively on real income (measured in terms of the overall pricelevel, which is nothing but the inverse of the purchasing power of one unit ofcurrency), and negatively on inflation (as a proxy for the opportunity costof holding money instead of real assets) and the rates of return on domesticand foreign deposits (which measure the opportunity cost of holding money

40

instead of interest-bearing financial assets):

Hd PLEV (PLEV) (1 + ID)-hD[(1 + IF)(1 + dev)]jPhF .(82)

(1 + PINF) -PPINF,

where dev denotes the expected devaluation rate, that is, the expected rateof change in ER, which is taken as exogenous. This specification of themoney demand function allows us not only to account for different elasticitiesbetween domestic and foreign deposits, but also for different elasticities acrosshouseholds.

An alternative, simpler specification is to assume that the demand formoney balances is proportional to total consumption, as a result of a "cash-in-advance" constraint:

Hh = CONSh.

Because of our assumptions of a fixed exchange rate and incomplete ster-ilization, the nominal money supply (which is derived below from the basemoney stock), HI, is determined endogenously. In equilibrium, this stock isequal to the total sum of money demanded by households: 24

Hs=EHg. (83)h

The portion of wealth that is not held in the form of noninterest-bearingcurrency is allocated between domestic and foreign deposits. The relativeproportions of holdings of each of these two categories of assets are taken todepend on their relative rates of return:

'fBh 1±+ID aBh

1 -YBh ( ( 1 + IF) (1 + dev) (84)

where 'YBh represents the proportion of domestic deposits held in total de-posits:

7Bh ~ DDh (85)=Bh=DDh+ER *FDh(

24When computing the solution of our model, this equation is dropped. Given Walras'law, if all other markets but the money market are in equilibrium, then the money marketmust be in equilibrium as well. Our computer program checks that this equation indeedholds continuously.

41

In solving the model, we use equation (84) to determine the optimallevel of domestic bank deposits, whereas we use equation (81) to determineresidually the level of foreign deposits.

3.8.2 Firms

Firms finance their investment plans (as defined above) through retainedearnings and domestic (DLp) and foreign (FLp) loans:

PK - Z = ADLp + ER - AFLp + re * YFp.

Solving this equation for DLp gives us the demand for bank loans:

DLd = DLp,_1- ER .AFLp + PK * Z-re - YFp. (86)

The path of foreign loans is set exogenously. This implicitly accounts forceilings that firms may face in their access to foreign markets.

3.8.3 Commercial Banks

Banks are at the heart of the financial system in our archetype economy,as is indeed the case in many low- and middle-income developing countries.Commercial banks in our framework are required to keep a portion 0 <rreq < 1 of the deposits that they collect as reserve requirements:

RR = rreq E DDh. (87)h

The balance sheet of commercial banks is

NWPB = DLP + DLG + RR-Z DDh- ER FLB, (88)h

where NWPB denotes banks' net worth, DLP loans to the private sector,DLG loans to the government, and ER * FLB foreign loans (measured indomestic currency terms).

Equation (86) represents the demand for loans. We assume that the actualstock of loans is demand determined, and that banks borrow on world capitalmarkets the required "shortfall" given their domestic sources of funds. The

42

commercial banks' balance sheet is thus used to determine FLB (see Figure4): 25

ER * AFLB = ADLp + LXDLG - (1- rreq) E ADDh. (89)h

Given (89), and given that all their profits are distributed to households,the net worth of commercial banks evolves over time according to,

NWPB = NWPB,-1 - AER. FLB,-1 , (90)

where the second term on the right-hand side represents again valuationeffects associated with nominal exchange rate changes.

Banks set the loan interest rate as a premium over the average costof funds-including the devaluation rate, which affects the cost of foreignborrowing-taking into account the (implicit) cost of holding reserve require-ments:

IL = PR ID' b[(1 + IF) (1 + dev) - 1]1-Qb (91)

1 -rreqwhere 0 < ab < 1 is measured by the initial share of domestic deposits inbanks' total funds (that is, Ceb = Eh DDh,o/(Eh DDh,O + FLB,oERo)), andPR denotes the finance premium, which is assumed to be set according to:

PR S [= (&c(NWp+DLP )) + (1 - 6,)PR_1, (92)

where 0 < p, < 1 is the speed of adjustment, 0 < 6, < 1, and NWp is thenet worth of private urban firms in nominal terms, defined as

NWp = PK Kp-DLp-ER FLp.

NWp changes over time according to

NWp = NWp,-1 + PK AKp - ADLp - ER AFLp - AER. FL,9.3)

+APK- Kp,l.

The last two terms on the right-hand side of this expression representcapital losses associated with depreciations of the nominal exchange rate andcapital gains associated with changes in the price of capital.

25 Note that capital losses associated with nominal exchange rate changes, AER*FLB, 1I,axe accounted for in equation (70).

43

Our specification of pricing decisions by commercial banks allows us tocapture balance sheet effects in the determination of loan rates. It is consis-tent with the current line of research emphasizing the role of collateral andthe impact of borrowers' net worth on the "finance premium," as illustratedin the work of Bernanke, Gertler, and Gilchrist (2000), Kiyotaki and Moore(1997), and Izquierdo (2000). The higher the value of the private capitalstock net of foreign borrowing (that is, "pledgeable" collateral, PK * Kp-ER- FLP, or an "effective" fraction 6, of that amount) relative to theamount of domestic loans, DLP, the higher the proportion of total lendingthat banks can recoup in the event of default by seizing borrowers' assets.This reduces the finance premium and the cost of borrowing, stimulatingthe demand for credit. The dependence of the cost of funds on net worthis a critical aspect of the model; for instance, a nominal exchange rate de-valuation (a rise in ER), reduces firms' net worth and may dampen privateinvestment by increasing the cost of capital.

An alternative justification for the finance premium equation (92) canbe found in the models of credit market imperfections recently developedby Agenor and Aizenrnan (1998, 1999b). These models, following Townsend(1979) and Helpman (1989), emphasize the importance of monitoring and en-forcement costs of loan contracts that lenders face in a weak legal environment-as is so often the case in developing countries. In such an environment, thesecosts may be an increasing function of the amount lent (even against "good"collateral) because of congestion in courts and the difficulty of settling legalclaims, which make it hard for lenders to actually seize borrowers' assets incase of default. This approach amounts to specifying the premium as a pos-itive function of the ratio of the amount lent DLP over "effective" collateral(as is done here) or separately as a negative function of PK KP- ER * FLP,with possibly a lower elasticity than that associated with DLP.

Finally, it should be noted that the assumption that banks borrow at willon world capital markets to satisfy the demand for domestic loans may notbe appropriate for all low-income countries, because domestic financial inter-mediaries (even local subsidiaries of foreign banks) may have either limitedaccess to these markets-and thus be subject to quantity rationing-or mayface themselves a rising risk premium on external funds (as, for instance, inAgenor and Aizenman (1998)). An alternative approach, which avoids in-troducing credit rationing of domestic borrowers and its complications, is toassume that commercial banks hold excess liquid reserves (above and overrequired reserves), and that movements in such reserves adjust endogenously

44

to equilibrate the credit market-with foreign borrowing taken as exogenous.

3.9 Public Sector

The public sector in our framework consists of the government and the centralbank. We consider them in turn and relate changes in official reserves to thebalance of payments.

3.9.1 Central Bank and the Balance of Payments

From the central bank's balance sheet, its net worth, NWCB, is given by

NWCB = DCG + ER FF-MB, (94)

where DCG denotes domestic credit to the government, FF the stock offoreign reserves, and MB the monetary base. Assuming that capital gainsand losses are not monetized, changes in the monetary base reflect changesin credit to the government, as well as changes in official reserves:

MB = MB-1 + ADCG + ER /AFF. (95)

Assuming no operating costs, net profits of the central bank, PROFCB,are given by the sum of interest payments on loans to the government, andinterest receipts on holdings of foreign assets:

PROFCB = IL- 1 DCG,_1 + IFG FF-1 . (96)

where IFG is the interest rate on foreign loans to the government. We assumethat net profits of the central bank are transferred to the government.

Whereas domestic credit to the government, DCG, is treated as an ex-ogenous policy variable, the accumulation of foreign reserves depends on thebalance of payments (see Figure 5), as any current account surplus (or deficit)must be compensated by a net flow of foreign capital:

AFF E(wpeiEj - wpmiMi) + IF 3 E FDh,_1 (97)i h

-IF FLP,-1-IFG(FLG,l - FF-1) - IF- FLB,-1

-ZAFDh+ AFLG+AFLp +AFLB.h

45

Equation (97) determines, the change in the foreign-currency value of offi-cial reserves, AFF, required to clear the balance of payments, given changesin households' holdings of foreign assets, Eh AFDh, changes in foreign loansmade to the government, AFLG, and to private firms, AFLp (both taken tobe exogenous), changes in loans to domestic banks, AFLB, and the currentaccount.

The monetary base consists of currency, H 8, and reserve requirements,RR. The supply of currency to households is thus given by

H8 = MB - RR. (98)

Note that the model can also be solved with a flexible exchange rate (asopposed to a fixed exchange rate), in which case FF would be kept con-stant (that is, AFF = 0) and equation (97) would be solved for the nominalexchange rate (which affects trade volumes). Under this regime, currencyfluctuations can have sizable effects not only on the banking system but alsoon private sector balance sheets and the functioning of the economy; as dis-cussed earlier, the "finance premium" depends not only on the capital stockbut more generally on the net worth of private borrowers, which accounts forforeign borrowing.

Finally, given (94) and (95), the central bank's net worth evolves overtime according to:

NWCB = NWCB,-1 + LER FF_1, (99)

where the last term represents valuation effects.

3.9.2 Government

We assume that government expenditures consist of government consump-tion, which only has demand-side effects, and public investment, which hasboth demand- and supply-side effects. Public investment consists of invest-ment in infrastructure, education, and health.2 6 We define investment ininfrastructure as the expenditure affecting the accumulation of public in-frastructure capital, which includes public assets such as roads, power plantsand railroads. Investment in education affects the stock of public education

2 61t should be noted that this treatment of public investment differs from standarddata classification reported in national accounts; in many instances these investments areclassified as current expenditures.

46

capital, which consists of assets such as school buildings and other infrastruc-ture affecting skills acquisition, but does not represent human capital. In asimilar fashion, investment in health adds to the stock of public assets suchas hospitals and other government infrastructure affecting health.

Government saving is defined as minus the government budget deficit:

-DEF = [PVGXG - WMUG - WsSG] + PROFCB (100)

+TXREV - TRH - G - IF0 ER. FLG,l

-IL-,(DCG,-1 + DLG,-l)

The term in square brackets represents profits by the government fromsales of the public good. TXREV denotes total tax revenues whereas TRHis government transfers to households. G represents total government ex-penditures. PROFCB represent profits from the central bank. The finaltwo terms in the government budget include interest payments on loans fromabroad, and interest payments on domestic loans by the central bank andcommercial banks (see Figure 6).

Using the definition of net profits of the central bank given in equation(96), government saving can be rewritten as

-DEF = PVGXG - WMUG -WSSG (101)

+TXREV - TRH - G -IFG - ER. (FLG,_1 - FF-1 )

-IL-DLG,-l.

Total tax revenues, TXREV, consist of revenue generated by importtariffs (net of export subsidies), sales taxes, and income taxes:

TXREV = (WPmANtmANMANER) + (wpmptmpMpER) (102)

- (wpeATteATEATER) - (wpeptepEpER) + indtaxiPXiXi

+inctaxr(YHAT + YHAN) + inctaxuu(YHuF + YHs)

+inctaXKAP (YHKAP).

Note that in this prototype framework we do not account explicitly forpayroll taxes, although this could be important to study complementaritiesbetween tax and labor market reforms.

Government expenditure is defined as investment in infrastructure, IINF,

investment in health, IHi investment in education, IE, and other current

47

expenditures besides labor costs, Gc, which are all considered exogenouspolicy variables:

G=IINF+IE+IH+Gc. (103)

Government investment increases the stock of public capital in eitherinfrastructure, education or health. The stock of public capital in educationincludes items such as school buildings, whereas the stock of health capitalincludes hospitals and the like. Infrastructure capital includes all other stocksof public property, such as roads, railroads, and power plants. Accumulationof each type of capital is defined as:

Ki = Kj,_1 (1 -i) + pQPl , where i = INF, H, E, and where 0 < 8i < 1.

(104)The main reason for treating government capital at a disaggregated level

is that, as noted earlier, we want to capture the different long-run effects thatthe allocation of public resources may have on the economy and ultimately onthe poor. For instance, in our model the stock of capital in education affectsskills acquisition according to equation (24). Infrastructure and health capi-tal affect the production process in the private sector as they both combineto produce the stock of government capital, KG:

KG = CES(KINF, KH). (105)

As discussed below, the various channels through which different formsof government investment flows affect the economy figure prominently in ouranalysis of the impact of debt-reduction strategies and expenditure realloca-tion. In particular, we discuss the issue of how, for a given reduction in thestock of foreign loans to the government (and thus lower interest paymentson external public debt), public spending should be reallocated in order toachieve specific poverty reduction goals and growth targets.

The government deficit is financed by either an increase in foreign loans,domestic loans, or domestic credit by the central bank:

DEF = ER L AFLc + ADLG + AXDC. (106)

In general, a variety of financing rules can be specified in IMMPA. Forinstance, it could be assumed that the deficit is financed by domestic loans(through either commercial banks or the central bank) or foreign borrowing,or that it be closed by cuts in government spending. Considering domestic

48

financing alternatives could be useful for analyzing the crowding-out effectsof public spending on private investment, as is done for instance by Agenorand El Aynaoui (2003). In the numerical results reported below, we assumethat the sources of deficit financing are set exogenously and thus that thepublic sector deficit is determined from "below the line." The variable thatadjusts expenditures to make them consistent with the available financingand the level of revenues is the level of lump-sum transfers to households.The choice of a specific financing rule (which is critical for simulations ofmany shocks) is an important aspect of adapting the IMMPA prototype tospecific countries.

The net worth of the government, NWG, is defined as:

NWG = PK(KG + KE) - (DLG + DCG) - ER. FLG, (107)

and evolves over time according to

NWG = NWG,-1 + PK(zAKG + AKE) - (ADLG + ADCG) - ER A

+APK(KG,-1 + KE,-1) - AER. FLG,-1 ,

with the last two terms on the right-hand side representing again valuationeffects associated with changes in the price of capital and the nominal ex-change rate.

Finally, from (94) and (107), the net worth of the consolidated publicsector, NWpS, is given by

NWpS = PK(KG + KE) - DLG + ER. (FF - FLG) - MB. (109)

4 Poverty and Income Distribution Indica-tors

There are several alternative approaches to the analysis of the poverty anddistributional effects of policy and exogenous shocks in applied general equli-librium models. A popular approach in the CGE literature consists in speci-fying a relatively large number of homogeneous household groups and calcu-lating average income for each group following a shock and treating the groupas a whole as being poor if average income is lower than a given poverty line.This is the procedure followed, for instance, by Lofgren (2001), in a study

49

based on a classification of households in fourteen groups, of the impact ofexternal shocks on poverty in Malawi. In IMMPA, the distributional andpoverty effects of shocks are assessed in two ways: first by calculating a setof indicators (for income distribution) based directly on the model's simu-lation results; second, by linking IMMPA simulation results to a householdexpenditure survey.

Specifically, two measures of income distribution are generated directlyfrom IMMPA: the Gini coefficient and the Theil inequality index (see, forinstance Litchfield (1999) or Cowell (1998)).27 Both are based on the sixcategories of households that were identified earlier, that is, workers locatedin the rural traded sector, rural nontraded sector, urban (unskilled) informaleconomy, urban unskilled formal sector, urban skilled formal sector, and cap-italists. Thus, these indicators allow the analyst to study changes in incomedistribution and poverty between groups, under the assumption of completehomogeneity within groups (or representative households). Formally, theyare defined as

Gini =2 2n2 H E E IYHi - YHjl, i, j = AN, AT, UI, UF, S, KAP,i 3

where n = 6 is the number of household categories and YH = YHi/n isthe arithmetic mean level of disposable income for household categories.

The Theil inequality index is measured as

Theil Y ffllog(YHi), i = AN, AT, UI, UF, S, KAP,n YH YH

and other variables are as defined above. We also calculate these two indica-tors using consumption, instead of disposable income.28

27Other commonly-used indicators include the Atkinson index which, like the Gini index,ranges from 0 to 1. For a detailed analytical discussion of the pros and cons of variousmneasures of income inequality, see Cowell (1998). For instance, the Atkinson index issensitive to inequality changes in the lowest part of the income distribution; the Ginicoefficient is sensitive to inequality changes around the median; and the Theil index andcoefficient of variation are both sensitive to inequality changes in the top part of the incomedistribution.

28The initial values from IMMPA are 0.512 and 0.515 for the consumption-based andincome-based Gini coefficients respectively, and 0.199 and 0.203 for the consumption-basedand income-based Theil indexes.

50

Following a shock, IMMPA generates three measures for these indicators(as well as those derived from household surveys, as discussed below): ashort-term measure (first two periods following a shock), a medium-termmeasure (between 3 and 5 periods), and a long-term measure (between 6and 10 periods). 2 9 While somewhat arbitrary in the choice of intervals, theimportance of these measures (which can obviously be calculated only witha dynamic model) is clear: they allow the analyst to identify and discusspossible dynamic trade-offs in the analysis of policy choices, by contrastingtheir short- and longer-run effects on the poor.

To assess the poverty effects of alternative shocks, we link IMMPA toa household income and expenditure survey, such as an Integrated Survey(IS) or Living Standards Measurement Survey (LSMS), which typically col-lect extensive information on migration, household income and expenditure(including own consumption), household assets, credits and savings, levelsof education, apprenticeship and training, employment, occupational char-acteristics and status, as well as geographical location.3 0 The various stepsinvolved in our approach are illustrated in Figure 7. To begin with, we assumethat the available survey consists of a relative large sample. This is neededto reduce non-sampling errors that may cause household income and expen-diture to be underestimated-particularly in the least monetized regions ofpredominantly nontraded agricultural production (see Fofack (2000))-andensure sound inference on poverty effects following changes in factor allo-cation and resource flows across sectors. The approach that we proposeproceeds as follows:

Step 1. The user uses the information provided in the household survey of29Specifically, let xz denote the initial (base period) value of consumption (or income) for

household group h and {gt}jt1= the (discounted) growth rate in consumption (or ijncome)generated by the model for the first 10 years following a shock. The short-run measureof consumption, xSR, is calculated as a geometric average of the period-1 and period-2values of xh, calculated at the average growth rate for the period, xZR(1) and xsR(2):

XSR = xhR(s)XhR( 2 ) = VxO(1 +g9SR) x (l + gsR) 2 , where 9SR = /(1 + 91)(1 + 92)-

1. Thus, xsR = Xo(1 + gsR)' 5 Similarly, it can be shown that the medium-run value ofcornsumption,xmR iS given by XhMR = ( + 9MR)

2 , where 9MR = vft.=3(1 + 9t) -1,whereas the long-run level of consumption, xLR, is given by xLR= x5( + 9LR) 3 with9LR = ft_06(1 + gt) - 1. See Chen et al. (2001) for more details on this procedure.

30 For more details on the scope and content of these surveys, see Delaine et al. (1992)for Integrated Surveys and Grosh and Glewwe (2000) for LSMS surveys. See also Deaton(1997) for a general discussion.

51

its choice (presumably the most recent information available) to classifythe available sample into the six categories of households contained inthe INMMPA framework (using, say, information on the main source ofincome of household heads) so as to establish an interface between themodel's predictions and actual household income and expenses.

Step 2. Following a shock to the model, IMMPA generates real growth ratesin per capita consumption and disposable income for all six categoriesof households in the economy, up to the end of the simulation horizon(say, T periods).

Step 3. These growth rates are applied separately to the per capita (dis-posable) income and consumption expenditure of each household (ineach of the six groups) in the survey. This gives absolute income andconsumption levels for each individual (and averages for each group)following the shock, for T periods.

Step 4. Assuming different initial poverty lines for the rural and urbansectors (expressed in monetary units and adjusted over time to reflectincreases in rural and urban price indexes), and using the new absolutenominal levels of income and consumption for each individual and eachgroup, the model calculates a poverty headcount index, a poverty gapindex, as well as the two indicators of income distribution describedabove (the Gini coefficient and the Theil inequality index). Thesecalculations are performed for the three different horizons identifiedearlier.

Step 5. Compare the post-shock poverty and income distribution indicatorswith the baseline values to assess the impact of the shock on the poor.These comparisons, as indicated above, are based on the assumptionthat the poverty line is constant in real terms in both the rural andurban sectors-an assumption that can obviously be relaxed.

The two poverty indexes that are described in Step 4 are defined as fol-lows. The poverty headcount index is the ratio of the number of individualsin the group whose income is below the poverty line to the total number of

52

individuals in that group.31 The poverty gap index is defined as:

1 nPG = YH. 1: (YH - YHk),

Ak=1

where k is an individual whose income is below the poverty line, n is the totalnumber of people in the group below the poverty line, YHk is the income ofindividual k, and YH* is the poverty line.

In practice, of course, whether income or expenditure data should be pre-ferred depends on the scope and quality of the data in the available householdsurvey. In the illustrative simulations reported below, we used instead of anactual household income and expenditure survey a "fictitious" one, that webuilt as follows. First, we calculated real per capita disposable income andconsumption expenditure for each of our six household categories, using theinitial values that are provided to solve the model numerically. Second, usinga random number generator and a log-normal distribution, we produced asample of 1,397 observations (corresponding to the total number of workersand capitalists in the initial data set). We thus considered each individualworker and capitalist to represent one household. Third, we used the ini-tial per capita income and consumption data as mean values and imposed astandard error of 0.5 for all household categories, except for skilled workersand capitalists, for which we chose a standard error of 0.35. We set (some-what arbitrarily) the income poverty line for the rural sector at 0.45 and0.5175 for the urban sector (or 15% higher than the rural poverty line). Forthe consumption data, we used initial poverty lines of 0.4 and 0.46 for therural and urban sectors (again, with the latter being 15% higher than theformer). We assume for simplicity that these poverty lines remain constantin real terms for the whole horizon of the simulation period that we considerbelow (10 periods). Figures 8 and 9 show a log-normal approximation tothe initial data on income and consumption that we generated for each of

31As is well known, the headcount index suffers from several limitations (see for instanceBlackwood and Lynch (1994) and Ravallion (1994)). In particular, it does not indicatehow poor the poor really are-it remains unchanged even if all people with incomes belowthe poverty line were to experience, say, a 50 percent drop in income. Put differently, whena poor person becomes poorer, the index does not increase. Moreover, it implies that thedistribution of income among the poor is homogeneous (it does not distinguish betweena poor person who earns one monetary unit less than the poverty line and a poor personwho earns 100 monetary units less than the poverty line). But to the extent that theanalyst is interested only in the number of poor, the headcount index is a useful measure.

53

our six categories of households. This parametric approximation is of coursevery good, given that the artificial samples that we generated were based onthe log-normal distribution itself. More generally, a number of alternativestatistical distributions-such as the beta distribution-can be specified inIMMPA to approximate the actual distribution of any given income group(see Agenor and Grimm (2003)).

These results produced an income-based headcount index of 61.8 % in theagricultural sector (86.1% in the nontraded sector and 37.2% in the tradedsector), 52.8% for urban unskilled households (78.8% in the informal sectorand 14.0% for unskilled workers in the formal sector), and 0% for skilled work-ers and capitalists. For the economy as a whole, the income-based povertyrate amounted to 53.2%. With the income-based poverty gap, the resultsobtained were 38.9 % for the agricultural sector (44.0% in the nontradedsector and 26.8% in the traded sector), 40.8% for urban unskilled households(43.8% in the informal sector and 15.7% for unskilled workers in the formalsector) and again 0% for skilled workers and capitalists.3 2 The aggregatepoverty gap index reached 38.7%. For the consumption-based headcount in-dex, the results are 64.4 % in the agricultural sector (88.7% in the nontradedsector and 39.8% in the traded sector), 62.3% for urban unskilled households(91.8% in the informal sector and 18.4% for unskilled workers in the formalsector), and 0% for skilled workers and capitalists. For the economy as awhole, the consumption-based poverty rate amounted to 56.9%. With theconsumption-based poverty gap, we obtained 40.6 % for the agricultural sec-tor (47.0% in the nontraded sector and 26.1% in the traded sector), 46.3%for urban unskilled households (50.0% in the informal sector and 18.9% forunskilled workers in the formal sector) and again 0% for skilled workers andcapitalists. The aggregate poverty gap index reached 41.4%. These numbersare broadly in line with the evidence for sub-Saharan Africa.

For each policy or exogenous shock, therefore, the user can assess theshort-, medium-, and long-term effects on poverty and income distributionand thus examine possible trade-offs with other policy objectives. The mainbenefit of this approach is that it allows us to link IMMPA simulation resultsdirectly with actual patterns of income and expenditure and to provide a

32For unemployed skilled workers (who do not receive wage income but are assumedto receive interest income on financial assets as well as government transfers and a shareof distributed profits by firms in the private sector), the average income is close to thepoverty line in the base period. For simplicity, we treat the group of skilled workers as ahomogeneous one and generate a distribution in which no skilled worker is poor.

54

more accurate derivation of poverty indicators. However, as noted above, itassumes that intra-group distribution is constant. Put differently, the within-group homogeneity assumption implies that the within-group rank orderingof households and individuals remains unchanged following any shock. Asnoted by Kanbur (1987) and Demery and Addison (1993) in a related con-text, the assumption that within-group distributions are unchanged and un-affected by policy shocks implies that workers are withdrawn from the sectorof origin in a representative manner (leaving the distribution of income thereunchanged) and that, as they move from one sector to another, they assumethe income distribution characteristics of the sector of destination (in par-ticular, the variance of income in that sector is assumed to apply to all newentrants).33 Thus, some workers may be poor not because of their personalcharacteristics but because of the economic circumstances that characterizetheir sector of employment. Whether the assumption of constant within-group distributions is always warranted is not entirely clear; it representstherefore a potential weakness in this approach.

The procedure described above assumes that the modeler matches house-holds as defined in the macro component of IMMPA and a household surveyusing information on the main source of income of household heads. Analternative treatment is also possible and depends on whether or not thehousehold survey provides sufficient detail regarding the composition of in-come among individual members of each household; "light surveys" tend toconcentrate on the household head, whereas more in-depth surveys providericher information. To the extent that the information is detailed enough,and as long as each member of a household can be "allocated" to one ofthe six income groups identified earlier, growth rates of income and con-sumption can be applied separately to each individual income earner (asin Step 3 above). A more accurate measure of the change in each house-hold's income can therefore be calculated and poverty and income indicatorscan be generated using either "individual" income earners or "composite"households. However, whether accounting for heterogeneity in the sourcesof income among individual household members makes a difference or not isgenerally case specific; it depends on both the characteristics of the intra-household distribution of income (which depends on each household's risk

33It also implies that income transfers between households in any given group are ig-nored. In practice, intra-group income reallocation may be large in periods of hardshipand may represent an important factor in understanding the poverty effects of adverseeconomic shocks.

55

diversification strategies) and the extent to which the growth rates of incomeand consurnption generated by the macro component of IMMPA following agiven shock differ among the various income groups on which it is based. If,for instance, the intra-household distribution as given in the survey is suchthat most of the income of each composite unit is generated by the householdhead, treating the household as a homogeneous unit and applying the samegrowth rate of income to each member should not result in significant errors.

Several recent studies have attempted to drop the assumption of a stablewithin-group distribution to analyze the poverty and distributional effects ofpolicy and exogenous shocks in applied general equilibrium models; they in-clude Cockburn (2001), Decaluwe, Dumont and Savard (1999), and Decaluwe,Patry, Savard, and Thorbecke (1999). Individual data in these studies areincluded directly in the general equilibrium model and (assuming that thewithin-group distribution follows a well-defined statistical distribution, suchas the lognormal or a beta distribution) micro-simulation techniques are usedto exploit intra-group information. This approach has the benefit of allowingthe analyst to distinguish, in the evolution of poverty indicators, the specificcontribution of three factors: changes in the poverty line (when it is treatedas endogenous), average income variations, and income distribution. It pro-vides therefore a potentially important direction for future research.3 4 Atthe same time, however, it must be recognized that it is relatively complexto implement (particularly in conjunction with a fully specified macroeco-nomic model, like ours), because it requires manipulating a sizable amountof data. Moreover, whether changes in the intra-group distribution mattera lot appears to be shock dependent; indeed, in some of the simulation re-sults reported by Decaluwe, Dumont and Savard (1999), such changes onlyaccount for a small proportion in changes in poverty measures. A possi-ble way of testing for whether this type of techniques should be used is tocompare the aggregated results of two (or more, where feasible) householdsurveys and test statistically (using parametric or non-parametric tests) forany evidence of a shift in intra-group distributions. This test is not, how-ever, fully satisfactory, because evidence of stability across surveys cannotnecessarily be construed as providing definite support to any model-basedexperiment-particularly those involving "atypical" shocks.

It is also worth noting that in the artificial survey that we constructed,34Another approach, based also on micro-simulation techniques, is pursued by Robil-

liard, Alatas, Bourguignon, and Robinson (2001).

56

we did not account for openly unemployed workers, of either variety. In prac-tice, however, surveys may report a head of household as being unemployed(and therefore with no declared wage income), while at the same time receiv-ing non-wage income from, say, holdings of financial assets and governmenttransfers. Instead of simply assuming that the rate of growth of income orconsumption is zero for the unemployed (an assumption that may lead tounrealistic results for medium-term exercises), our inclination would be totreat these observations as follows. For unskilled workers, it would seemreasonable to assume that the openly unemployed are actually employed inthe informal sector or the nontradable agricultural sector (even if they don'tdeclare it), depending on the sector of occupation, at the going wage. Onecan thus apply the growth rates of consumption and "full" disposable incometaken from the macro component of IMMPA for that category of workers,as in the sequence described previously. By contrast, for skilled workers, theassumption that the unemployed are actually working in the informal sectormay not be very satisfactory, for the reasons discussed earlier. We wouldsuggest using the growth rate of pre-tax, non-wage income only, for that cat-egory of workers, and adjust the rate of growth of consumption taken fromthe macro component of IMMPA in proportion to the differential betweendisposable income of employed workers and pre-tax non-wage income.

Finally, it should be noted that we have abstracted in the above discussionfrom issues associated with differences between national accounts data (onwhich IMMPA is based) and survey data (from which poverty measures arecalculated). However, in practice, large discrepancies can arise between thesetwo sets of data. In particular, it is possible that the composition of employ-ment, output, and the inter-group income distribution generated by IMMPAand the household survey (following step 1 above) are different.3 5 Indeed,it is well recognized that national accounts and survey estimates of incomeand consumption patterns can differ significantly. Moreover, these differencesmay even be increasing in some cases, as appears to be the case in India (seeDeaton (2001)). In general, the evidence suggests that nominal consumptiongrowth rates estimated from survey data tends to be substantially lower than

35As indicated earlier, for the simulation exercises reported in this paper, we generatedan artificial sample (based on a log-normal distribution) using the mean per capita realincome data generated by the initial calibration of the model. The inter-group Gini coef-ficients that are calculated from the model and from the household survey are thus veryclose, given the relatively large number of replications. In practice, however, this may ormay not be the case.

57

those estimated from national accounts data. Various factors may accountfor these discrepancies, as discussed by Deaton (2001) and Ravallion (2000).For instance, consumption in national accounts is typically determined asa residual and is thus contaminated by errors and omissions elsewhere inthe accounts. In practice, researchers often end up treating one source orthe other as the "correct" or "most reliable" one-despite the fact that it islikely that both sources of information are subject to error. In the presentcontext, because both sources are used jointly, the issue of reconciling themarises. For instance, if it is believed that the national accounts data providean accurate measure of the level of consumption, one approach could be toscale up. survey data so as to match the former, and use the rescaled datafor poverty assessment. There are several potential problems, of course, withthis approach-the assumption that household consumption levels as mea-sured in the survey are correct up to a multiplicative constant is by no meansa reliable one, given the likely discrepancies between urban and rural data.More generally, the decision as to which data are correct is a difficult one,and the reconciliation process is likely to be country specific.

5 Calibration

Assessing the properties of the model presented in the previous section re-quires calibration and numerical simulations. Given that the objective ofthe model is to analyze the poverty impact of adjustment policies in highly-indebted, low-income countries, we have calibrated it to reflect what we be-lieve to be a "representative" country. While parameters and initial valuesfor each of the variables in the model are provided in Appendix C, we alsoprovide a brief summary for the key variables below. Many of these para-meters (such as demand and supply elasticities) reflect conventional values.used in the literature. A financial social accounting matrix (FSAM), follow-ing the lines of Thorbecke et al. (1992), Easterly (1990), Rosensweig andTaylor (1990), and Thissen (1999, 2000), presents all initial values in a morepedagogical format.

We now provide a brief overview of the calibration of initial values andparameters. A detailed analysis is provided in Chen et al. (2001).

58

5.1 Inlitial Values

5.1.1 Endogenous Variables

For the first period, we assume that our "representative" economy has anominal GDP of approximately 850 units and we set agricultural productionto be about 30 percent of GDP, with slightly more than four-fifths of it beingtraded and the remainder being nontraded. We further assume that theprivate, formal urban sector produces 15 percent, the informal sector about43 percent, and the public sector 12 percent of the country's total output.

The size of the total workforce in the first period has been set to 1416, with829 persons residing in the rural area, about 40 percent of which are employedin the traded agricultural sector. There are about 467 urban unskilled work-ers, eighty percent of which are employed in the informal economy. Thereare 121 skilled workers, with 46 of them being employed in the private urbanformal sector, and another 50 are working in the public sector. This impliesthat the initial open unemployment rate for skilled workers is about 20 per-cent. The reproductive growth rate of urban unskilled individuals has beenset to 2.2 percent in the initial period.

The skilled nominal wage rate has been set to approximately 30 percentabove that of the binding minimum wage paid to unskilled workers employedin the formal-urban economy. Rural workers employed in the traded agricul-tural sector receive about 35 percent of the urban minimum wage.

Aggregate demand consists of households' consumption, private invest-ment and government expenditures. Demand for the formal, private urbangood is approximately 600 units, whereas demand for the nontraded agri-cultural good is 190 units. Demand for the public good is 286 units, anddemand for informal goods is 152 units.

The level of agricultural goods exported is 371 units and imports of non-traded agricultural goods have been assumed to be 75 units in the first period.Domestic demand for nontraded agricultural goods is set at 123 units. Im-ports of the formal private good are assumed to be 486 units. Also, we haveassumed exports and domestic demand of the private formal good to be 215units and 114 units, respectively.

First-period values for household incomes range from a value of 61 for thenontraded agricultural sector household to 217 for the capitalist household.Private urban formal firms made the highest profits with a profit level of 179.The nontraded agricultural fi,ns were assumed to have the lowest profit levels

59

of about 30. Incomes of firms mirror that of profits with private urban firmsearning the highest income (168 units), and nontraded agricultural firms thelowest income (30 units).

The return to capital is initially set to 12 percent, whereas the lendingrate is set to 7 percent, with a financial premium of 1.07. The amount ofdomestic loans to private urban firms is assumed to be 154, whereas bankshave foreign liabilities of 62. The stock of private capital has been set to 1500,whereas the initial capital stock in infrastructure has been set to 500, andthat for education and health have been set to 50 each. The initial stock ofpublic capital has also been set at 50 units. With a reserve requirement ratioof 10 percent, and total bank deposits of 213, the level of required reserves is21. The money supply is initially at 152. The initial values of tax revenue,transfers to households, the budget deficit and total government expenditureas shares of GDP are 23 percent, 15 percent, 1.3 percent and 37 percent,respectively.

5.1.2 Exogenous Variables

Both the initial numbers of skilled and unskilled workers in the public sec-tor have been set to 50 persons and grow at a constant rate of 2 percentannually. The real minimum wage rate of urban workers and the real wagerate paid to workers in the rural traded sector are kept constant throughout.The reproductive growth rate of rural workers has been set to 2 percent perannum.

Constant income tax rates of 15 percent, 23 percent and 33 percent arebeing levied on unskilled (both agricultural and urban), skilled and capital-ists. There are no tariffs or export subsidies for agricultural goods. Privateurban goods have a 4 percent import tariff imposed but also enjoy a 5 percentexport subsidy.

Foreign loans to private firms amount initially to 136.6 and grow at 9percent in every period. The initial values for investment in infrastructure,health and education have'been set to 10 and each grows at a fixed rate of2 percent per period. We have assumed that depreciation for infrastructurecapital occurs at 1 percent per annum.

Government consumption has a initial value of 40 and grows at 2 percentthereafter. Domestic credit to government has an initial value of 50 andgrows also at an annual rate of 2 percent thereafter. Domestic loans to thegovernment has an initial value of 100 and is kept constant, whereas foreign

60

loans to the government have a starting value of 1567 and grow at 1 percentannually. Indirect tax rates on output have been all set to zero.

5.2 Parameter Values

Recall that all sectors, with the exception of the informal sector, have CESproduction functions. Elasticities of substitution between the various inputsin these production functions have been calibrated to values between 0.4 and1.2, reflecting low to medium values.

The input share parameters have been chosen to reflect labor-intensiveproduction technologies typical of a developing country. In the agriculturalsector, the labor share parameter is 0.92 for the traded good and 0.63 for thenontraded good. Similarly, the labor (versus public capital) share parameterin the public good production function, I3 xG, is also relatively high at 0.90.Production of the private good not only leans heavily toward usage of labor,with the skilled labor-private capital composite input share parameter being0.90, but also uses more unskilled labor rather than skilled labor or capital,with the unskilled labor share parameter having a value of 0.97. The re-productive growth rate of the urban unskilled workers has an elasticity torelative expected wages of 0.3.

Recall that only the agricultural goods and private formal goods can betraded. The demand functions for these goods are therefore expressed asArmington functions over import and domestic demand. The values for theshare parameter (for import demand) and elasticity of substitution for thedemand for agricultural goods are 0.5 and 0.8, respectively, whereas thosefor the demand for private formal goods are 0.8 and 1.01, respectively. Forall households, the largest consumption share is on goods produced by thesectors in which they are employed. With the exception of the urban informalhouseholds, for all households the smallest share of consumption went toinformal goods.

For the construction of the consumer price index, private urban goodswere assigned the largest share whereas in the construction of the rural priceindex, agricultural goods had the largest weight. For the urban unskilledprice index, informal goods had the largest share. Lastly, for the construc-tion of the urban skilled price index, private urban goods were weightedmost heavily. Money demand elasticities to the domestic and foreign interestrates have all been set at 0.5, whereas the elasticity to real income has beenassigned a uniform value of 1.0.

61

6 Some Illustrative Experiments

This section presents and discusses the numerical results associated withthree types of shocks: a temporary terms-of-trade shock, a permanent cutin domestic credit to the government, and a poverty reduction program con-sisting of partial external debt forgiveness coupled with a reallocation ofsavings on debt service payments to three alternative forms of governmentexpenditure: lump-sum transfers to households, spending on infrastructure,and outlays on education. 3 6 Both the short- and longer-run effects of theseshocks are analyzed, with a particular focus on poverty as measured by theindicators described above. 37

6.1 Terms-of-Trade Shock

We first simulate the impact of a temporary terms-of-trade shock that takesthe form of a transitory (one period only) 10 percent increase in the worldmarket price of the agricultural traded good (see Tables 2 and 3). Thistype of shocks has indeed played a pervasive role in explaining changes inreal incomes in sub-Saharan Africa and has been analyzed in a number ofempirical studies (see, most recently, Dorosh and Sahn (2000)). Because wehave assumed that product wages of unskilled agricultural traded workersare fixed (that is, the ratio of the nominal wage to the value added price ofthe good is constant), nominal wages will tend to match the increase in theproducer price.3 8 Had we excluded financial sector effects from the model,the demand for labor would remain constant and so would production of thetraded good. But because we assume that firms in this sector borrow tofinance wage payments, the effective price of labor includes the loan rate, asnoted earlier. As we shall explain later, interest rates experience an initialdecline after the positive terms-of-trade shock, implying that the effectiveprice of labor goes down. In itself, this tends to increase the demand forunskilled labor. However, a decline in the real value of public investment,

3 6For a description of the IMMPA simulation program, which consists of both Eviewsand GAMS versions combined with Ecel input and output sheets, see Chen et al. (2001).

371t should be kept in mind that the choice of a particular set of poverty measuresalways involves a value judgement and can have considerable bearing on simulation resultsand policy choices. The open architecture of IMMPA, however, allows users to programalternative measures if deemed necessary.

38 They do not match the increase in price of the agricultural good one to one becauseproducer prices also reflect changes in intermediate input prices.

62

due to increasing formal sector goods prices, leads to an overall reduction invalue added and output in this sector, and lowers unskilled labor demand.Given that at any point in time the total supply of unskilled labor in theagricultural sector is predetermined and that this segment of the labor marketmust clear at the aggregate level, workers laid off from the agricultural tradedsector are absorbed by the nontraded sector. But because of the decline inpublic investment, output also declines in the nontraded agricultural sector.All these dynamics occur in the period following that of the shock, given ourassumption that the cost of credit in loan contracts is agreed upon one periodin advance and therefore interest rate effects take one period to materialize.At the time of the shock though, interest rate effects are absent and as a resultvalue added in both rural sectors remains constant, implying no changes inthe rural traded and rural nontraded workforce. This occurs because productwages in both sectors remain constant, implying that nominal wages matchthe increase in value added price of their respective good.

The increase in wages in the agricultural sector yields higher income andhigher consumption for households in that sector; this raises aggregate de-mand and put upward pressure on domestic prices. A strong price increaseactually leads to a switch in the demand for formal sector goods, away fromdomestic production and toward imported goods. Production in the privateformal sector therefore decreases in response to the expansion in spending.The supply response is partly brought about by the decline in the public cap-ital stock and partly by price-driven increases in the skilled (product) wage,which lowers the demand for skilled labor. As a result, unemployment ofskilled workers increases. The decline in skilled employment, in turn, lowersthe marginal product of unskilled labor in the private formal sector, loweringdemand for unskilled labor as well. Because a constant minimum real wageprevails for unskilled formal workers, there is a shift in the supply of unskilledlabor from the formal into the informal economy, until the marginal productof labor matches the minimum wage. As a consequence, the product wagefalls in that sector and the size of the informal economy expands.

We next focus on the effects of this shock on migration between rural andurban areas. The major determinant of this decision is the ratio of averageexpected rural consumption wages to average expected urban consumptionwages.39 In the case at hand, the increase in the price of the agricultural

39As defined earlier, consumption wages are defined as nominal wages deflated by theprice index corresponding to the consumption basket of workers in a particular sector.

63

traded good raises the average rural wage by more than the average urbanwage, providing incentives to remain in the agricultural sector. This bringsmigration down after the shock, but the situation is gradually reversed as theeffect of the shock dies out. In a similar fashion, skills acquisition dependson relative consumption wages of urban unskilled versus skilled workers. Thefact that the expected consumption wage of skilled workers goes up whereasthat of unskilled workers goes down leads to an increase in the rate of skillsacquisition and an increase in unemployment. This increase is graduallyreversed as the effect of the shock fades away.

Higher incomes and aggregate demand lead to a rise in tax collection. Be-cause the public sector deficit is determined by its sources of financing (whichare taken as exogenous), and the selected closure rule implies that additionalrevenue is devoted to an increase in government transfers to households, theinitial positive impact of the terms-of-trade shock is reinforced by an expan-sionary fiscal policy. But there is also an additional effect of public sectorfinances: higher taxes relative to the existing debt service are interpretedby investors as a reduction in confiscation risk, resulting thereby in a risein private investment. This in turn generates an increase in the demand forloans by firms which, in itself, puts upward pressure on the domestic banklending rate (as noted earlier, banks charge a premium over the cost of funds,which depends inversely on the effective value of collateral relative to the sizeof loans). Nevertheless, because the price of capital jumps up by more thanthe increase in loans, the premium goes down and so does the lending rateon impact.

On the financial side, money holdings for all households are higher onimpact, and eventually go back to their original levels as the effects of theshock vanish. Although savings across households increase as a consequenceof higher incomes, they do not rise proportionally to income because sav-ings rates are smaller on impact due to the temporary increase in inflation.Given that the allocation of savings between domestic and foreign depositsis made after choosing the desired level of money holdings, and that the lat-ter increases at a higher rate than total financial savings, then holdings ofdomestic and foreign deposits fall on impact, but quickly increase relative tothe base scenario once inflation drops and savings rates go up again.

Higher export prices for the agricultural good, coupled with the increasein foreign loans from domestic banks to finance the higher level of privateinvestment, more than offset the increase in imports. As a result, the centralbank accumulates foreign reserves and this in turn leads to an expansion of

64

money supply. These variables also return close to their baseline values asthe effects of the shock fade away.

We now turn to poverty and income distribution indices based on our fic-titious household survey data. The impact and short-run effects of the terms-of-trade shock generally result in reductions in consumption-based povertyheadcount indices. There is a decrease in the poverty headcount index of theagricultural nontraded sector, but the largest decrease occurs in the agricul-tural traded sector. The decrease in the latter index is expected, given the10 percent increase in the world price of the exported agricultural good andthe existence of fixed real wages in that sector. The decrease in poverty inthe rural nontraded sector is mainly due to higher wages, because increasingnontraded agricultural goods demand leads to increasing labor demand inthat sector. Because there are only smaller changes in consumption povertyheadcount indices for the remaining sectors, there is a negative net effecton the economy-wide consumption poverty index in the short run. In thelong run, however, there is only a slight decrease in the poverty index for theagricultural traded and nontraded sectors (given the size and temporary na-ture of the shock), which leads to a slight decrease in the consumption-basedpoverty headcount indices for the rural sector and the economy as a whole.Poverty headcount indices based on disposable income depict similar results:there is a relatively strong decline in the economy-wide poverty index in theshort run, and a small decrease in the long run.

Consumption-based poverty gap indices indicate that poverty decreasesin the agricultural traded and nontraded sectors and in the urban informalsectors in the short run. Given these decreases, the economy-wide povertygap index also decreases despite an increase in poverty in the formal unskilledlabor group. In the long run, there is also a decrease in the poverty gap index,although its magnitude is much smaller. A similar picture emerges whenincome-based poverty gap indices are used instead: the economy experiencesa decrease in poverty in the short run, which reverses itself in the long run.

In summary, a temporary terms-of-trade shock has relatively unambigu-ous effects on poverty. Both headcount and poverty gap indices indicate thatdeclines in poverty rates in the short run are reversed in the longer run.

With regard to income distribution indicators, the Gini coefficients gen-erally indicate short and long run declines in income inequality, whereas theTheil indices shows short- and long-run increases in income inequality. Long-run effects are generally much smaller than short-run effects. Again, this isnot surprising, given the temporary nature of the shock.

65

6.2 Cut in Domestic Credit to Government

We next exploit the financial nature of the model to track the impact of apermanent, 30 percent cut in the level of domestic credit by the central bankto the non-financial public sector, keeping its growth rate constant after thefirst period (see Tables 4 and 5). This has two effects on impact. First, thefall in credit reduces the financing available for the public sector, thereforerequiring a reduction in the deficit "above the line"; and second, there is a fallin the monetary base (and consequently the money supply), which createsdeflationary pressures.

The reduction in the deficit is accomplished by a proportional cut in to-tal lump-sum transfers to households.40 This lowers households' income andconsumption of all goods.41 Lower government revenue relative to the exist-ing level of debt service payments increases the perceived risk of confiscationon impact and leads to a fall in both investment and the demand for domesticloans. The drop in firms' loans is smaller than the initial fall in the priceof collateral, providing incentives for banks to increase the premium on thecost of funds, thereby increasing the loan rate. This has a negative effecton output in production sectors that rely on bank credit to finance work-ing capital needs (wage payments here), namely the agricultural traded andurban private formal sectors, because the effective price of labor increasesafter the initial jump in the loan rate (given our assumption that the costof credit in loan contracts is agreed upon one period in advance). Neverthe-less, production expands in both the traded agricultural and urban privateformal sectors. A lower price of formal sector goods leads to an increase inreal investment by the government, and the expansion of public capital dom-inates the increase in working capital costs in both sectors. In addition, theproduct wage of skilled labor declines, given that the fall in the price of thecomposite factor is much smaller than that of nominal skilled wages. Thelatter effect dominates over the interest rate effect, leading to an increasein demand for formal skilled labor, and an increase in output in the privateformal sector. The urban unskilled workforce also increases, because the in-crease in the (expected) urban consumption wage is larger than the increasein the (expected) agricultural consumption wage.

In spite of higher interest rates, the increase in the public capital stock4 0Transfers could be negative, in which case they can be interpreted as non-distortionary

taxes.4'With the exception of the informal good.

66

leads to a higher demand of unskilled labor in the rural traded sector, and atransfer of unskilled workers from the rural nontraded into the rural tradedsector. Notwithstanding the decline in labor supply in the nontraded agri-cultural sector, output in that sector expands due to the increase in publiccapital. In subsequent periods, wage increases lead to increases in the prod-uct wage. But because households in the rural nontraded sector consumea bundle of goods whose price index increases more than nominal wages,consumption wages fall for workers employed there.

The contraction in aggregate demand, together with the increase in out-put in the urban private formal sector, leads to a strong decline in importsand a smaller expansion in exports; the improvement in the trade balanceleads to higher official reserves, which tend to mitigate the impact of the orig-inal reduction in domestic credit to the government on domestic liquidity. Insubsequent periods, the fall in both private investment and the associateddemand for domestic loans reduces the need for external financing of banks,thereby reducing the rate of accumulation of foreign reserves. The initialfall in money demand leads to a reallocation of financial assets, namely, anincrease in holdings of both domestic and foreign deposits.4 2 The increase indomestic deposits lessens the need for external financing of domestic banks,whereas the increase in foreign deposits abroad by domestic households gen-erates a capital outflow. Both factors put downward pressure on foreignreserves and the expansion of money supply. The final outcome of all theseforces is a fall in money supply, which has deflationary effects that are con-sistent with the observed initial fall in the price level.

Because we assume that domestic credit grows at the same rate as in thebase scenario beyond the initial period, most variables converge to their pre-shock levels in the long run. The reason for a relatively rapid convergenceis that the initial value of domestic credit to the government as a share ofGDP is relatively small, implying that (given our assumption of a constantdeficit) the initial level cut does not alter significantly the path of domesticcredit for government budget financing purposes in future periods.

Consumption per capita falls temporarily for all categories of workers. Onthe one hand, consumption wages of unskilled workers in the rural traded sec-tor as well as in the urban formal sector increase, whereas those of unskilled

42Even though savings fall as a consequence of lower income, the reduction in the infla-tion rate increases savings rates on impact. This, coupled with the fall in money demand,lead to higher resources available for investment into other financial assets (domestic andforeign deposits).

67

workers in the rural nontraded and urban informal sectors and skilled work-ers in the formal sector decrease. On the other, all households face lowertransfer receipts from the government, mainly as a direct consequence of thedrop in government domestic credit, but also because of lower tax revenuesresulting from the fall in aggregate demand. This impact of lower transferreceipts counteracts any wage gains, leading to falls in household income andlower consumption levels.

Consumption-based poverty headcount indices show that poverty de-creases only in the formal unskilled labor group in the short run. Never-theless, poverty is unchanged for all groups in the long run. Turning toincome-based poverty headcount indices, we observe that poverty decreasesin the formal unskilled labor group in both the short and the long run. Theeconomy-wide income-based poverty headcount index increases less than itsconsumption-based counterpart in the short run, whereas both indices areunchanged in the long run. Poverty gap indices, both consumption andincome-based, show that poverty at the aggregate level increases both in theshort and the long run. However, they also indicate that the increase inpoverty is greater in the short run than in the long run. Hence, all fourpoverty indices show that poverty changes, as a result of the domestic creditshock, are relatively small in the long run.

Both consumption- and income-based Gini coefficients indicate that thedomestic credit shock has a negative effect on income distribution, both in theshort and the long run. However, Theil indices generally imply that incomedistribution becomes more equal with the domestic credit shock. Thus, theeffect of the shock on income distribution is ambiguous.

6.3 Debt Reduction and Expenditure Reallocation

An important policy experiment for highly-indebted low-income countriesinvolves debt relief and fiscal adjustment. We analyze three different deficit-neutral scenarios that differ in the allocation of the savings resulting from apermanent, 5 percent reduction in the stock of public external debt.43 Thefirst scenario (which we take as our benchmark) corresponds to the case inwhich savings are allocated to an increase in lump-sum transfers to house-holds, in proportion to their initial shares. The next two scenarios focus

4 3 Because we assume that contracted foreign public sector debt has infinite maturity,debt service consists of interest payments only.

68

on the allocation of savings to investment in education and investment in in-frastructure, respectively.44 One of the attractive features of our model is thatit also permits the analysis of the effects of mixed policies-a combination,for instance, of investment in education with investment in infrastructure-aparticularly relevant consideration for policymakers who must determine theallocation of public expenditure. We contrast the effects of a mixed policypackage with those of "pure" policies that imply full allocation to just onetype of investment, but for brevity we do not go over a detailed account ofthis additional simulation.

Throughout the discussion we assume that the non-financial public sec-tor deficit remains constant at base scenario levels. For many developingcountries, the initial position may be one in which the fiscal deficit is unsus-tainable and creating undue pressure on inflation. In such conditions, thereis a good case for using savings from debt reduction to bring the deficit downto sustainable levels, if increases in taxation are not feasible. We abstract,however, from these considerations and examine the effects of alternativeways of spending the income saved from debt reduction, because our interestlies in understanding the effects of alternative strategies for expenditure allo-cation. We implicitly assume that the starting fiscal position is sustainable,be it because of continuing foreign aid or proper fiscal management, althoughthis can be easily modified (by considering alternative sources of financing)to consider jointly the case of debt reduction coupled with a cut in the fiscaldeficit.

6.3.1 Transfers to Households

We first consider the case in which the savings (current and future) associatedwith debt reduction are rebated to all households in the form of lump-sumtransfers (see Tables 6 and 7). Specifically, we assume that the governmentallocates these transfers according to initial household shares.45 This policyhas immediate demand-side effects: the positive impact on households' in-comes leads to an increase in some components of consumption and money

44As noted earlier, the model also incorporates investment in health; but because itseffects are similar to investment in infrastructure (that is, both types of investment increaselabor productivity) they are not reported separately.

45This implies that as long as groups differ in size, transfers per capita are not the sameacross groups. An alternative scenario would be to assume that only poor householdsreceive transfers.

69

holdings. 4 6 At the same time, the reduction in the stock of external debtreduces interest payments, and the increase in income yields higher govern-ment revenues. These two effects have a positive impact on private capitalformation, because both contribute to a reduction in confiscation risk. Higherprivate investment implies a higher demand for domestic loans to firms, whichin turn leads to an increase in the premium charged by domestic banks4 7 anda proportional increase in the loan rate.

The higher lending rate increases the effective price of labor in the agri-cultural traded sector, thereby reducing labor demand and output in thatsector. Laid-off workers are absorbed in the agricultural nontraded sector,leading to an increase in output and a reduction in the product wage inthat sector. By contrast, in the formal urban sector, the rise in investmentincreases the private sector capital stock; because skilled labor is (to somedegree at least) a substitute to private physical capital, this implies that thedemand for skilled labor goes down, resulting in an initial decrease in out-put of the formal private good. However, the subsequent switch of unskilledworkers from the informal economy to the formal sector leads to a rise in out-put in the private formal sector. At the same time, it reduces output in theinformal sector, but increases both the product wage and the consumptionwage in the informal economy.

The resulting smaller differential between average expected urban con-sumption wages and rural wages eventually leads to a dampening of migra-tion flows into urban areas, which tend to support the initial increase ininformal sector wages. There is also an increase in the expected skilled con-sumption wage and a fall in the expected unskilled consumption wage (due,in the case of the latter, to the increase in the price index of the unskilledconsumption basket). This leads to a higher rate of skills acquisition andincreased unemployment relative to the baseline scenario. In terms of realper capita consumption, all working households benefit from the increase intransfers, except the urban unskilled labor group.

Poverty, as measured by the economy-wide consumption-based povertyheadcount index, decreases in the short run when savings from lower debtservice are applied to transfers to households. The reduction in povertybecomes somewhat larger in the medium run, but is reduced in the long run.

46Specifically, consumption of the agricultural traded and private formal sector goodsincreases, whereas consumption of informal and public sector goods decreases.

47See our discussion of the terms-of-trade shock section for a more detailed explanationof this effect on the premium.

70

In both the short and the long run, poverty reduction only occurs in therural and urban informal sectors.48 Short- and long-run poverty reductionis somewhat larger when using income-based poverty measures, but remainsconfined to the rural sectors.

When using consumption-based poverty gap indices as indicators, povertyagain falls in the short and the long run, with a larger decrease at a longerhorizon. However, in contrast to headcount indices, poverty reduction hereoccurs across all unskilled labor groups in the long run. In the case ofincome-based poverty gap measures, economy-wide poverty decreases in theshort and long run, confirming results obtained from the consumption-basedpoverty gap measure. Among the four indices of income distribution, the Ginicoefficients show that debt reduction reduces income inequality, whereas theTheil indices shows the opposite result.

6.3.2 Investment in Infrastructure

The use of savings from debt reduction to increase the stock of capital ininfrastructure has not only demand-side effects but also supply-side effects(see Tables 8 and 9). We focus on the latter, because demand-side effectsare similar to those already described above. In particular, higher infrastruc-ture provides a boost to production in the rural traded, rural nontraded, andprivate formal sectors. It increases the marginal product of all factors of pro-duction in these sectors, given our assumption that infrastructure facilitatesa more efficient use of available resources. Therefore demand for labor goesup, as well as private investment. But there is also an additional channelthat contributes to the rise in capital formation, and that is the reduction inconfiscation risk resulting from the long-run increase in tax revenue, pairedwith a cut in interest payments following the reduction in external debt.

These two channels of transmission have a compounded effect on outputin the private formal sector, which increases strongly when compared to thecase in which savings from debt relief are used to finance lump-sum transfers.Skilled labor unemployment is reduced in contrast to the transfers scenario.The behavior of the rural sector is quite different as well: increased pro-ductivity both in the agricultural traded and nontraded sectors leads to anexpansion in total agricultural output. Similar results to those of the trans-fers policy apply to consumption per capita levels among working households,

48This is mainly due to.the fact that, in our base-period calibration, the share of totaltransfers allocated to rural households is assumed to be higher.

71

except for agricultural traded households who experience declining consump-tion levels.

Consumption-based poverty headcount indices show that debt reductionhas no effect on poverty in the short run, but it does have some positive effectsin the long run. The latter is mainly due to poverty reduction in the ruralnontraded sector. Similarly, income-based poverty headcount indices showthat poverty is reduced only in the long run, and the main gains accrue tohouseholds in the rural nontraded sector. The consumption-based povertygap index indicates that poverty decreases in both the short and the longrun, with long-run gains being significantly larger. This reduction in povertyoccurs for all unskilled labor groups in the long run. Looking at income-based poverty gap measures, we observe that poverty increases in the ruraltraded and urban unskilled households almost outweigh poverty decreasesamong rural nontraded and urban informal labor households in the long run,resulting in a small decrease in the aggregate poverty gap index. Finally,the results indicate that income distribution, both in the short and the longrun, becomes more (less) egalitarian when we use the Gini coefficient (Theilindex), and that improvements are more significant at a longer horizon.

Interestingly enough, on comparing the effects of poverty reduction whensavings are allocated to infrastructure relative to the effects under a transferspolicy, we find that every poverty indicator under the first scenario, exceptthe income-based poverty gap, outperforms indicators under the transfersscenario in the long run. The fact that investment in infrastructure shiftsthe private sector production possibility frontier (whereas the transfers pol-icy does not) explains why poverty reduction may be more significant whenresources are devoted to infrastructure instead of transfers.

6.3.3 Investment in Education

Investment in education provides incentives for higher skills acquisition byunskilled urban workers and this has a direct impact on the supply of skilledlabor (see Tables 10 and 11). Debt reduction leads to an increase in openunemployment of skilled labor, in spite of increasing investment due to lowerconfiscation risk, because private capital and skilled labor are substitutes inproduction. Moreover, skills acquisition grows at a faster pace than labordemand over time, implying that open unemployment starts accelerating.This highlights the importance of considering both demand- and supply-sideeffects in designing policy reforms. For instance, we have performed a simula-

72

tion in which half the savings from debt reduction are applied to investment ineducation whereas the other half is allocated to investment in infrastructure.This mixed policy yields better results (in the sense of skilled unemploymentbeing much lower) than the case in which public savings associated with debtrelief are fully allocated to investment in education.

As urban unskilled workers enter the skilled labor force, and informalworkers are absorbed into the formal sector, the expected average wage forunskilled workers in the urban labor market decreases (despite the fact thatexcess demand for labor in the informal sector tends to put upward pressureon wages there), but a larger initial decline in agricultural wages createsincentives to migrate from rural areas. This pushes workers out of the ruralsector and into the urban unskilled labor market. However, migration flowsare reversed in the long run as (expected) urban unskilled wages starts todecline.

Consumption-based poverty headcount indices show no decrease in povertyin the short run, and only workers in the agricultural nontraded sector benefitfrom poverty reduction in the long run. The aggregate income-based povertyheadcount index shows little change in either the short or the long run. Theeconomy-wide, consumption-based poverty gap index shows a decrease inboth the short and the long run. In the short run, decreases in povertygaps take place for rural non-traded and urban informal groups; in the longrun, decreases are only observed for urban unskilled household categories,including formal and informal unskilled workers. Income-based poverty gapindices indicate a decrease in the poverty gap in the rural non-traded andurban informal sectors in the short run; this translates into a decrease in theaggregate poverty gap, despite an increase in the value of that indicator forrural traded and urban formal unskilled households. However, in the longrun poverty increases in the rural and urban unskilled groups are greaterthan the poverty reduction in the urban informal group. This leads to anincrease in the aggregate poverty gap in the long run. Finally, indices of in-come distribution again convey a mixed picture; the Gini coefficient shows amore egalitarian distribution whereas the Theil index indicates the opposite.

Poverty reduction under this scenario is clearly not as substantial as in thecase where savings are allocated to infrastructure. This is due in part to thefact that investment in education does not necessarily translate into a shiftin the private sector production possibilities frontier (PPF) as long as newly-skilled workers remain unemployed, whereas investment in infrastructure (asindicated earlier) does so unambiguously. Comparison between investment in

73

education vs. a transfers policy is more ambiguous, as results vary dependingon the chosen indicators.

7 Conclusions

The purpose of this paper has been to present an integrated quantitativemacroeconomic framework developed recently at the World Bank for thepurpose of analyzing the impact of policy and external shocks on incomedistribution and poverty in developing countries. The prototype describedin this paper captures important structural features of low-income, highly-indebted countries: the existence of a negative relation between external debtand private investment, a large urban informal economy, a limited menu offinancial assets, the impact of credit constraints on the decision to acquireskills, and a predominant role of banks in the economy. Taken together, thesefeatures create a variety of channels through which adjustment policies mayaffect growth and poverty in the short and the long run.4 9

Section II discussed the main features of the model and how they af-fect the transmission channels of policy and exogenous shocks. Section IIIprovided a detailed analytical presentation of the model's structure and Sec-tion IV discussed calibration and solution issues. Section V analyzed theresults of several simulation experiments. In particular, we used the modelto analyze a temporary terms-of-trade shock, a permanent cut in domes-tic credit to the government, and a poverty-reduction program consistingof partial external debt forgiveness coupled with a reallocation of savingson debt service payments. The latter experiment is particularly importantfor many low-income countries. Recent international evidence suggests thatexternal debt remains at unsustainable levels in many poor countries, evenafter accounting for aid from international debt relief programs. For sure, thedirect short-term impact of debt reduction is to reduce pressure on the gov-ernment budget constraint; and if one considers a country starting from aninitially high fiscal deficit (with a high degree of monetization and inflation,or a significant crowding-out effect on the private sector) then it is indeedbe beneficial to allocate savings from debt reduction to reducing the deficit.

49We described in another paper (see Agenor, Fernandes, Haddad, and van der Mens-brugghe (2003)) how the prototype version described here should be amended or modifiedto make it more suitable for the analysis of poverty-reduction strategies in middle-incomecountries.

74

But in more general circumstances, the question remains as to what the al-location of expenditure should be in order to maximize the impact of debtrelief on poverty reduction. Our model allows us to address this question.Specifically, we considered an experiment that consists of a permanent cut inthe level of external debt, with the savings resulting from lower debt servicebeing allocated to either lump-sum transfers to households, expenditure oninfrastructure (which has complementary effects on private investment and adirect effect on the productivity of private inputs), education (which affectsincentives to acquire skills), or education and health (which raises productiv-ity of the labor force). The results illustrated the importance of accountingfor the various channels through which poverty alleviation programs based ondebt reduction and expenditure reallocation may ultimately affect the poor.

The model that we have developed can be used to analyze a variety ofother policy shocks-such as a reduction in external tariffs, fiscal adjustment(such as public sector layoffs or changes in the structure of income taxes, forinstance), labor market reforms (such as changes in the minimum wage), a de-valuation of the nominal exchange rate, or a financial liberalization programbased on an increase in bank deposit rates. An important point that we wantto emphasize is that our model is not meant to be applied "blindly" to anyparticular country. Although the "prototype" version described in this paperis general enough to be applied to a variety of cases (at least as a "first pass")we view our framework as a flexible tool that can (and should) be amendedor extended to fit particular circumstances and needs. For instance, regard-ing the credit market, whether the "collateral" view or the "monitoring andenforcement costs" view are appropriate should be gauged on the basis of theevidence on the determinants of interest rate spreads.i0 More generally, wehave assumed that the credit market clears because commercial banks canborrow (and lend) as much as they desire on world capital markets; if banksare unable to do so, it could be assumed that they carry significant excessliquid reserves that would play the equilibrating role. Alternatively, creditrationing can be introduced (see, for instance, Decaluwe and Nsengiyumva(1994)), with possibly significant implications for the behavior of private in-vestment and the behavior of output in the short term. Finally, we havecompletely ignored informal financial markets. A good argument for doing

5 0 Note that, in practice, although interest rate spreads reflect the riskiness of loans theyare not strictly speaking measures of borrowers' risk, because the amount lent may berationed or may reflect lenders' perceptions of risk.

75

so is that financial liberalization in many developing countries has proceededto such an extent that these markets play a much less prominent role thanthey used to. However, it is also true that in some countries (mostly in sub-Saharan Africa) informal credit markets continue to play a significant rolein the financial system. Modeling the channels through which such marketsoperate could proceed along the lines of Agenor, Montiel and Haque (1993)but would add another layer of complexity.

There are several other changes in, or extensions of, our prototype frame-work that may be worth pondering, depending on country circumstances.In particular, we did not account in our analysis for the possibility of un-skilled unemployment. We assumed, as does the "conventional" view, thatthe urban informal labor market is characterized by ease of entry, a highdegree of wage flexibility, and limited labor protection. As a result, adverseshocks to the formal economy tend to translate into lower (average) produc-tivity in the informal sector. However, despite the absence of restrictionsto entry in the informal sector, urban open unemployment is often high indeveloping countries, and tends to affect both skilled and unskilled workers(which account for the majority of the poor). This evidence suggests thatthe extent of labor mobility between the formal and the informal sectors, al-though very high, is not perfect. Our analysis could be extended to accountfor unskilled unemployment along the lines of Agenor (1999, 2003) for in-stance, who argued that informational frictions may force unskilled workersto remain unemployed while they are searching for a job in the formal sector.As a result, a Harris-Todaro type mechanism can be used to determine thesupply of unskilled labor to the formal sector.

Another issue that requires some more thinking is how to endogenizeproductivity growth, in addition to the level effects that efficiency wage con-siderations bring in the model and the endogenous fertility rate. Recentstudies, most notably by Easterly and Levine (2001), have emphasized theimportance of total factor productivity (TFP) growth in accounting for out-put growth in developing countries. One possibility could be to relate effortto the capital-labor ratio, through an Arrow-type "learning by doing" mech-anism (see for instance Villanueva (1994)). A second possibility is to relateTFP growth to policy variables that may affect incentives to acquire anduse new technologies or to use resources efficiently. Yet another possibility,following Lee (1995), is to endogenize growth by accounting for the transferof technology and know-how that occurs through imports of capital goods.Such goods account for an important share part of total imports and domes-

76

tic investment of developing countries (see, for instance, Agenor and Montiel(1999, Chapter 1)) and have been shown to have a significant impact on percapita income growth in cross-section and panel data studies (see Lee (1995)and Mazumdar (2001)).

These possible modifications and extensions to adapt our framework toparticular country circumstances also lead us to emphasize the importanceof adequate econometric research to estimate key behavioral relationshipsand provide reliable parameters for calibration purposes. For instance, as-sessing whether private investment is responsive to the debt-to-taxes ratio ornot matters a great deal in assessing the incentive effects of debt reduction;and determining whether the stock of public capital in infrastructure hasa large complementarity effect on private investment is crucial to calculatethe growth implications of alternative investment allocation strategies by thepublic sector. Likewise, estimating the parameters of the rural-urban migra-tion function (and whether factors that income differentials matter), as wellas the skills acquisition function, is essential to assess the medium- and long-term effects of policy shocks on poverty. Finally, assessing the sensitivityof the finance premium to the ratio of borrowers' net worth to the amountlent is also critical to assess the degree of interaction between the real andfinancial sectors.

77

References

Adam, Christopher, and David Bevan, "Costs and Benefits of IncorporatingAsset Markets into CGE Models-Evidence and Design Issues," WorkingPaper No. 202, Institute of Economics and Statistics, University of Oxford(October 1998).

Agenor, Pierre-Richard, "The Labor Market and Economic Adjustment," IMFStaff Papers, 43 (June 1996), 261-335.

-, "Fiscal Adjustment and Labor Market Dynamics in an Open Economy,"unpublished, the World Bank (March 1999).

-, The Economics of Adjustment and Growth, Academic Press (San Diego,Cal.: 2000). Second edition forthcoming, MIT Press.

", "Employment Effects of Stabilization Policies," European Journal of Polit-ical Economy, (November 2001), 853-75.

-, "Macroeconomic Adjustment and the Poor: Analytical Issues and Cross-Country Evidence," Policy Research Working Paper, the World Bank (Jan-uary 2002). Forthcoming, Journal of Economic Surveys.

-, "Mini-IMMPA: A R+amework for Analyzing the Unemployment and PovertyEffects of Fiscal and Labor Market Reforms," Policy Research Working PaperNo. 3067, the World Bank (May 2003).

Ag6nor, Pierre-Richard, and Joshua Aizenman, "Contagion and Volatility withImperfect Credit Markets," IMF Staff Papers, 45 (June 1998), 207-35.

- "Macroeconomic Adjustment with Segmented Labor Markets," Journal ofDevelopment Economics, 58 (April 1999a), 277-96.

- "Volatility and the Welfare Costs of Financial Market Integration," inThe Asian Financial Crisis: Causes, Contagion and Consequences, ed. byPierre-Richard Ag6nor, Marcus Miller, David Vines, and Axel Weber (Cam-bridge University Press: 1999b).

Ag6nor, Pierre-Richard, and Karim El Aynaoui, "Labor Market Policies and Un-employment in Morocco: A Quantitative Analysis," unpublished, the WorldBank (March 2003).

Agenor, Pierre-Richard, Reynaldo Fernandes, Eduardo Haddad, and Dominiquevan der Mensbrugghe, "Analyzing the Impact of Adjustment Policies on thePoor: An IMMPA Framework for Brazil," unpublished, the World Bank(May 2003).

78

Ag6nor, Pierre-Richard, and Michael Grimm, "Linking Representative House-hold Models with Household Surveys: Implications for Quantifying PovertyReduction Strategies," work in progress, the World Bank (May 2003).

Agenor, Pierre-Richard, Nadeem Ul Haque, and Peter J. Montiel, "Macroeco-nomic Effects of Anticipated Devaluations with Informal Financial Markets,"Journal of Development Economics, 42 (October 1993), 133-53.

Agenor, Pierre-Richard, and Peter J. Montiel, Development Macroeconomics,Princeton University Press, 2nd ed. (Princeton, New Jersey: 1999).

Barro, Robert J., and Gary S. Becker, "Fertility Choice in a Model of EconomicGrowth," Econometrica, 57 (March 1989), 481-501.

Becker, Gary S., Kevin M. Murphy, and Robert Tamura, "Human Capital, Fer-tility, and Economic Growth," Journal of Political Economy, 98 (October1990), s12-s37.

Bernanke, Ben S., Mark Gertler, and Simon Gilchrist, "The Financial Accelera-tor in a Quantitative Business Cycle Framework," in Handbook of Macroeco-nomics, ed. by John B. Taylor and Mark Woodford, North Holland (Ams-terdam: 2000).

Bhattacharya, Amar, Peter J. Montiel, and Sunil Sharma, "Private Capital In-flows to Sub-Saharan Africa: An Overview of Trends and Determinants," inExternal Finance for Low-Income Countries, ed. by Zubair Iqbal and RaviKanbur, the World Bank (Washington DC: 1997).

Bigsten, Arne, and Susan Horton, "Labor Markets in sub-Saharan Africa," un-published, the World Bank (June 1998).

Blackwood, D. L., and R. G. Lynch, "The Measurement of Inequality andPoverty: A Policy Maker's Guide to the Literature," World Development,22 (April 1994), 567-78.

Bodart, Vincent, and Jean Le Dem, "Labor Market Representation in Quanti-tative Macroeconomic Models for Developing Countries: An Application toC6te d'Ivoire," IMF Staff Papers, 43 (June 1996), 419-51.

Bourguignon, Francois, Jaime de Melo, and Akiko Suwa, "Modeling the Effectsof Adjustment Programs on Income Distribution," World Development, 19(November 1991), 1527-44.

Bourguignon, Francois, William H. Branson, and Jaime de Melo, "Adjustmentand Income Distribution," Journal of Development Economics, 38 (January1992), 17-39.

Chen, Derek, Hippolyte Fofack, Henning Jensen, Alejandro Izquierdo, and DaoudaSembene, "IMMPA: Operational Manual," unpublished, the World Bank(November 2001).

79

Cockburn, John, "Trade Liberalisation and Poverty in Nepal: A ComputableGeneral Equilibrium (CGE) Micro Simulation Analysis," unpublished, Uni-versite Laval (April 2001).

Cowell, Frank A., "Measuring inequality," Discussion Paper No. 36, LondonSchool of Economics (July 1998).

Deaton, Angus, The Analysis of Household Surveys, Johns Hopkins UniversityPress (Baltimore, Md: 1997)., "Counting the World's Poor: Problems and Possible Solutions," WorldBank Research Observer, 16 (November 2001), 125-48.

Deaton, Angus, and John Muellbauer, Economics and Consumer Behavior,Cambridge University Press (Cambridge: 1980).

Decaluw6, Bernard, Jean-Christophe Dumont, and Luc Savard, "Measuring Povertyand Inequality in a Computable General Equilibrium Model," Working PaperNo. 99-20, Universite Laval (September 1999).

Decaluwe, Bernard, A. Patry, Luc Savard, and Erik Thorbecke, "Poverty Analy-sis within a General Equilibrium Framework," Working Paper No. 99-09,African Economic Research Consortium (June 1999).

Decaluwe, Bernard, and Fabien Nsengiyumva, "Policy Impact under CreditRationing: A Real and Financial CGE for Rwanda," Journal of AfricanEconomies, 3 (October 1994), 263-308.

De Gregorio, Jos6, "Borrowing Constraints, Human Capital Accumulation, andGrowth," Journal of Monetary Economics, 37 (February 1996), 49-72.

Delaine, Ghislaine, et al., "The Social Dimensions of Adjustment Integrated Sur-vey: A Survey to measure poverty an understand the effects of Policy changeon households," SDA Working Paper No. 14, the World Bank (WashingtonDC: 1992).

Demery, Lionel, and Tony Addison, "Impact of Macroeconomic Adjustment onPoverty in the presence of Wage Rigidities," Journal of Development Eco-nomics, 40 (April 1993), 331-48.

Dervis, Kemal, Jaime De Melo, and Sherman Robinson, General EquilibriumModels for Development Policy, Cambridge University Press (Cambridge:1982).

Devarajan, Shantayanan, Hafez Ghanem, and Karen Thierfelder, "Labor Mar-ket Regulations, Trade Liberalization and the Distribution of Income inBangladesh," Journal of Policy Reform, 3 (- 1999), 1-28.

Devarajan, Shantayanan, Delfin S. Go, Jeffrey D. Lewis, Sherman Robinson, andPekka Sinko, "Simple General Equilibrium Modeling," in Applied Methods for

80

Trade Policy Analysis, ed. by Joseph F. Francois and Kenneth A. Reinhert,Cambridge University Press (Cambridge: 1997).

Dorosh, Paul A., and David E. Sahn, "A General Equilibrium Analysis of theEffect of Macroeconomic Adjustment on Poverty in Africa," Journal of PolicyModeling, 22 (November 2000), 753-76.

Easterly, William, "Portfolio Effects in a CGE Model: Devaluation in a Dollar-ized Economy," in Socially Relevant Policy Analysis, ed. by Lance Taylor,MIT Press (Cambridge, Mass.: 1990).

Easterly, William, and Ross Levine, "It's not Factor Accumulation: StylizedFacts and Growth Models," World Bank Economic Review, 15 (May 2001),177-220.

Elbadawi, Ibrahim A., Benno J. Ndulu, and Njuguna Ndungu, "Debt Overhangand Economic Growth in sub-Saharan Africa," in External Finance for Low-Income Countries, ed. By Zubair Iqbal and Ravi Kanbur, InternationalMonetary Fund (Washington DC: 1997).

Fargeix, Andr6, and Elisabeth Sadoulet, "A Financial Computable General Equi-librium Model for the Analysis of Stabilization Programs," in Applied GeneralEquilibrium and Economic Development: Present Achievements and FutureIrends, ed. by Jean Mercenier and T. N. Srinivasan, University of MichiganPress (Ann Arbor, Michigan: 1994).

Fofack, Hippolyte, "Combining Light Monitoring Surveys with Integrated Sur-veys to Improve Targeting for Poverty Reduction: The Case of Ghana," TheWorld Bank Economic Review, 14 (January 2000), 1-25.

Fortin, Bernard, Nicolas Marceau, and Luc Savard, "Taxation, Wage Controlsand the Informal Sector," Journal of Public Economics, 66 (November 1997),293-312.

Ghatak, Subrata, Paul Levine, and Stephen W. Price, "Migration Theories andEvidence: An Assessment," Journal of Economic Surveys, 10 (April 1996),159-98.

Gottfries, Nils, and Barry McCormick, "Discrimination and Open Unemploy-ment in a Segmented Labour Market," European Economic Review, 39 (Jan-uary 1995), 1-15.

Grosh Margaret and Paul Glewwe, eds., Designing Household Survey Question-naires: Lessons from Ten Years of LSMS experience for Developing Coun-tries, Oxford University Press (Oxford: 2000).

Harris, John, and Michael P. Todaro, "Migration, Unemployment and Devel-opment: A Two-Sector Analysis," American Economic Review, 60 (March1970), 126-43.

81

Hernandez-Cata, Ernesto, "Raising Growth and Investment in Sub-SaharanAfrica: What Can Be Done?," Policy Discussion Paper No. 00/4, Inter-national Monetary Fund (December 2000).

Helpman, Elhanan, "Voluntary Debt Reduction: Incentives and Welfare," IMFStaff Papers, 36 (September 1989), 580-611.

Hoddinot, John, "Wages and Unemployment in an Urban African Labour Mar-ket," Economic Journal, 106 (November 1996), 1610-26.

Husain, Aasim M., "Domestic Taxes and the External Debt Laffer Curve," Eco-nomica, 64 (August 1997), 519-25.

lyoha, Milton A., "An Econometric Analysis of External Debt and EconomicGrowth in sub-Saharan African Countries," in External Debt and CapitalFlight in Sub-Saharan Africa, ed. by S. Ibi Ajayi and Mohsin Khan, Inter-national Monetary Fund (Washington DC: 2000).

Izquierdo, Alejandro, "Credit Constraints and the Asymmetric Behavior of AssetPrices and Output under External Shocks," doctoral dissertation, Universityof Maryland, unpublished (November 2000).

Jung, Hong-Sang, and Erik Thorbecke, "The Impact of Public Education Expen-diture on Human Capital, Growth, and Poverty in Tanzania and Zambia: AGeneral Equilibrium Approach," Working Paper No. 01/106, InternationalMonetary Fund (September 2001).

Kanbur, Ravi, "Structural Adjustment, Macroeconomic and Poverty: A Method-ology for Analysis," World Development, 15 (December 1987), 1515-26.

Khan, Mohsin S., and Nadeem U. Haque, "Foreign Borrowing and Capital Flight:A Formal Analysis," IMF Staff Papers, 32 (December 1985), 606-28.

Krugman, Paul, "Financing vs. Forgiving a Debt Overhang," Journal of Devel-opment Economics, 29 (November 1988), 253-68.

Kiyotaki, Nobuhiro, and John Moore, "Credit Cycles," Journal of Political Econ-omy, 105 (April 1997), 211-48.

Lee, Jong-Wha, "Capital Goods Imports and Long-Run Growth," Journal ofDevelopment Economics, 48 (October 1995), 91-110.

Lewis, Jeffrey D., "Financial Repression and Liberalization in a General Equilib-rium Model with Financial Markets," Journal of Policy Modeling, 14 (April1992), 135-66.

-, "Macroeconomic Stabilization and Adjustment Policies in a General Equi-librium Model with Financial Markets: Turkey," in Applied General Equilib-rium and Economic Development, ed. By Jean Mercenier and T. N. Srini-vasan, University of Michigan Press (Ann Arbor, Michigan: 1994).

82

Litchfield, Julie A., "Inequality: Methods and Tools," the World Bank (March1999).

Loayza, Norman V., Klaus Schmidt-Hebbel, and Luis Serven, "What DrivesSaving across the World?," in The Economics of Saving and Growth, ed. byKlaus Schmidt-Hebbel and Luis Serven, Cambridge University Press (Cam-bridge: 1999).

Lofgren, Hans, "External Shocks and Domestic Poverty Alleviation: Simulationswith a CGE Model of Malawi," TMD Discussion Paper No. 71, InternationalFood Policy Research Institute (February 2001).

Maechler, Andrea, and David W. Roland-Host, "Empirical Specifications for aGeneral Equilibrium Analysis of Labor Market Policies and Adjustments,"Technical Paper No. 106, OECD (May 1995)., "Labor Market Structure and Conduct," in Applied Methods for Trade Pol-icy Analysis, ed. by Joseph F. FRancois and Kenneth A. Reinhert, CambridgeUniversity Press (Cambridge: 1997).

Mazumdar, Joy, "Imported Machinery and Growth in LDCs," Journal of Devel-opment Economics, 65 (June 2001), 209-24.

Pattillo, Catherine, Helene Poirson, and Luca Ricci, "External Debt and Growth,"Working Paper No. 02/69, International Monetary Fund (April 2002).

Ravallion, Martin, Poverty Comparisons, Harwood Academic Press (Chur: 1994)."Measuring Aggregate Welfare in Developing Countries: How Well Do Na-

tional Accounts and Surveys Agree?," unpublished, the World Bank (May2000).

Robilliard, Anne-Sophie, Vivi Alatas, Francois Bourguignon, and Sherman Robin-son, "Crisis and Income Distribution: A Micro-Macro Model for Indonesia,"unpublished, Food Policy Research Institute (January 2001).

Robinson, Sherman, "Macroeconomics, Financial Variables, and ComputableGeneral Equilibrium Models," World Development, 19 (November 1991),1509-25.

Robinson, Sherman, Antonio YOnez-Naude, Radil Hinojosa-Ojeda, Jeffrey D.Lewis, Shantayanan Devarajan, "FRom Stylized to Applied Models: BuildingMultisector CGE Models for Policy Analysis," North American Journal ofEconomics and Finance, 10 (March 1999) 5-38.

Rosensweig, Jeffrey A., and Lance Taylor, "Devaluation, Capital Flows, andCrowding-Out: A CGE Model with Portfolio Choice for Thailand," in So-cially Relevant Policy Analysis, ed. by Lance Taylor, MIT Press (Cambridge,Mass.: 1990).

83

Sachs, Jeffrey, "The Debt Overhang of Developing Countries," in Debt, Stabi-lization and Development, ed. by Guillermo A. Calvo et al., Basil Blackwell(Oxford: 1989).

Serven, Luis, "Uncertainty, Instability, and Irreversible Investment : Theory,Evidence, and Lessons for Africa," PRE Working Paper No. 1722, the WorldBank (February 1997).

"Macroeconomic Uncertainty and Private Investment in Developing Coun-tries," PRE Working Paper No. 2035, the World Bank (December 1998).

Shapiro, Carl, and Joseph E. Stiglitz, "Equilibrium Unemployment as a WorkerDiscipline Device," American Economic Review, 74 (June 1984), 433-44.

Stark, Oded, The Migration of Labor, Basil Blackwell (Oxford: 1991).Taylor, Lance, Structuralist Macroeconomics, Basic Books (New York: 1983).Thissen, Mark, "Financial CGE Models: Two Decades of Research," unpub-

lished (June 1999).-, Building Financial CGE Models: Data, Parameters, and the Role of Expec-

tations, University of Groningen (Groningen: 2000).Thissen, Mark, and Robert Lensink, "Macroeconomic Effects of a Currency De-

valuation in Egypt: An Analysis with a Computable General EquilibriumModel with Financial Markets and Forward-Looking Expectations," Journalof Policy Modeling, 23 (May 2001), 411-19.

Thorbecke, Erik, in collaboration with B. Kim, Daniel Roland-Holst, and D.Berrian, "A Computable General Equilibrium Model integrating Real andFinancial Transactions," in Adjustment and Equity in Indonesia, ed. ByErik Thorbecke, OECD Development Centre (Paris: 1992).

Townsend, Robert M., "Optimal Contracts and Competitive Markets with CostlyState Verification," Journal of Economic Theory, 21 (October 1979), 265-93.

Villanueva, Delano, "Openness, Human Development, and Fiscal Policies: Ef-fects on Economic Growth and Speed of Adjustment," IMF Staff Papers, 41(March 1994), 1-29.

World Bank, Workers in An Integrating World, World Development Report(Washington DC: 1995).

Yeldan, A. Erinc, "Financial Liberalization and Fiscal Repression in Turkey:Policy Analysis in a CGE Model With Financial Markets," Journal of PolicyModeling, 19 (February 1997), 79-117.

Zeufack, Albert G., "Structure de propriete et comportement d'investissementen environnement incertain," Revue d'Economie du Developpement, 5 (March1997), 29-59.

84

Appendix AList of Equations

PRODUCTION

V¾ = [aXi{,l/flxiUT PXs + (1 -,B i)KG -xi] xi for i = AN, AT (Al)

Xi = V. + ai,jXi for i = AN, AT (A2)i

X o CX 1 UfXI + E iX (A3)

XG = C'XG{I3Xc1[XG2SGPXG2 + (1 -OX/ 2 )UGPXG2]PXG2 (A4)

+(1- 3XGl)K1 } PXG1 + ZaiGXG

Xp = axp{l#XpT1PXP + (1 - /Xp)KGPXP} PXP + E aipXp (A5)i

T1 = axp1 {j#Xp 1 TjPXP1 + (1 - /Xp1 )U;PXP1}-Pxlpi (A6)

T2 = CaXP2{3XP2 (ef SP)-PXP2 + (1 -PXP2 (A7)

Xp = CeTP{/TpEp'p + (1- Tp)DPTP}PTP (solved for Dp) (A8)

XAN = DAN (A9)

EAT = (1- aAT,AT)XAT (A10)

RGDPFC = X.iPViXi/PGDPFc (All)

EMPLOYMENTUR = UR,-1(1 + 9R) - MIGR (A12)

rd -(vl+1-7XAT 1 TXT O X AT l+PXAT (A13)UAT VAT (1 + IL-_).WAT CPX)AT

VAT = XAT - INTAT (A14)

UAN =UR-UAT (A15)

85

Ud= T WMX1P±Tl))+PXP1 (A16)

Uu Uu,-,-(1 + gu) + MIGR - SKL (A17)

(SP + SG)WsSS -ggu = Agag.[U W + (UP + UG)WM]/UU] + (1 A9g)gu,- (A18)

Uj = UU - UG- Ud (A19)

Sdp T2rs /3XP2XPPT21 +PXP2 A0

2 (aXXP22WS( + IL 1)) (A20)S = S-1 + SKL (A21)

S-_SG-_Sd A2UNEMPs = S P (A22)

WAT = WATPVAT (A23)

I3 XAN(l - 71XAN) fVA N1`7XAN

WAN PVAN PXAN ( I)1PAN) (A24)

VAN = XAN - E aiANXAN (A25)

WM = WMPT1 (A26)

WS = WSPT2 (A27)

OXP2Yef(1 - ef) ( T 2 \ 1+,OXP2 PuS

aPXP2 ef SpJ P 2 (A28)

ef =1-efm I R'S/PUS) (A29)

WI= =,xiU, PV' (A30)I i

VI = XI - aiIXj (A31)

MIGR- Am [UR,-1'Mln EW + U R' (1 - Am)MIGR-l (A32)86A]UR,2

86

Ewu = uWM1- + (1u-1 (A33)PULI,-1

EwA = ORWAT,- 1 + (1 - OR)WAN,-1 (A34)PR_1

SKL = (1-As)SKL-1 (A35)[(WTuI'-i + WTUF1 )edu (EWs)w ( ]

+As ~~Uu,-i1 w KE-)

Ews = s Ws,1 ,(A36)

SUPPLY AND DEMAND

INTj =ZE auiXi, where i, j = AN, AT, I, P,aG (A37)

Q =AN aQA{ 3 QAMAN + (1-QA)DANA}PQA (A38)

Q, = XI (A39)

QG = XG (A40)G~~~~~~

QP = QP{/QPMPPQp + (1 - 3Qp)DPQ} PQP (solved for Mp) (A41)

QAN = CAN + INTAN (A42)

I = Cr + Gi + INT, (solved for CI) (A43)

G = CG + ZG + INTG (solved for CG) (A44)

Qp=Cp+Gp+Zp+INTp (A45)

C = Eh CC,,hCONh for i = AN, I, G, P (A46)PQi

Gi = ggi G for i = AN, I, G, P (A47)

ZPQKZi = Zz PK for i = G,P (A48)

PQ,-

TRADE

Ep Dp (PEp ) 3 Tp) (A49)

87

MP = Dp (PMP 1 QP) (solved for PDp) (A50)

(PDAN /3QA UQAMAN =DAN (PMAN 1 -/3QA) (A51)

PRICES

PVi= Vi-1{PX (1-indtaxi)- ajiPQj Xi, where i,j=AN, AT, I, P, G

(A52)PEAT wpeA(l + teA)ER (A53)

PEp = wpep(1 + tep)ER (A54)

PMAN = WpMA(l + tmA)ER (A55)

PMp = wpmp(1 + tmp)ER (A56)

PXAN = PDAN (A57)

PXAT = PEAT (A58)

PXi= PQi for i = I,G (A59)

PXp = PDpDp + PEpEp (A60)xP

PQi = PDiDi + PMiMi for i = AN, P (A61)

PT, =PT2 T 2 + (1 + IL-l)WMUp (A62)PT1 =~~~T

PT PROFp + (1 + IL-I)WsSp (A63)PT2 =T2(A63

PK - PQGZG + PQPZP (A64)

PLEV - PLEV-1 (A65)PLEV-1

PLEV = CPI = E wtiPQi (A66)i

PR = E wriPQi (A67)

88

Puu = S wuiPQi (A68)

Pus = 5wsiPQi (A69)

PGDPFc = 5 viPVi (A70)

INCOMEPROF, = PVV - WiUi for i = AN, I (A71)

PROFAT = PVATVAT - (1 + IL_1)WATUAT (A72)

PROFP = PVpVp - (1 + IL-)WMUp - (1 + IL-1 )WsSp (A73)

YF,= PROF, for i = AN, AT, I (A74)

YFP = PROFP - IL-DLP,-1 - IF- FLP,-1 ER (A75)

YFPB = IL-1 [DLp,-1 + DLG,-1 + UATWAT + WMUP + WsSp{A76)

-ID 5 DDh,l - IF. ER FLB,-1h

YHAN =YANTRH + UANWAN + ID. DDAN,-1 + IF * FDAN,-1ER + YFAN(A77)

YHAT = 7ATTRH + UATWAT + ID DDAT,-1 + IF* FDAT,-IER (A78)

YHuI = ,yTRH + UIWI + ID DDuI,-1 + IF * FDUI,-1 ER + YFI (A79)

YHUF =YUFTRH+(Up++UG)WM+ID*DDUF,-1 +IF FDuF,-1 ER (A80)

YH5 = ysTRH + (SP + SG)Ws + ID DDS,-1,.+ IF FDS,-1 ER (A81)

YHKAP ID .DDKAP,-1 + IF FDKAP,-1ER + YFAT (A82)+(- re)YFp + YFPB + _YKApTRH

CONh (1 - inctaXh)YHh - SAVh (A83)

SAVh = savratehYHh(i - inctaXh) (A84)

( 1±ID U7S,h

savrateh = So,h 1±PIF (A85)(1 + PINF)(A5

WTh = WTh,_l + SAVh + AER FDh,-1 (A86)

89

FINANCIAL SECTOR AND INVESTMENT

Hh (YHh )(1 + ID) -'hD (A87)P LEV (P LEV)(A7

[(1 + IF)(1 + dev)]-6hF (1 + PINF)-hPINF

H= Hh (A88)h

'YBh _1±ID OB

1- =B Bh (1 + IF)(1 + dev)) (A89)

7Bh DDhD+ER FDh (A90)

ER * FDh = WTh-Hh-DDh (A91)

IK=PKKp (A92)

DLp = DLp,1 - ERAFLp + PK Z - re YFp (A93)

KP = Kp,-8(l- p) + Z-1 (A94)

Z Kp,. ( KINF K {(1+ ARGDPFC )CAcc (A95)KINF,-1 I. RGDPFc,_l

ez ((1 + IK)(1 - inctaXKAP) a\K

(1 + PINF1 l)aP k 1+ IL J

-ODIFG * FLG,-,ER ODDIFG * FLc,-IER 21

-kD ' )TREV )XREV )V

RR = rreq E DDh (A96)h

AFLBER = ADLG - (1 - rreq) E ADDh - ADLp (A97)h

IL= PR . ID-b[(l + IF)(1 + dev) - 1]1b (A98)IL =~1 - rreq A8

PR = [p ( c(NW P+DLP) ) ]y + (1-)PR_ 1 (A99)

90

NWPB = NWPB,.l - AER- FLB,-l, (A100)

NWP = NWp,-, + PK AKp- ADLp - ER * AFLp - AER. * A;,QA)-APK - Kp,1

PUBLIC SECTOR

AFF = E(wpejEj - wpmiMi) + IF * FDh,1l (A102)i h

-IF * FLp,l- IFG(FLG, 1- FF-1) - IF FLB,l

- 1 AFDh + AFLG + AFLp + AFLBh

NWCB = NWCB,-1 + AER * FF-1 (A103)

MB = MB-1 + ADCG + ER * AFF (A104)

Hs = MB - RR (A105)

-DEF = PVGXG-WMUG-WSSG (A106)

+TXREV - TRH - G-IFG * (FLG, 1 - FF_1 )ER - ILIDLG,1l

G = IINF + IH + IE + Gc (A107)

TXREV = ERE(wpmitmiMi-wpeiteiEi) + E indtaxjPX(408)i i

+inctaxr(YHAT + YHAN) + inctaxuu(YHuF + YHs)

+inctaXKAP(YHKAP)

AFLGER = DEF - ADCG - ADLG (A109)

NWG = NWG,- 1 + PK(AKG + AKE) - (ADLG + ADCG) - ER

+APK(KG,-1 + KE,-1) - AER * FLG,_1

K G = aG{IPGKINF + (1 -3G)KWP} PG (Alll)

91

Km = Km,i(1- lm) + Iwm, for m = INF, H, E (A112)PQp,1l

NWps = PK(KG + KE) - DLG + ER * (FF - FLG) - MB (A113)

92

Appendix BVariable Names and Definitions

Endogenous Variables5 'Name DefinitionCi Aggregate consumption of good i = AN, G, I, PCONh Consumption by household hCPI Weighted average of composite good pricesDi Domestic demand for domestic good i = AN, PDDh Domestic deposits by household hDEF Government deficitdev Expected devaluation rateDLP Domestic loan by private urban formal firmEi Export of traded good i = AT, Pef EffortEwu Expected urban unskilled wagesEWA Expected agricultural wagesEws Expected skilled wagesFDh Foreign deposits by household hFF Foreign reservesFLB Banks' foreign liabilitiesG Government expendituresGi Government spending in good i AN, G, I, PHS Money supplyHh Money held by household hHhd Money demand by household hIK Return from equitiesIL Interest rate for domestic loanINTi Intermediate good demand for good iKE Capital in educationKG Public capitalKH Capital in healthKINF Capital in infrastructureKP Private capital

51The index i (respectively, h) is used below to refer to all production sectors (householdgroups, respectively), that is, AT, AN, G, I, P (AN, AT, UI, UF, KAP, R, US, UU,respectively), unless otherwise indicated.

93

Mi Import of good i = AN, PMB Money baseMIGR Migration to urban areaNWPB Net worth of banksNWps Net worth of the consolidated public sectorNWCB Net worth of the central bankNWi Net worth of sector i = P, GQw Opportunity cost (wage) for skilled workersPR Rural price indexPDi Domestic price of domestic sales of good i = AN, PPEi Price of exported traded good i = AT, PPGDPFC Price deflator for RGDP at factor costPINF Inflation ratePINDi Price indeces used to index nominal wages where i = S, AT, MPK Price of capitalPLEV Price level ( PLEV_EQ with partial adjustments)PLEVe Price level equilibrating the money marketPMi Price of imported good i = AN, PPQi Composite good price of good i = AN, G, I, PPR PremiumPROF, Profit by good i firm where i = AN, AT, I, PPT1 Price of T1

PT2 Price of T2Ph Urban price index for household h = US, UUPVi Value added price of good iPXi Sale price of good iQi Demand of composite good i = AN, G, I, PQid Demand of good i = AN, G, I, PQ5 Supply of good i = AN, G, I, PRGDPFC Real GDP at factor costRR Reserve requirementsS Skilled workersSAVh Saving by household hSavrateh Saving rate for household hSKL New skilled workersSP Skilled labor employed in private urban formalSdp Demand for skilled labor in private urban formal

94

T, Composite input from T2 and unskilled laborT2 Composite input from capital and skilled laborTRH Transfers to householdsTXREV Tax revenuesUi Unskilled labor employed in sectors i = AN, AT, I, PUid Unskilled labor demand in sectors i = AN, AT, I, PUis Unskilled labor supply in sectors i = AN, IUNEMPs Unemployment of skilled workersUR Unskilled workers in rural economyUu Unskilled workers in urban economyVi Value added in good iWi Nominal wage for unskilled labor in sector i = AN, AT, IWi Real wage rate for unskilled labor in sector i = IWM Nominal wage rate for unskilled labor in the private formal sectorWM Real wage rate for unskilled labor in the private formal sectorWs Nominal wage rate for skilled laborwS Real wage rate for skilled laborWTh Total wealth by household hXi Production of good iYFi Income by good i firm where i = AN, AT, I, PYFPB Income by private bankYHh Household income for hZ Total investmentZi Investment demand for good i = G, P

95

Exogenous VariablesName DefinitionDCG Domestic credit to governmentDLG Domestic loans to governmentER Nominal exchange rateFLi Foreign loans to sector i = G, PGc Government consumption9R Population growth in rural economygu Population growth in urban economyID Interest rate on domestic depositsIE Investment in educationIF Interest rate on foreign depositIF0 Interest rate on government foreign loans

IH Investment in healthIINF Investment in infrastructureinctaXh Income tax for household h = KAP, R, US, UUindtaxi Sales tax rate on good i = AN, AT, P, GSG Skilled workers in public sectortei Export subsidy for good i = A(agricultural good), Ptmj Import tariff for good i = A (agricultural good), PUG Unskilled workers in public sectorwi Real wage rate in sector i = AN, ATwpei World price of export of good i = A (agricultural good), Pwpmi World price of import of good i = A (agricultural good), P

96

ParametersName Definitionaij Input-output coefficient for good i and j(ab Share of domestic funding in total bank fundingaedu Elasticity of skills acquisition to cost of educationaXG Shift parameter for public capitalag9u Shift parameter for urban unskilled worker population growth rateaQi Shift parameter in composite good i = A (agricultural good), ParTP Shift parameter in transformation function between exported

and domestic private productionaixi Shift parameter in production of good iaxpi Shift parameter in composite input of unskilled and skilled/capital

composite inputcxXP2 Shift parameter in composite input of skilled workers and private

capitalfiG Shift parameter for public capital/ 3QA Shift parameter in agricultural composite goodI3 QP Shift parameter in urban composite good/3 TP Shift parameter between exported and domestic private production13xi Shift parameter in production of good i = AN, ATI3 XG1 Shift parameter between labor and public capital in public productionO1XG2 Shift parameter between skilled and unskilled workers

in public productionI3 xI Shift parameter in informal productionB3xp Share parameter between inputs and public capital in

private productionI3xpl Share parameter between unskilled and skilled/capital

composite input3 XP2 Share parameter between skilled workers and private capitalfhD Money demand elasticity on domestic rate/3 hF Money demand elasticity on foreign rate/3 hPINF Money demand elasticity on inflationccij Shares of household consumption in goods i and jbc Collateral parameter6E Depreciation of education capital6H Depreciation of health capital6 INF Depreciation of infrastructure

97

6 p Private capital's depreciation rateefm Minimum effort level

Partial adjustment coefficient for premium??xi Coefficient of returns to scale for good i = AN, AT

YBi Share of domestic deposits in total deposits for household h-1gU Elasticity of urban unskilled worker population growth rate to

the proportion of skilled workers in total urban population'Yef Elasticity of of effort to wages'ah Share of transfers allocated to household h-Yp,, Elasticity of premium to firms' net worth positionggi Share of government expenditure on good i = AN, I, G, PKsE Shift parameter in skills acquisition functionIC5 Shift parameter for skilled private sector employmentAg Coefficient of distributed lag of effect of past income on growth rate

of urban unskilled populationAm Partial adjustment rate on migrationA,PI Premium shift parameterA, Partial adjustment rate on skills acquisitionQw Shift parameter of skilled reservation wage function

OD Shift parameter of external debt effectODD Shift parameter of external debt effectq5u Elasticity of skilled reservation wage to the unemployment rate

of skilled workersinds Elasticity of skilled nominal wage to the skilled price index(O)k Parameters used in calculating skilled nominal wage k = 1, 2,3

Shift parameter of rate of return to capitalOkBh Domestic/foreign deposits shift parameter for household hre Percentage of profits retainedPG Substitution parameter for public capital

PQi Substitution parameter in composite good i = A (agricultural good), P

PTP Substitution parameter between exported and domestic privateproduction

Pxi Substitution parameter in production of good i = AN, ATPXG1 Substitution parameter between workers and public capital

in public productionPXG2 Substitution parameter between skilled and unskilled workers

in public production

98

Pxp Substitution parameter between inputs and public capital inprivate production

Pxp1 Substitution parameter between unskilled and skilled/capitalcomposite input

PXP2 Substitution parameter between skilled workers and private capitalrreq Reserve requirement ratiouACC Elasticity of investment to growth rate of real GDP at factor costUBh Domestic/foreign deposits elasticity for household haE Elasticity of skills acquisition to capital in educationaH Money demand elasticity on real incomealj Elasticity of rate of return to capitalUIK Elasticity of investment to return to capitala'K Elasticity of investment to gross growth rate of infrastructure capitalUM Elasticity of migration to wage differentialsap Elasticity of inflation on investmentUQi Elasticity of composite good i = A (agricultural good), POS,h Elasticity of saving rate to deposit rateUTP Elasticity of transformation between exported and

domestic private productionaw Elasticity of skills acquisition to wage differentialUxpi Elasticity of substitution between unskilled workers and composite

input of skilled workers and private capitalUXp2 Elasticity of substitution between skilled workers and private capitalSo,h Saving coefficient for household hOR Share of rural workers employed in traded sectorou Share of urban unskilled workers employed in formal sector0, Initial ratio of the number of workers employed in the private sectorvi Weight for good i real GDP at factor cost price deflatorwti Initial share of good i in aggregate consumption for i = A, I, P, Gwr1 Initial share of good i in rural consumption for i = A, I, P, Gwsi Initial share of good i in skilled workers' consumptionwui Initial share of good i in urban unskilled workers' consumption

for i = A, I, P, Gzz1 Share of investment expenditure on good i = AN, I, P, G

99

Appendix CParameter Values

Productionaxpl Elasticity of substitution between unskilled workers and 1.2000

composite input of skilled workers and private capitalCXP2 Elasticity of substitution between skilled workers 0.4000

and private capitala7TP Elasticity of transformation between exported and 1.0100

domestic private productionaXAN Shift parameter in nontraded agricultural production 0.5407,/XAN Share parameter in nontraded agricultural production 0.6310PXAN Substitution parameter in nontraded agricultural production 0.3333aXAT Shift parameter in traded agricultural production 0.7944/3 XAT Share parameter in traded agricultural production 0.9151PXAT Substitution parameter in traded agricultural production 0.3333C!x/ Shift parameter in informal production 0.1315oxi Share parameter.in informal production 0.1500QXG Shift parameter in public production 4.0000,3 XGi Share parameter between labor and public capital 0.8953

in public production1XG2 Share parameter between skilled and unskilled 0.7605

workers in public productionPXGI Substitution parameter between workers and 0.3333

public capital in public productionPXG2 Substitution parameter between skilled and 0.1667

unskilled workers in public production°xP Shift parameter in private production 0.3393/3 xp Share parameter between inputs and public 0.8953

capital in private production

Pxp Substitution parameter between inputs and public 0.3333capital in private production

°xpi Shift parameter in composite input of unskilled 28.5784and skilled/capital composite input

(3xpl Share parameter between unskilled and 0.0310skilled/capital composite input

PxP1 Substitution parameter between unskilled 0.1667and skilled/capital composite input

100

aXP2 Shift parameter in composite input of skilled 4.6174workers and private capital

13 XP2 Share parameter between skilled workers 0.0350and private capital

PXP2 Substitution parameter between skilled workers 1.5000and private capital

aTP Shift parameter in transformation function between 2.1201exported and domestic private production

/TP Share parameter between exported and domestic 0.3349private production

PTP Substitution parameter between exported and 1.9901domestic private production

aAN,AN Input-output coefficient 0.0500aAT,AN Input-output coefficient 0.0000aI,AN Input-output coefficient 0.0000aP,AN Input-output coefficient 0.1000aG,AN Input-output coefficient 0.1000aAN,AT Input-output coefficient 0.0000aAT,AT Input-output coefficient 0.0500aI,AT Input-output coefficient 0.0000aP,AT Input-output coefficient 0.1000aG,AT Input-output coefficient 0.2000aAN,I Input-output coefficient 0.0000aAT,I Input-output coefficient 0.0000a1 ,, Input-output coefficient 0.0500ap,j Input-output coefficient 0.1000aG,I Input-output coefficient 0.1000aAN,P Input-output coefficient O4000aAT,P Input-output coefficient 0.0000aj,p Input-output coefficient 0.0000ap,p Input-output coefficient 0.2000aG,P Input-output coefficient 0.2000aAN,G Input-output coefficient 0.0000aAT,G Input-output coefficient 0.0000aj,G Input-output coefficient 0.0000ap,G Input-output coefficient 0.1000aG,G Input-output coefficient 0.0000

101

817XAN Coefficient of returns to scale 0.301XAT Coefficient of returns to scale 0.30

Employment'Yef Elasticity of effort to wages 0.1701efm Minimum effort level 0.23569!R Population growth in rural economy 0.0222ou Share of urban unskilled workers employed 0.2222

in formal sectorOR Share of rural workers employed in traded sector 0.5000a'M Elasticity of migration to wage differentials 0.5000ICE Shift parameter in skills acquisition function 0.0150ow Elasticity of skills acquisition to wage differential 0.5000UJE Elasticity of skills acqusition to capital in education 0.8273A\m Partial adjustment rate on migration 0.3000A, Partial adjustment rate on skills acquisition 0.3000Kcs Shift parameter for skilled private employment 0.3946agu Shift parameter for urban unskilled worker 0.0210

population growth rate'Y9U Elasticity of urban unskilled worker population 0.1000

growth rate to the proportion of skilled workersin total urban population

Ceedu Elasticity of skills acqusition to cost of education 0.1000Qw Shift parameter for skilled reservation wage function 1.0000ou Elasticity of skilled reservation wage to 0.0000

the unemployment rate of skilled workersinds Elasticity of skilled nominal wage to the skilled price index 1.00Ag Coefficient of distributed lag of effect of past 0.0000

income on growth rate of urban unskilled population08 Initial ratio of the number of workers employed 0.1000

in the private sectorDemand

aQA Elasticity of agricultural composite good 0.8000CYQP Elasticity of private urban composite good 1.0100akQA Shift parameter in agricultural composite good 2.0000fQA Share parameter in agricultural composite good 0.5000PQA Substitution parameter in agricultural composite good 0.25000 eQp Shift parameter in urban composite good 1.6519

102

OQP Share parameter in urban composite good 0.7978PQP Substitution parameter in urban composite good -0.0100CCAN,A Shares of household consumption in goods 0.5000CCAT,A Shares of household consumption in goods 0.5000CCA,UI Shares of household consumption in goods 0.1500CCA,UF Shares of household consumption in goods 0.2000CCA,US Shares of household consumption in goods 0.1000CCA,KAp Shares of household consumption in goods 0.1000CCI,NA Shares of household consumption in goods 0.0500CCI,TA Shares of household consumption in goods 0.0500CCi,UI Shares of household consumption in goods 0.7992CCI,UF Shares of household consumption in goods 0.1158ccI,US Shares of household consumption in goods 0.0600CCI,KAP Shares of household consumption in goods 0.0200CCG,NA Shares of household consumption in goods 0.1711CCG,TA Shares of household consumption in goods 0.1711CCG,UI Shares of household consumption in goods 0.0200CCG,UF Shares of household consumption in goods 0.2500CCG,US Shares of household consumption in goods 0.1832CCG,KAP Shares of household consumption in goods 0.1950CCP,NA Shares of household consumption in goods 0.2789CCP,TA Shares of household consumption in goods 0.2789CCp,UI Shares of household consumption in goods 0.0308CCP,UF Shares of household consumption in goods 0.4342CCP,US Shares of household consumption in goods 0.6568CCP,KAP Shares of household consumption in goods 0.6850

99AN Share of government expenditure on nontraded 0.0000agricultural goods

99I Share of government expenditure on informal goods 0.000099G Share of government expenditure on public goods 0.0000ggp Share of government expenditure on formal private goods 1.0000ZZAN Share of investment expenditure on agricultural goods 0.0000zz, Share of investment expenditure on informal goods 0.0000ZZG Share of investment expenditure on public goods 0.3684zzp Share of investment expenditure on formal private goods 0.6316

PricesWtA Share of agriculture in aggregate demand 0.2000wtg Share of government in aggregate demand 0.2000

103

wtp Share of private urban good in aggregate demand 0.5000wt1 Share of informal good in aggregate demand 0.1000wrA Share of agricultural good in rural consumption 0.5000wrG Share of government good in rural consumption 0.1711wrp Share of private urban good in rural consumption 0.2789wr1 Share of informal good in rural consumption 0.0500WUA Share of agricultural good in urban unskilled 0.1750

workers' consumptionWUG Share of government good in urban unskilled 0.1350

workers' consumptionwup Share of private urban good in urban unskilled 0.2325

workers' consumptionW 1 s Share of informal good in urban unskilled 0.4575

workers' consumptionWSA Share of agricultural good in skilled workers' 0.1000

consumptionWSG Share of government good in skilled workers' 0.1832

consumptionWSp Share of private urban good in skilled workers' 0.6568

consumptionws1 Share of informal good in skilled workers' 0.0600

consumptionVan Weight for AN real GDP at factor cost price deflator 0.0589Vat Weight for AT real GDP at factor cost price deflator 0.2347vi Weight for I real GDP at factor cost price deflator 0.0721vp Weight for P real GDP at factor cost price deflator 0.3852Vg Weight for G real GDP at factor cost price deflator 0.2492

Incomere Percentage of profits retained 0.2020US,h Elasticity of saving rate to deposit rate 0.5000So,AN Saving coefficient for agricultural 0.1623

workers in nontraded sectorSo,AT Saving coefficient for agricultural 0.1063

workers in traded sectorSo,UI Saving coefficient for unskilled urban 0.2027

workers in informal economySo,UF Saving coefficient for unskilled urban 0.1063

workers in formal economy

104

sO,US Saving coefficient for skilled workers 0.1063So,KAP Saving coefficient for capitalists 0.1063

'YAN Share of transfers allocated to workers 0.1667in agricultural nontraded sector

'YAT Share of transfers allocated to workers 0.5000in agricultural traded sector

'Yui Share of transfers allocated to workers 0.0000in urban informal sector

YUF Share of transfers allocated to workers 0.0833in urban formal sector

'Ys Share of transfers allocated to workers 0.2500in skilled sector

'YKAP Share of transfers allocated to capitalists 0.0000Financial Sector

16p Private capital's depreciation rate 0.1092ABAN Domestic/foreign deposits shift parameter 2.9603(BAT Domestic/foreign deposits shift parameter 2.9603OBUI Domestic/foreign deposits shift parameter 2.9603OBUF Domestic/foreign deposits shift parameter 2.9603OBS Domestic/foreign deposits shift parameter 2.9603OBKAP Domestic/foreign deposits shift parameter 2.96030'BAN Domestic/foreign deposits elasticity 0.7000OJBAT Domestic/foreign deposits elasticity 0.7000uBUI Domestic/foreign deposits elasticity 0.7000uBUF Domestic/foreign deposits elasticity 0.7000C'BS Domestic/foreign deposits elasticity 0.7000

'7 BKAP Domestic/foreign deposits elasticity 0.7000Oz Shift parameter of rate of return to capital 0.1738OD Shift parameter of external debt effect 0.0092ODD Shift parameter of external debt effect 0.0199UI Elasticity of rate of return to capital 0.8000ap Elasticity of inflation on investment 0.1000

IhD Money demand elasticity on domestic rate 0.5000/hF Money demand elasticity on foreign rate 0.5000/3 hPINF Money demand elasticity on inflation 0.5000(aH Money demand elasticity on real income 1.0000

'YBAN Share of domestic deposits in total deposits 0.75'YBAT Share of domestic deposits in total deposits 0.75

105

YBKAP Share of domestic deposits in total deposits 0.75'YBS Share of domestic deposits in total deposits 0.75'YBUF Share of domestic deposits in total deposits 0.75'YBUI Share of domestic deposits in total deposits 0.75C'IK Elasticity of investment to return to capital 0.8000aYb Share of domestic funding in total bank funding 0.9000

Ypr Elasticity of premium to firms' net worth position 0.0500PIpr Premium shift parameter 1.0934

',pr Partial adjustment coefficient for premium 1.0000bc Collateral parameter 0.1475UK Elasticity of investment to gross growth rate of 0.1000

infrastructure capital

C7ACC Elasticity of investment to growth rate of real 0.05GDP at factor cost

Public SectoraUG Shift parameter for public capital 0.2140/BG Share parameter for public capital 0.7500PG Substitution parameter for public capital 0.33336 INF Depreciation of infrastructure 0.01006 H Depreciation of health capital 0.14006E Depreciation of education capital .0.1400rreq Reserve requirement ratio 0.1000

106

Table 1IMMPA: Financial Balance Sheets

(in domestic-currency terms, at current prices)

HouseholdsAssets Liabilities

Cash holdings (H) Financial wealth (WT)Domestic bank deposits (DD)Foreign bank deposits (ER . FD)

FirmsAssets Liabilities

Stock of private capital (PK. Kp) 1 Domestic borrowing from banks (DLP)Foreign borrowing (ER. FLp)Net worth (NWp)

Commercial BanksAssets Liabilities

Domestic loans to government (DLG) Domestic bank deposits (DD)Domestic loans to firms (DLp) Banks' foreign liabilities (ER. FLB)

Reserve requirements (RR) Net worth (NWB)Central Bank

Assets LiabilitiesLoans to government (DCG) Cash in circulation (H)Foreign reserves (ER. FF) l Reserve requirements (RR)

Net worth (NWCB)Government

Assets LiabilitiesCapital in education (PK. KE) Loans from central bank (DCG)Capital in health (PK KH) Loans from commercial banks (DLG)Capital in infrastructure (PK * KINF) Foreign borrowing (ER.FLG)

Net worth (NWG)Consolidated Public Sector

Assets LiabilitiesCapital in education (PK KE) Cash in circulation (H)Capital in health (PK KH) Reserve requirements (RR)Capital in infrastructure (PK * KINF) Loans from commercial banks (DLG)Foreign reserves (ER * FF) Foreign loans to government (ER. FLG)

Net worth (NWps)

107

Table 2lUUA.: Lwroeconomoc kndicators

la percen Increas In the world prkce of the agrkcultunl teded good(Absokute elaln fromo baseine, urnles othervAse Indicated)____________________________

Ba Values ~~~~~~~~~~~~~~~~~~~~~Conos VIe. I20AsCoe BI0.oeso Conoo so Bus VolJs7 2 3 4 5 0 7 5 8 7~~~ ~~ ~~ ~~ ~~ ~~ ~~~0 1 2 2 I 5 1 5 5 0 10

307017526 1 22 991 0007 30A 2577 27102 2142 2177 Root. 2244.7 31 2566 M.O 304 3 A 27708 2144 2175 Z22772 22052 2214 'M.0 .2.0 0.7 02 007 043 045 047 0O4 057O- C4..ou '22M 72443 12371 72002 12100 122. 1204 I270. 132 14. 1303. 124.7 1370a 123i 13114 122 12221 13"2. -733 7445 no. -0.5 040 0.3 0.2 020 0.3 0.3 0.4 0.6700-6dg0*Os 7M07 MA. 770. 700 M tW.oj U& S 0102 022,2.5 MI77 703. MID MJ8 1500 0114 822A4 0330 0414 MA 40 t.t54 0 3 0.0 02W 0.0 02 0.3 0. 0.01

7o50.~~~~,d0s 0000.2~ING 25007 2000.1 3774 27102 254A 2177. 22707 2204 2313 2090. MA 20044 20774 2IIO. 204.0 2777.5 7 312 3340.2 23147000O~~~~o77,37 07~~~910 0747 0244 0572 30. 500A 70601 102i 1220 7072 077.4 010.2 030 14A 9304 0074 1307 1a.0 10404 027 32A7 20 042 07 0253 0 50 057 020 O.3 0A

-wa-0001 42 404 470 42.4A 42 44 404 40.0 44 40 00. 20. 472 .42 432 441 " 454 5 44 a 0.3 03 0.3 03 000m 0.00 03 0 3a 000n 0a31,la 80031064 370~~~~2" 240 MA04 2774 2747 2704 l 372 3 2 30.4 2 71.5A 20.2 MI 1.5 2747 270. MI.A 204. 200. M .AM 02 .3 0.42 .0.0 -007 AD.0 .0.2 .4 4.0

Po690876.O 07 32.7 227 234 240 007 3. 30. 30. 3 37.4 32.1 227 214 M4 a 2M7 35.4 A7 20A 3. OM 03 0.m 00W ON 03 0.3A 0 50 0.3 04EgaOdg387's 242.3 750A 704 M70 M 700 Waj 07t' Wig2 A2M2 3 770. 7002 70.- 703 7002J MI. 0770 C27. 032.7 0MA 32270 .4.0 040 .0.13 01 , .11 .011 .010 .010 .07

411

o.hwoi. 5- 0705 72204 7000. 1060 130A 750.4 11232 114,07 I I70 A2A 7302 130.7 130. A000 1302 ,0734 I 1734 07 01000 1200 7702 212 000 00 00a 02 0Do 0.0 OM 0A

04075004076456 80 .000050) 73. 727 740.7 -7.4 770. 7062 73.0 1 0 272 5 02.0 7I0.2 737.0 7-02 -700 70 7MA. 760.0 0014. 0710 8122. 32.2 454 000 I=.1 am7 4.7 .07 W1 am410

cipn D-, 002 65S MA. 67A 202 007 "A2 7low 1072 I77 107.5 NJ2 0 072 gas 0"A 002 1N.1 107 117, 7015 .7.2 .0.7 02 W 00 03 041 041 00 041M75*0606 I0.7 12 702 l01 W 102 102 I02 174 Ila7 & 10. 5.2 1.02 0.1 102 1MA 102 15.0 174 77 000 OM 000 0CW 00D 00 MOD 000 0m

C."idba'a636.0 0 GA a07 04 04 0A2 o 04 04 0 DA 2 04 04D 04 0 0 04 0A 0.0 04 0.. 000 000 a 0 M a0. 000 00 03 00 00 0m3170,3 tows 70.~~~~~~7 007 077 7723 722 M0 ALI 2 002 20A a" 04 03.3 77.7 74A 702 032 NJ2 202 MA I's .7.2 .7 OM3 0GM0 am0 " 'A07 001 0'M

h,Uht6nmfl 27.4~~~ A 4. It, 10. 732 10 a 122 5A 4I 142 &3. 311 72. 0.4 732 122 IR0 141 54. 542 7. 104 .42 a3n 022 0.3 047 007 0.1 0.0 047

Taoi000 2MI 4324 432 430. 427. 432 W 4A0 WA0 MIA. 42.7 402 42.2 47. WAS 437 7 43.4 43 4-04 4014 42 27I7 OM 043 .00 .44 .0.0 ID7 .407 .047 04M400707230402. 440 0222 403.7 450.1 4MS 407 400. 4004 MA 203 400 4A 4222 422 400. 4500. 4274 0704 4205 401 V a74 0 OM 441 .0.0 .4.0 .027 .00 .41 .4.0,

c ... I.- Q ~0 444 412 42.4 40. M4 I 5W 402 404 40. 404 446 4.0 4222 435 M4 I 04 45MA 4 40 0 C0 000 0 0 000 000 0.0 000 00 75 ~~~~ ~~~~~~~302 32. M?. 22. 040 24.7 3M4 3.7 202 DU 372 035 . 02A 04 3.7 3.4 20.7A 202 3 OM 0.0 0.0 0.m 0 0 000 OM 00 W. 0.

T3044064311.5 30.7 30.l 304O M". 303 34 201. 02 MI7 320. 30.0 WAS 30 3052 AM2 M.3 204.7 W02 W 272 025 022 .041 .44 .4 .41 .00 .40 .00007Stln40707~~~~~~0 0~.5 77 72 73 7?4 75 72 74 T4 7.7 AS 7.7 72 7.3 72 74 7s 72 74 7. 0.3 000 000 0.0 0.3 000 03 000 000

o503

3K 70.7 102 174 173 17 4 17A 174 10.0 10.1 70 M?. 10 17.0 172 17.4 77 ?17 1M4 10.1 Is 000 0.3 000 0.3 000 0.3 0.20 000 000M

7.0 0057 060 5.7 154 70. 10.7 70 702 1 I27" 77 177 70.7 00. 70.0 10. 75. 7 A5 102 77 17. 00 0.2 000 OM AMA 0.0 041 0.3 0.MC) la0500057 740 t0 7.1 7., I., I., II 72 A1 74 10 70 I., 1. 7.7 7.1 A2 A 000 0.0 0. 03 0.3 000 0.3 000 000M

00.347020 34.0A0 . 0D 00 04 00 00 04 04 04 0. 02 CA 02 04 GA 0.0 0.0 0.A M0. 03 0.2 000 0.3 000 0.00 000

Fb.~~~ CA 0 02s 02 02 02 0.2 00 GA0 0. GA 0.0 0 0 OA 04 02 04 0 0 04 0 000 000 O 000 0.3 0 ox00 03 000 0.0M

AM ~~~~~2434 30.7 270. 304 305.1 ;2 1 302 3004 2004 ML 300 37.0 274.4 201 300 2214 3.2 n7. W7 30 70.40 4. 044 047 020 0.2 0.0 02, 002NW 0017102.4 070.7 2274 m 230 000 310 27 2130.22 3M 3L 320 210. 23.4 227 702 34.7 2702 M3 A 3W77 727 14 4 0.4 047 0.3 020 020 0.07 0.2M

30*Xi 552~~~~~~~~~t WA4 3.0 A4 on .3. PAa72 M 000D002A57 I. M I MI on 072 502 002 OM 000 OM 0.3 000 041 0.0 000 000 07a60.6.0no 55I 320G5.7 00 on o 072 M 502 02 07 320 32 007 on on3 0.4 00. 70. 000 0W 0. 0.2 000 .0.0 000 000 03W

l.4760873 ~~~~ ~~~~0.0 04 00 AM 04 04 00 0.0 04 0 04 0.0 00 04O 00C 0A 02 00C 00 0 0M 03 000 000 03 000 000 000 0300 04 0.0 040 00 000 0 0.0" 04 4 0C 0. 00 A0.0 0.0 04 02 0 04 0 000 03 000 0.3 020 000 0.3 000 0.3

Tofkoow,i-o.inw 20 2 27 270 220 M,1 320. 30. 304 302 30. 2032 37 270. 373 034 3214 32 251. 307 31 70.40 4I 0.54 047 0.3 020 0.0 0.7 .0 00-17 7 2o 2 20 342 MIA MIA. 3625 2 203.5 2472 2072 MA 5. 7.4 337 277 M30. 220 2624 2077 312 M 10.7 3. .0.2 0.72 0.17 0.7 0 12 0.12 07 02 7

750fl703 ~~~~ ~~~~20.5 3.7 40. 00.0 00.7 20. 01A 53.3 114* 13 232 3.6 402 007 702 670 024 M0I4 170.1 IN .022 048 073 0.74 070 0.77 0.70 07 0.3M

d.aefl(5002 ~~~~~27412234A 2274 2714 2054 2304 0253 2474 ZI 7 2 3 322 22.4 27 22 14M' 275. Z302 203. 207. 312 30. 70.70 20 4.0.2 012 0.77 0il 0.77 0.72 002 07

'004540874 0~~~~~~~~~.1 3. 20 7 .7 32 4 0 42 43 43 4. 32 3 543 3.7 32 44 4 2 4 4 42 4 04a- 077 007 0.00 0.07 0OM 0.01 007 001 0.07

305 477.0~~~~~~~~V 0507 0074 7002. 1202 14472 7302 10743 304 307.4 4702 00.7 532 70404 72402 54A 7200. 1001 22 2120 0 420 AM0 0.5 a.2 6.0 M 0.2 020 712 7.5056n2.aW ~~~ ~~~~~0 0 0.0 02 0.4 00 0 0 04 0.0 0 0 A C 00 00 04O 0 0 04 0.0 0.0 0. 3 00 00 00 0. 0 0b0 .5.007w406 477~~~~~~1.0 M I8 0074 1062. 7242 5474 7I2 .074 2362 357A 470.3 002.7 200.2 50002 12002 54A AMA4 737I 27070 304 43 -53 0.70A 025M 006W 05M :3 020 0712 CAM

AM706402006 206 Ml 402 04 007; 32 MA74 132 752 73.1 32 MA 402 007 MA4 072 02. 14 710.7 In .022 0.4 0.7 0.74 070 077 070 0.70 020 0m

TO DAMM 47109 3? 004 1002. 0244 547 A74502 10740 3082 2372 470.2 432.7 80.3.2 10002 I 724.0 7400. 7302 07J 27090 270 420 52 6709 0G5 OM0 GAS C 030 772 7265788 1071~~~~~~~~V 27025 362.4 4577 05.7 0002 7400 MI2 WA7S 71 1022 270. 0. 4514 004 MI 04 7MA rj 1.4 MIA~ 113 0.IN C .3 7.4 7.30 7.3 .71M .00 I.42 .020 .07

OW-ft MA~~0. XIJ 4002 30A 400. 87.3 0702 143.6 1543 7390 3127 000 402.2 00 70I 0154 302 70004 77514 13. 217 M2 7.50 7A4 740 7 77 747 72 527

SsSOOOpu.57 02~ ~~~~~~ ~ :7 16 74 I 1.4 0 7.4 12 14 74 07 12 74 1 74 7A 'A4 ¶4 12 5 4.72 0.07 041 03 0.0 000 020 000 000 0x75.4375'7 02~~~~~~G 'A 02 0A 02 0.7 07 0.7 0. 0. 102 22 02 04 02 0.7 0.7 0.7 07 0. 27 .3.0 021 I0 am 00 0.3 000 02 0m

5o7 04~~~~~~~~~~~*A A GA CA GO GA 0.0 0 A DO 2 GA 0.0 0.0 GO GA0 GO 0.0 4 640 0. 00 06 0. 000 0.3 00 000a 000 0.3 (A40- 77 , 72 As 4 '72 72 'A 74 77 I 7 02 7.7 72 2 1 72 74 72 72 7'A 7. 20. IIs7 .10 .0 04 C.4SO .0.3 .00 404 044

IiiiatF 1.10 113 75 7I0 7.77 Is 7.70 779 AIC 121 1A0 A173 A04 AA0 A717 1.10 A7 77 1.39AI 72 .430 0.300DSOA 0.MIA 0. =Q W O-M CAM2 0=

I0 000 00- S00 * ZOKO 000 O 0 O0Z000000000000

00000 W W 000 ZO 000 00 O 00 O04 AZO0Z0000090,00 00O-O 1-O -0 ZOO Wa StO 'CO 00 PO00OOO0Z0000 ZO 00 0000000 0S0- 0400- 100 IT0 SI0 000 W0 0 00 000000Z0000 00000 000 00000 0000 00- alo 00 0000 000 000 00 0090000000 00 0-0'00

0000 WWO 00000W ZO ZOO 0 MO ZOO W00 004PZPI000 H0OO' 0000000000000Cor 0000 000-0- 000 ZoO 0 00 HO 000 S 000 000 000000000 40 00400000-- O 0000 0000 OO, 000 00 00 W 000 ZO0O00O0000 00,000000000000 SOM COM WOO 000 MO ZOO ZOO 00W.4000000000000Z00O00oO M 0oo 00- W00 W0 WO WO MO 00 OC"-Id00(P -PA..0000 0000 00 00 00Wel0 O 0000 U10000 000 00- 000 000 00 00000000004000000- 00 00

000- 00000 000- 000 00 0W ZOO 000 90-0 40000 00 0 00 00 *0000000- OO 00- 000 00 0W 0W0 000 0W0 0Coo 000 0 *000 00ocO0000000

WOV .0 o M0 ZO 0 00" 000 000 ZO 000000000000000000000000000- 0 0000- 0000 000e 000 C0 OO 00 000000000000004o M00V00-

O O 0 -O 000M 00 ~ 0 00 00-0 ZOO ZOOP .....0000-0ZO 00A 0000000 000 00 00 000 900-0 000 0-0 g EO000 40000000000Oo0*. 1000

wo- C, O' 0, W 0 t G000 00000el0 - t, .I 0000a,00 W0"ZOO 000& M, 00 000004000004000

I.t I.t IS. 10S. 0. S. o- - - C" 1 " 1 " 1 . -o.- I- n .-..

000 00- 00 0-0- w 0 C W00 000 000 W00 0 00 v00 ,0 000 000 000 000 000 0000 0000 0000 0000 0000 0000 0 000 0 100 0000 0000 1000 4000400000-0- 000- 000 000 00- 00-0- L 000- ZO 000-St0.00, I0 Co 0M 00 I'l 000 00 W0 C00 O0 0,00 '0, 000 000 0-0 000 00 r0 0- 0 fo4orOooOWOO00000000 W0 S O ZO ZOO9 O GS ZO 0 OO-- 0 000 I0-e00. a-0 0 IW0 000 000 06 00 0 000 000 000 0'a 00Z M0 000 I6a 0 0~0000000WV . .0 W. So .0 W* W: v" la C. a "" V. C. cm M C' C' r. r :M ra~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~40000000ZO 00 ZO ZO 00 OO OO ZO

CO 0 - - - - 0 0 0 0 0 0 - 0 - - - 0 0 0 0 00400000.0l

00 0 0 oO 00M0 WO 00 MZOO C 000 '000 000 00 00 000 00e ITV t 0 LWI0 000 000 000 000 000 00 I00 00 00 000 00 a0 000 000o 00 'WoOOoWoO000000.

W.0 ZeO O 000 00 W0 MO 00 0-0 000- I 000 0 . -0 000 000 0-0 C-0 00 C00 00 0-00 00 000 0 C0 0 -0 000 00 000 000 C0 0000000000000- 000 W0 000 00 000 000 00 000 00- 0000 0I00 " 000 000 '000 It,00 0-00 000 "I0 000 00I 000 000l l00 000 000 0-00. 00 000000I0J0WoO WO 00 000 00 0 o W.0 Co ZOO 0` ZOO 000 0 00 00I0 00 0 00 00 00 0 0-0 0- 00 00 I0 I 00 I0 00 I0 I 0000000 ,.4000000"009'O0 000 ZO ISO ZOO, lo0 ZOO :Z0, WO 000 00 001 00 00 C0 00 C0 00 I0 00 Il0 -0 00 00 00 00 00 00 00 00000000000000000CDOO M00 OO 000 a0 MO 00 S, . a0 ZOO 000 I00 -0M C.0 K0 M00 00 00 000 000 0I-0,000 0' 00 00a00 00 00 00 00 000 00 00

0IfoF00004ao0O

00 0 OO ZOO " . 000 ZO WO OO ZOWO 0K 0 IK 00 M0 00 0 00 00 W. 00 00 OM0 000 00 00 0. 0 )00000000000000400IC040MD0M0- 000- 00* 000 a00- 000 00* 00- 00 0000 000, O CW ' 000 OH 000r' M0 0 000 000 000 000 00 00 W0 000 "00 000 re, 00 000003000%000000000000 00- 00 00- 00 00- 0* 00 00-00 000 00 I.. 000No00M0 0,0 a00TOK

000 000 060 00 W.0 on0 000 000 000 000- 0L.00000400.000000

00 0 0 0 0 0 0 0 0 0 00 0 0 0 9~~~~~~~~~~~~~~~~~~~~00 O04 O0 0 -0 0p t 00 0f u O

00001301 LA0j00C1(4010 00000000u90 A1J9ood 10002000S 0-VdVdYta 01000

TobW 4OWA. Agacroaconomle lndJcators

30 P-"t Cut In dofmatic cmtM to(Abaokdo ftM bmth", uniggs othenvim WjcaW)

I z 3 BASS Vh- C-at VGGMG DffM,CWC B.,- c-ot M 8_ VAj_.A 6 a 7 4 2 M 3 1 G a A 9 10 1 2 3 4 5 6 7 G

P, "D 11.1 114M 111. Ulm? ZWJ MM 'M' M"3 211-A 2.32 2177. MIII MASS 73, 4m- m 0.23 03, on 0- 03, OM Om an,j I24,A 12" 1 2a&a 1310 6 1=6 I,W2 Vas ,Ms ,,, 121AA IMA IM2 IZNI 1310A 13321 IWA Jiro IMm, - -ISA, I 76 DO All 0-12 012 0.13 0.13 ..-1 alas AUX GUA ,, M M I M- M I MA MA .112 MG OA, .11 W, 0.15T- MCA I MIA I M"a 2"0- 21-3A V"G UM? MGAI�F MIX M0 I "117 2WI-3 20773 21IDA 211" 21771 MO, M, .110 mW 021 G 1. 0 21 Oil OM Om am CIAM A ",2 SICIA COCA im., MIA MCA wa.QA .,A o'? A M 0 "'A M I "C" ICU2 IW I 10 -3OA7 3M OM OM OM 02S 0.2g AM AMAx. 03 15.0 a QA 03 452 4U a Om 0-00 Om 0.00 ODO ODD am ODD am 0 002"A M2 M OA OA, AL ALL QA .1 C.,A VIA V. 7 MO ,, ,, 'A M MCI 2M 7 2"A 271 3 2147 2M. .14 MA 0.18 0.03 Om 0 as ADS OAS 0 M amIMA n, 32 7 MA U 0 M' 11 XI I I IIA M 1 32-7 33A U 0 M , M. M' In .103 75" 7674 MA X3 G, I MA M Om a W Om CM Om 0 00 mc, G ou GM mooI I 12is faza 85A TMA Ml- MA TWA Goal all, CUL UA M 155 mW ODI AM 0.02 042 am 0.1 Om am

10193 'MA IMS M. MMA -, IIM2 III&I 'M., 1=5 Gama MOO- 'M' 1=2 1-1 M1 11=5 1-4 IM.4 1 n AID Om C W Ca, Sal 0 M aA .27

41 -70A ?. .", -00A -W 7 -SO? 4MO -102.0 4002El*�dqoaa, A Ws (--ftwd-o riu ni 744.7 -671 .106 -"A -"A Cma S,13" 1 MCA MO -A MIA M 5 :;" 031 My ny 7,m? 70.4 TMA MA n,a `CSa "Ie "M 12M IM 11, GM a 35 Om an oil mat OMkm- d 0- a Ws Wft) MA "IA MA "Ie MA MA MA 7 CIL, A .1 Ml 5 812D M lm OA. ALI 0.03.G. "A To 7 MIA MA M, 0.02 Om am at, OA, 0.014M.3 44.0 M, IC, I 4" �a - ,, M, MID M, -113, M52 am 01. O" GM 021 Al, 0 M 0.23C.Pow- a "'A 'Ma -731 .13 .. ,6 422 ALI am CZ, QA3

a GA AD GA I" ITA VA I 0 CDO OLD M AM 'W am GM, OmNFPS fi-n-4 el"" 81&"O 411.51 w1al "16.75 InGA lo" 107.5 I'll 05 U152 I&I 17&21 M163. 'we A? I.AAC

OD GA OL oh OD 00 00 G,,t,CGIfi- g 78 7 ay VA 71I 74�. 'As Ml G62 Gas MA 4A o 0 'Ob o 0 AD G 0 O C a. G Om mw a. OLL GAS M M. M OLO M02 II 74A 18.2 X -2 WA a .,M 2 M Z. -CM 44, 4m D2 4m 442 maz2IA 163 121 I&S 13. 131 . I U. 146 1-1 -3 M I UA UA 13 0 112 1.4 141 II 9 I& MAa XAa Z. O- Ca, 0 U DN GM OM OX

ToIt- 435.1 433.0 1152 OAS W I ODA Aag 7 IMS .,A ,Xj, Alas OX, Alas .,, AX C OD 7 AID , .,A IX, II, 4AL AM Om CIA .17 AM "I CAO ODA M.Too, IGRA -2 432 4MY An 1 OGA -A -0 ASIA MI 4.J Wa 442.1 ASAI AS&I AWL 4.1 4XI A&I "I 4110 g U OLD OM 0AA 0 OA .10 AQA 43 AGA "A AAA a I QA AD a I's 12A aj . I ." "S A - Om am GM, moo am CLO 0 Go Om aw 0.C--P� GA OM 057 OM GM 0 a31.4 3zi IL? MA U 0 NJ A XI AGA M-3 31A U I -? MA aAL U 7 n' M I MA W OM OM CAO 0 M OM OM CM a M a Do a 0T� 3111 3OSJ 306 0 XOGA 30" WAI Mb W.? 3-3 MI 287A W? 0 .1i W&A MCI MA I 3OXA SIX. MO 9 .7 N,O AM .6, mW SM 0 W CZ, OM GM OA0- Pft- 'A 7 1 7.2 Ti TA TA rA T G FA 7 as ?.I 72 72 7 A 7.5 7A 1 a 7A T. GAS O-OD Om 0 W G W ovo 0 Do am ODO ALall 633 &A "I B" O&O WA 57.2 67A 0. OXI a 3 WA "A 0.2 Gas a a .12 9?.$ a Om 0 M CAL 0-M M Om a W CM am 0 a

TCW Ar- a o;T &Q 'T.0 172 IF. 17A PA I&D ISO M 1.7 I.-O ITA 112 174 17A 11. ISA el la. Om GM Om a Ca G W G . 0 Go G Ca 0 W DO13-7 I" I&D W I" MA It M 17.0 17� 15, IGA MO 161 16 3 I" MG ICA i'A 17. CAL AO OM ALL OM aM OM mat GAS 0.0I 0 IA I 0 I., 1.1 I., I f I 1 12 ` IA I 0 ix I I I I I., I., 1.1 12 1. am) 0 W Om ... G Do GM 0 Go - am A.C) G O OD G0 OA 0 0 AL OA OL CA G OA 0 a 0 C CZ 0 0 0.0 OA CA OA m am 0 OD Om OM G W 0 Ca Om am Om a aa 0 a 0 &C OA 0 0 0.0 0.0 MO OA 0. -1" 4I 4 3 Cl 4I 3 a 4A 4 1 4A .0 -153a ai .031 4m 432 4 M .0aI 4) 35 43

lw� MA M? 21AS 2903 MG I =1 33U Was M&A taL MIA 274 I a' I WCA Ml. =1 35Z-2 Wmi W -I7. 1.0 13M 15-7l ISOO 1624 16 so 'M 17M 170 1-. W, 192.4 2107 ZZZA 2� 249A M 2" =1 "A MA MaA M I Mao 2522 2011.2 MCA MA X95 U46 3M IDM 1330 15." MAO IUI ISW I&M 1721 ?W 1::1.0 52 S&I W 552 543 57 GAL MA IOI SID U ., .1 SaiI Cal 571 U MA G2. OM .-OG OM OW OM OM OM OX OWA uz 53, S. I W 5" 57A MA MA 612 SID SID n UI 552 Ma 57 4 M 6 9 M. Om ALL 0.00 a W Om OM moo 0. 0.GA 0.0 0.0 CZ 00 04 Oh OA OP Al U .0 0.0 Oh 0 G GA OA aG GA a 0. AM .00 A W CAL G W M Om C W0- d.� G 0 0 C 0 0 OD 0 0 mo 04 GA Oa 0 a 04 DA a a GA OD 'D DO 0 0 'D C 0M ADS GW GW G W 0 Go G W Om moo 0MA 262-7 275 : 2903 WS, 3201, WS M MA M 0 ZNI M 5 2M.? M; W4 319A =5 0 3GOA MG 3 M �ZT -ZN �0.15 .33 . U M 4 �l W, 4C.� .� 21" MIA W MIA MSG M MA M7.6 Z, 20 210 6 MA Ml-I M, M Mg 20 2471 n' MG, M -2.00 0 25 mO5 ..7 G V 0. W 0R- ,.*� Ms n, OA Sal a 7 we Oll 103-3 1119 1311 283 3719 44.1 U, Ml WI .11 MZA 114A IM Om .024 -GAS 4M -OM .35 'M 'M -am

M--a-" dI-C (W) 2.9 MIA 2" 4 MIA M 5 2WA MA 241.0 ZSIA M : 21:Al M : W, 014 2XI maz 2Q a 217A Z17 Ma � U -XW 025 0 M 0 ol OA7 0 O? Om 00G- WW, .'� - .� Clknp- 3 1 3A 34 37 36 A a 1.2 4 3 4.5 4 3 3 a 4A 4 1 43 4 I 4a 4A G. DM oil 025 0 25 0- GM 025 OM OM .2

99mmimmmmiuLTCoO 471.9 eW "? 'WI, IN" I., Ian IA14 MMA -71 4n Wa WA 8 MCJ I2399 I.5a IMI IVIa 2M 4 255S. .1 III -2AI I" .2.45 -1.5 .2 IS -10 415 -2A4 4 402 OA 0 0 G C GA G G

G 0 0

0 G

-0 W

OM

411 . r .1. 1.11 IN" 1111 185:14 IWO' go MOCII 255�0 A 44�* .0 03 M4A ICIC`10 12n"A - O"b ISCO-I I.no 5 M�44 M50 -10MdI .02 41 �O M. -2.15 I IMS -2AS I's 1=5 CM I�4L- log- 'MA 'wa 'MD 'MA IWO M Ma. M. MA M. ML 100-0 IWD IWD MA IN. 'MO MA 'MO wo- Om GM 0.00 OLD Om Om Om Om 000 LAL- IC 2- U34 515A MA M2 'M 'W2 MC: VI 7 1881.0 M 4 Mll 5101 AWL U1.1 JOM 6 1265.1 1464.7 IMA lanA MIG. -IM .2.10 .2.04 -2.05 .2W -2M 4V -IM 4M Zooa-� - 285 XI 'S 5" 69 $Oa el IOI3 "42 In' Ml 31 9 O-I S" Sal $02 91 4 oXg I, A Ix am _0 U ZAO 4M OM 4M 438 4A mall 4ar- othoW. it a Mr W. INZ, INZ 1. MOO '87" 'W" 2W' A W 3 W. 1=7 1=9 -53 Iowa 1872 5 2OIK4 Mn. - �41 4m -2AS 4 45 -. 5 -1-C -ZO 4AA -Z.8~ ft� 'a,I 27 MIA AM? GOj UIA Ta 4 WIS GA 3 I'M8 Ina ZMS WA 45I 1 647 I .3.0 1415 4437 GO. , IM. 0-M .M I" III 1-0 M 0 135 12D-.C"A M0 U13 OAA MCA MG 5052 919.2 IGUL 11194 1=6 MA VBB 43CA MA MA COCA 9144 IMA 110.7 INCA It -I. Z -XV av 4AS 4m .&S, ,70 ,7

�GaaanaanI�. ,- -ot I? 'Z 13 14 1-1 IA IA IA 11 01 11 16 13 14 IA 14 1; 1. IA m0a 4M . OW OW mAO 0. DAC OM Ob-GCG- As II G Ga OA OA 0' G' G" CI G , A 'A 0. 0.6 0 6 M? a? A I a 7 0 1 _2- X12 418 0 01 OM LAO GAG OM C M maD.." GA AL ma Sa .4 C 6 a-0 - G a AD G G " A G G 0 'A " " CAL SM ODD GLa G- 0. 0 - 0. CCa ALI 1 12 1 3 7 1 7A % IA I I I I 1 2 73 1 4 IJ5 ' G 'A 1 7 GN .0 15 4XII 4) 09 LA 4m 4. AIM 4)105 4b�P� IAO 113 1 4 15 1.17 IAO 1.19 II: 1'2'0 II' I "a 113 114 'Ac 117 Il 119 1.19 I" 11, OAOOa .0.0003 A= A, 0001 'Mol , owl 'AM, , owl 4 MO, , M

*MPA. RSItuoarol, Paowdy and bacoo,a 11EsuftbonSa Idloator30 pOrsool cot 1n dwnaao50 credit55o govOlrmsa.t

(Abosolute d.stoSi,aU 7 bssolln. saab. 0toA1ot.6 IradlootoBoo V.05A0- 1 C"`e Voo I 7503W'""a.0. BaO.O CU1701 .00 Boo Vo4sa1 2 a A 5 0 1 0 0 0 1 a a 5A0* ~ 0 0 I A I 0 0 a5070ss.oa~~~~~~~C500000.A) ~~~~~12.24 1 54'1 133 13., 'IO II4 I4 151 15. "A. It? 3 1404I Ia IO M4 47 1. I ss .15O 5III 40 4.07 441 451 4 01 44, 441 44190

0 0.o.o.50.54ao75dtnasapjsssj 557~~~60 51.5 a,. 520 Sa 5504 55 5M4 54 54 505. SI 51 MI. 524 6 45 55 55 54 MA4 559 40 5.05 , .547 5 4A1 4.07 451 4.51 450 .0

0Il050571~~~a51.W0I%a0500..s5as.w5 04.5SI sti Aa 54A 414 504 07 504 55 55A 54 O5A AU 55 414 504 a,, 5MA MI. 5A AW 0 00 50 M 0 0 .50- 055 050 5.5 55 05WEslwSowwShlnbtlsaAoSo0=7%dssO.asios.oos) 74N U4 5 5 70 45.5 57.0 4A 504 V7. 50. N0A 55. M0A 474 45 04 50 05 V7. 502 O50 AM 401, 401 .001 4.5 41 40, 410 .01 44SWS..571.50s'550017a0.00I.1.boj pa~~~, al 054 n5 504 04 40 45O.1 0.1 .74 7.0 51.0 55I 504 454 452 OA4 45.1 -2 'IO 447 5.50 5.5 05 5.50 OM5 AM5 5.50 0 550MEI.PwIs, kboa..a(%O .l45sl.I 77 7.7 71 7I1 771 77 74 74 T. 74 7.1 7 1 71I 7., 71I 77I 74 72 7 14 5.5 050 0A0 050 50 5.5 5.M 05 5

5 .lhl70=770.0o75d0I50s5as.l.,0) 75.5~~~I 144 157 I2. 15.1 117 575 77I5 112 -I 754 14 13 14 III 70 11 7 57. 114 Ill 11. 5.5 5.5 050 050 050 05 00 050 5 55 0540504.54Uw,..aAflI4.5040 ~~~~ ~~~~0.0 52 54 5.5 54 5 5. 5 02 52 0. 05I 04 54 05 54 04 0 5.2 04o 050 050 05 05 5.5 050 am5 5.50 5.0 5.51555717.551115555*710.0=5.507..,00 00 00~~~~A 04 5.5 5. 0 5.0 4 A4 AS 5.0 0 5A 0S so 00 5.0 5.0 A 5 M050 5. 55 055 55 050 050 M 05 550

7s50=a.a750o705ooa51050a) 550~~~~~. 054 2514 n 2 XI 50. 504 ,50. 574 57 504 514 MD4 5. 517 2. MA4 MA. ra 77.0 54 04 551 50 050 50M AM5 050 M5 050 500-1) n~~~~75 734 757 75J 75.0 74 7A I 7A2 744 iA 75. 75.7 7 In 70.0 755 71 7A00 7 05 .740 51.7 050 541 00 4.01 5.I .51 54 557.lsa.n50i..0(%10557 277~~~~~~~~~W 014 01 277 M?4 50. 5. M5 25.0 WI4 MI 014 51 21Y 71 0545K 0 M?7 WI 04 I 2 407 507 05 050 050 ON 5.5 O50 050

P05015000%aoOO5.55s.IO) 000~~~~~~~11 05, 451 001 004 00. MA4 M04 057 0WA 0.5 "A4 VI 01 Po5 "A4 MA. 00 MA. M0 4Om 45 507 05A 550 0- M5 550 050 05

7~5007=b54k14.a05555I0000.0055l1.5I.l5.2 55 554 V 5 as " a5 ma U. 55 5 as. MI ma aa MI4 V54 33 554 M4 550 550 050 050 W5 550 550 5.50 550 050051751107452 050 777~~~~~~~~~~~~~31 574 Ws 55 05. 27. 704 104 70" 4.50 VA5 5, W 25, no0 274 704 104 100 .55 . ,0 4 074 OIl 0.50 07 040 5.5 0A

0075515007 500 457 554~~~~~~~~~~~~~~~~W U,. 751 75. Me. no5 507 1514 50 AU 5MA 000 75 704 MA 007 50 750 050 45 4.- 450 4.5 - 4.5 4.5 4A5 45*5511075510=0 54~~~~~~~~~~~~~A 04 0 5.7 04 15 72 13 7A 75 5 5.5 5.0 5.7 5.0 10 7414741 10 5.50 55 050 5.5 0M 550 55 OM AM5 050LSv .75,`a7 50 54o 04 04 0D 04 54 0S AS 5.0 54 5 50 04 O 00 00 0a AS4 04 0 a 05 50 500 AD550 'M5 5.55 0.50 0.5 050

401a500.q57%dooI.o,a7 501 504 55~~~~~~~~~~~~7 AS. M~ A0 A8. 074I74 0.I s 44 40. 0. M 0. As so Iv. 57. 575 4.1 05 5 00 055 50m50W '00 0W 5.5

5505aI100000g505075o1h7W57 557 001 007~~~~~~~~I 554 55.5 554 55.0 004. ss4 oo 50 5 557 55 oA .5 U.4 -5 5224 -34 5.` .05 0S. 55 0Al ' 5.' 0 .555 .55 050 5.5= 00(o0= 7550 100 174 750. 712.7 asI.. 750 24 1514 7la,55 71 57, .A1 14.1 755 5555 12541.53 51 14 410 A.55 51 4.5 451 457 40 41 441 A

- -~~~~~~~~~~~~~~~~~~~~~~~. -. TT4 Z -.0- -0- -0 -00054 40=7505 45am1s761 sf0. 045~~~~Ca 045 5.45 5.5 040 0M 5.= 5505 0

0540 000017 00 155 4001 7.0 55750 5~~~~.5 5.5 00 0.70 04 05 04M 05050 ASCI=

45'po-', .....0w400 0,5455. an0 A45 M.0 5.5 . 00=ol 540= 5.50100.7 p5b017 01. SlSIa..A000700,50 5.00~~O 5.7 5.s, 5.0 Ol5 oSn 455 0QOO 0t50507.47000 1l5055510751750 ~~~~~~~04 074 54 0 0 075. 5.70030 55540 045057,0a5000 (505051550)17.0 040~~~AS M 054 040 051 on4 0555 0M, 54O7050770550 S7510(005005001054)055170400545 0M Om on 5.8 .2 on om . 5.-L.O 70*5050 001 (minsm54~7.55.5I.s550.fl0o 05 545 am5 050 550 55Om0 d.0= 00=7l~0,c,0.504 75105050 (5s78015i57170100 040 75 0`.5A 040 055 04200 .507 5505100505t50555751b50dd751415505.50.070051177555 05 029 0Am 0'M 0.51 m.W" 05 7 4500 0W)15507 0705 75705*4040550.105005000105555o00 000 OA4 Om 5.5 547 0.n 0AS OW` SCP 0101so50 tbcsf.w&.ssas4000.oyown 045 Om 5.IA 044 055 040 05577 4051 .00=1450555500005 S515a550 755150150.5)00501 0,0.0 05 5.5 0Do AM0 M4 050 50 5 50.....50.05 0555015 07500777 5 0 W5 545 5.5 045 000 000 5.0 ODOMA o755A- 05 5 (--- p-00,57 70777W0a. 55o 05 07 54 W5 5.7 04 5550 0*750 05051Ap-00005475 5.S500005 00-55 hn 5000-7 0A 0II .57 04 0Am 0ome 01550104.tlI 55.05750 005457 005077 7.,55.s.s 04 502 04 544 04 Om5 4074 41 .v 0eO7C. 705750 0075051 507d775145 4550 550~~A 50. 050 550 050 040 04 00D0M.50.55504 I51.0.O0507a~S5055W050514..0 W 04 547 05 045 54 54 0470 5.05m 45075707 05077770 7555050 50.50.5550705057074 04 0A2 044 050 04 0cm Son, 00 5011055055150 005 75100450 f504557po5077050 05105 04 04 OM 040 0am 5A5 5= 050 550010S50 00150 *4 70.00050 750505000055051005 050 050 045 A.5 045 050 0W0= 5O= 00=0*00=50.0400 7151750.I 0050175500005 050~o 050 045 mm5 045 050 04= 5 ) 04=

-l - -- -'s - -n50

A5105 0

5557511A,W.o,,55a.07 040 055~~~O 055A 00 05' 5.00455 00= 00

Table 6IMUA. Macroeconomilc WI,dctr

S pecen govetnnet foreign debtreduclo7w88th inreasedbensfts to hosoalods(Absolute deitosfrom baslitne, unless otherwlea Indicated) ____________________________

1 2 3 4 B:ass vkaes ClIY-d VI" 1*sI DWVGVMno Beales.., CWolOs oo n Base VSloos

7 2 3 4 5 ~ ~~~~ ~~0 8 a 0 10 7 2 3 4 I 8 7 3 4 5 6 7 6a 7

TrM -00 IMn. 20007I 20441 20?77 2110.2 24.5" 01772 M107 22447 M231 m Z.0 0015 017 0504 "0777 0151A 216530 210.3 MW53 Z423.1 7II AM6 720 734 ?A I85 70 O*5 85D 5m.:3

Go. 87i 8.7 'MI2 ¶24.3 1257.1 7282A 73106 ¶0322 7304 376* 12. 744. 72361 2483 7277.4 7212 137. M7337 I304 7302 3 50.* 7450 425 277 427 4.44 450 427 54 3.37 5W28

6i,pfldsoo 76 78 MA2 775. 7042 MA8 5704 = 4233 on A 84M. 304 707. 77 757 I 82>0 5142 825.3 5, 5487 877 2* 268 2i7 0.87 2.8 3.05 5.7 3.23 3.33 4

Toas .0~~~~ 7283 22887 U 204 30772 21102 2743 20172 Z73.7 23447 2373. 25282 t 075 2057.7 25643 21774 01515 0103 23783 2333 2323.1

P77,8044p8 017 8147 3080 8512 MA8 480N 708 73027 7301 7075 8177. 8228 U084 2872 877* 283. 77173 773 1047.4 O 1 77 873 8.4 8.78 8.70 72D 7.2 742 78a2 A

P.t8 -l27 470 40 47A 4 4 44.7 452 40. 4"A 40 442 342 47A* 42 0 3 07 45.0 45A as* 48. 020 028 028 02 020 028 028 0O 020m 0

P"-,0882 87 2832 2284 2772 2747 274. 28* 274 2834 M. 266. MI6. 2503 0722 2757 M,0 230. 20.1 2887 30 M.3 45 02 m3 730 70 17 3.7 .2 a

7flsO..4374 32.7 y 30.7 30. 00 4.7 A M4. I0 042 M. 37.4 30_1 37 31.4 u 340 04 I0 M8 28.8MA 028 028 1 020 02 02 3.0 O'W 004 20, 020 3.

0n 2-722 .47 .740 .74 .77.4 .28 .-2 .887 402 .7'02 -.77 .0 .20.7 n.7 .772 -00 .44 .847 .82. .7827I 024 0 0 18 010 0.70 028 02 02 .02 011

00 0 a 74737.44. ..t. 7253I 7772 7TM7 ?78A 7702 728. 751 287 822 83 734 n77 74a 77 M 770 7824 7510 281 5 012.7 8n. .. .2D .028 .0.21 A01 .02 .00, '02 030 0.7

F-~~~~~~ ~43! .643 .46.0 -70* -737 .25A 4807 Mo .8. .8 .02 .42.7 .64 47A .702 .737 .75.0 -283 .82A 4.81 3713 3713 3.74 3.15 3.78 3 07 3.76 3.77 3.77 3.70

clat4 0" I4 MA2 820 872 28 04W7 "A 1300 107 777.7 Ml. 658 a 02. 64 28,8 04 a8 730.2 1077 717l 000 00 .071 .0, 03 020M 010 3.77 018 02

7*0IM .f 77.7 I5 160 78.7 76 3 702 76* -5 772 V 17 5 0 G 760 70.7 703 18.0 78* 18* I 172 3 OM7 020 02 02 3.2 020 02 020 0.2 02 0

C4r045ift7¶ 02 3.0 02 02 3.0 02 02 0 0 02 0 02 00 00 0U 0.0 00G 0A 0.0 02 028A 020 020 0M 020 020 0.2 02 020a 0

777288 707 a07 67. 773 740a M 70. 82 WA. 280 50. 742 75 87 773 748 78. 823 28. GM7 7287 0 00 .0.,, .0.7 030 020 070 0I5 0.18 GA

Ch 77074f- 82.6o 214 7813 727 732 73.6 133 147 743 14* 77. 23l 10. 703 73* 737 722 74. 742 47I 73. a004 0257 02M 0 14 0 12 3.7 0.14 3.7 015 016

T.W IM I~~~28 03. 4302 438 07.7 4M6 430.7 408 74 442. 437* 430 43725 43* 440 'I 473 4422 423 4430 & 2.1 2.7 22 2 232 2.0 2.4 244 222 2.57 0

304oI .O26 4524 44* 402.2 453.7 2457 4 45 4572 418 4442 447 404 43, 4-4 47 477 458 2800 4477 4627 403. 2.7 031 no 233 204 2.43 2.4 222 2.07

C ---,o 440 4'O4 472 02. 433 447 4'2 Mg. A 28. 402 440 410 42. 4.3 V47 48O 4* 48 48. 000 028 020 028 020 028 3.2 020 02

0084817824 77~~~~~~~.4 32. 32. MA 342 347 MA 381 38* 3 31A4 32. 02. 33. U.0 U4 70. M, MA 02 020 m 020 02 02 020 00 3.28 020 0o

115fl8 a3 3172 387 W* 308. 380.2 388. 383 317 3803 207 3762 372.0 370.2 3170 2707 700 me0 407A 20.0 53.7 5G8 3.47 GM0 557 'M GA, La 075 5

D 6o.607807y,-oll 62 71 7I 73 74 7.5 72 7* 'A 7. 0.4 7 1 73 73 74 72 7.5 72 7* 7. 028 028 020 025 0. 028 000 OM 028

Toad & 30 187 I5* 77 1730 17I.4 77* I7* Ig0 78, 76 18.7 78.4 17 0 27 74 17* 772 1620 78 7 8 020 M M Om 0200 00 00 00 000 3.28 02 0

74.0040 , 73.7 I5 784 471.0 761 11 151 166l 18*A 1712 77 757 5 100 I 0 763 78.5 788 78 72 1 0G 004 720 7.28 000 728 020 028 020 3.

7' 7287 45048. 784 0 02 30 72 77 3.0 72 0 0 02 00a 00 70 3. 02O 02a 02 00 0.0 02 3. 020 0O O 720 0.0 0.2 020 0.0 020 00 0D

7506:62.57 0252 7 54l1 55.2 503 574 NA8 U MI* 512 82. 041 54U 503- 503 574 02. 782 720 0Do a 000 72 05 m2 000 02OM 020 70 0M o

7.2S3007004 ~~~ ~~~~0 0 00 70 02 0G00A0 4 0 0 02 02 02 0 0 02 2.0 00 02 000 028 520 0'M 02 020M 000 0.2 028 0

OOlindalD 02 02 02 02 7 07 04 0 02 0A 72 0.0 02 02 00a 02 70 00 0 2G O 020 000 028 000 020 000 020 000 0D

TMG%We0mh .s M04 0827 0758 MA,. 20.7 3007 M0. MA0 '380* 300 04. 83 7732. 2 270M 5 22.7 lt MZ 3378 W 7437 300 *07 0*4 75 7.8 7.7 1. 20IA 3.4 27* 2.2i 24 2.

074 u87045 278 2348 2374 372 2302 2MA 24.5 2476 51* M 278 25 M, l 225 73 2370 M77 243.7 2403 2303 257 3.04 A57 744 IA* - I*4 a478 78 I.7 787a

64.760*.o,00 282~~~~~n 33. 445 Gas 507 506 87* 7303 74* 7N2I7 282 38. 44 783 750 872 023 7302 7757 54 020 000 774 0G 23 0M 40 722 .82 Of 0

JAloal`l,l, boa1407546ol-d 08) 074* 2240a 2Z74 2IA, 235. MI0 243.5 2472 2515 258 2153 235. 230. 2332 2370 0477 245.7 2403 2034 517 038 745 I4 U 740a M5 I0 122" 78 771 I

ak5447eoo7 3 1 .4 20G 07 soA6 4024 43 42 41 37 34. 38 3 30 42 432 4 4 5 I 020 007, 0 01 007 0.07 020 30 M 0202 03 0 00

Tw-. ~~~~~4778 828. 6472 7042. 7242.3 I74* 7674 .7674 2888.8 250A 472A 8M8 000. 7044 12477 74542 7000. 7504 2700.7 0072 081 1M 287 4.75 5_47 8.75 877 OM 71120 147

5.7t,.1017865a 02 02 00 02 00 05 42 02 0.0 0~~~~~~~~~~~~~~~'Do 02 00 3. 0 00 02 00 00 02D 0W 020 038 00 00 000 720 03 020 0 00 0m

L0f05000.0728.o7 7000 7000~~W 1000 7002 7042 702 72 227002'W I' 70'02 70M78A20 10 70 00 782 70 2 028I0WO7A8 00C0.M 020 02 00 0

LMo, G*77o 3~034 0782 506. 843. 7072.7 7007 28 76717 10474 2318 063 524 7074 8071 10775 7273.5 1474 5 1650 755.4 2U 0*5 12 20 3 002 500 6.2 728 827 70.57 I3

04500e ~~~~ ~~~~202 30.1 48. 58. 287 MA* 87* 733 7742 7301 282 30.2 40.8 78A 750 8102 83 7308 775 740 020 00 074 033 0.3 0.42 052 022 070 7

7048 471 878 0470 7042 72723 7447 7850*A 70740 248* 2707 4* 670 042D 70443 7247.7 745,5 7880 100* 27387 227 025 78 23 4 15 .47 5.75 817 08 77*5 1477

B8wtk.w 50 77 0755 24 452.7 543.7 Ul76 7444 8423 047.3 1764 707 276. 384 44 G47LD 844 7434 04 MIA04 12. * 728 178 78 2.27 278 3.3 05 420 5

0600628 ~~~~~~~~~2048 U"7 44046 A 504.4 805 80.2076 7442 7705 7380A 284 3817 400 60477 8004 810A 23 "0'4 775.7 40 24 047'M 2.3I 2 1 75 37 3

=R-GW0w2 M a o I 0 I I 15 72 74 7 4 741 1 '4A7. I OA "7 72 72 74 71 74 7 4 14 1 .02 07 020W 028 02 020x 000 02 020x 0M

3.Ca0.0 -oWo7.o.) 6. . 06 0* 0* 037 07 07, 07 0. 72 .0 08a 0* 06 0.7 07 03 0 7 0 3.8 204 .0 07, G007 001 007 007 007 001

0.776- 60 5A0 02 62 62 5A 62 00 '2 G6 02 02 a0 8 0 `a a 0 GA * 6 8. 00 0000 0 07M 028 02000 020 0 00 a7

Oo.70

,7 ?I 7.2 73 7* 7.5 75 A 76 77 7 4. 71 72 73 74 7*. 7 7 78 I 72. I I 03 D07 .0- .38 .20 .0.W04M .00 0, .02

Rlk plotI 7.70 773 7.74 7.78 177 778 778 77 7*5 77 7130 I73i 1.14 7.76 17.7 7 77011 II0I.2 7.01 a~ 00007 021 0M, 0200 020 0 1 0-MD7 027 20

IMupk Strusoltr,. P~soy wnd 051.00 Distribdo.e Olndlcto

(Abo dot. d sol n -' - fnb olm , U nto.. oth,o,wO o Indl-OcQ0.)_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _a-. V.04. CAl,s Olt... OhtIto.o 9.0w.. C,-wO .0049. v.Oo-I 2 4 0 S 0 10 2 4 0 A 0 7 0 0 10 I 1 2 A 0 . I 4 1050300001000.223*040.4 ~~~~~~~~~12. ISA 1310 11. 14I Ut 14 144 11, SO) ISO It. 132 I3A I- 142 14 Iso 153 111 021 a1.0 010 010 0AM 0AM 020 12 0 21 031S0304204002in0%fl,,00

0220.w003 504~~~~ 5th"2A 173 a 54WI 2 04I 0 0435J 045 04 10 -1 52 I 01. W I43 4 34 2.0 1.0I .00 M '0. M 03 -002 .02 00 .04 0M0 .0 .00 400

E'0

11..1441020 0

n{%40400IolOs.0204 52.1. 40A AS .2 42 41.2 003 44T. A0 02 0. 032 4.0 402 44A 12 14 2V7 20. 2WA 24. OM O00 0,0 0O 0M 000 000 OM 0M 0010.ns21l..100.m,w..4.inI'bm.ooi.t.w0, 50.0t 5M. $00. AT0 41 '2A Q0 00. 53 402 000 5U 500 ITS 4'L' 423 40A M 23 00A 0.0 000 000 0 001 3m 00 021 001 0040401w05- 0b000.2212.14,w 40 1 53 0. 403 23 4.0 401 401 41.0 21 400 5. 003 401 42 410 441 21 40 000 1.0 1.0 00 0 0 . 001 .00W1 A..0 .000 ow004m00.inO%.0o5.Io.o4405.osoI 844~~ M, 05s, 07. 003 003 703 700 7A III 007 011 04 27 0At 005 123 70t 113 1 037 04 1.0 0.0 00 00 .00 .0 .01 1.0

E0.~Om.400-oiO .JII I.1 I., 41 I., T1, 13 43 43 73 4.1 II 1 41 07 4 13 ?2 42 C"04 0M 00 0.0 000 00 040 040 00 000E.w0.,.o.0k,t.Afn IS d.40..rotosl..oo ~153 14 141, I23 111 I II 113 113 11.5 153 100 15.1 -2 121 -I 0 "10 113 I 0 000 000 1.0 040 040 00 040 000 0000.4005..44m00w00444..O.aw 03~~~~~~O 03 03 03 0 0 0. U3 03 02 0 03 3 003 3. 03 03 03 03 0. 0 040 0.0 000 1.0 1.0 M4 000 000 00L .wo...daoo.in.0.e.0fo 04~~~~~~~D 0.4 04 1.0 AS . 00 0.0 0 1.0 00 00 S. 00 1. 0.0 00 00 00 00 04 1.0 1.oo 000 000 00 00 00W 0Po.in003%dSp.0t.q,.uo3 240 253~~~~~~~~n 053 0.7 001 00M O 04 7 002 VA4 0V. 23 254 I0 M0 2. 47. 27 243 .00 .3 .002 .002 .0 02A00 I00 .0 402

P022001000W 140 IL3 41. 73 4.0 U 403 'I 743 743 44.4 442 7 73 741. 742 743 44.4 .44 0 714 0AM 035 1.M 1.23 OM5 2.2 31 OM 12 035P4Wo±.w04%0400001213 000 007 057 00,4~~~~~M 007'0.4 00 W' 4 04 05. 053 000 00 00 W.0 004 007 01. 007 07 32 03 1.2 02 00 .2 4 .2 0 00P40lw00I0044%~~~~~~~~~~ 414~~" 00.4 053 211 210 214 21. 04 000 203 21J 013 012 211 010 304 004 004 M.0 21. .001 040 000 300 040 000 001 1.1 00 03OAPOnootOSaoSOsoa 005 003 0~~~~~~~~45 00 00 00 SO 504 00.4 01. 03 002 a, 001 no0 00. 031.0 . 7 00 002 AM0 02 002w 04 030 04A.4 000 'AS

4100012200510022140..70022100000101010007 0.4A 02A WA. 004 -0 WA0 54 W U7.0 343 024 0 00 00A 00 a' 050 M 04A 24 OM 000 0.00 00 1.0 04 000 040 - 0400

2105.07S40 423 07.1 44 203 331 U.4 21. 10 10, ISA2 QS. AI3 023 05 123. 00 213 50 -A3 107 OIl 312 4a 04 000 002 00 03 CIA 000a1t122f0000097 53~~~~~~~~~~~D 0A 00 1. 0.0 1. 13 13 12 10 03 0 AS 0.7 0.0 13 13 03 13 1. 00 0.0 000 3 04 .0 001 0.0 a01 'A'L.-. -d w,IIA0.0 03 02 2 0.0 AS 0.0 00 AS 0 0.0 O3 0.30 0 0 0 0.0 0 A a 00 A 03 0 000 00 300A 340 0.0 0.0M 0.00 040 040

AwflbOOinflt%002 4~~~~~~~~~~051 40 a 4 44.0 443 40. 4AS278 443 OA. 401 US4 40. 00 404 443 4.0 044 074 444 0.M 0.AM 1.0 04 040 A 00 00 A.40 40.0

- - ~ ~~ ~~ ~~ ~~ ~~ ~~ ~~~~~~~~- -1 6A A A - A -- - - -A -A - u - - . W A"- . 4w .. 4t.00in0010030041500)010011.0000 03I=4 :Mz ;V 240 040 0441A on- o2n1 04050 .0054 04014T

P-"0.t0 OwiObi 0.0 040 030 040 OM 040 Sm. Om am,

145000441043311020024010200 00400 037~O. 0.4 031 0.4 22 0051 022, 04041 0.0014500 08501 Ow wwO.03 01 .00 tow n 00 0.0 0.00AM0 30 -0 0A OSMA2050 0 vOw wn.22b 0 -0040 A0O 1.7t 01404 OM .75 an4 0424 an2 aw.,Aq, n02.00000002114d000 I5 0 OA,.7 0.0AS0 01.4 Os, 4.25 4.0 4.0.0000(40001b071005 14 02 .0 0.4 0"2 033 allow 00 04000I 500010010002050000001,004310001l.0WO 034 1.0 MS 0.0 0.0 034"4 4MA 0.22

Aq, . oOo oaaO4o)uw05.A4 o OA OM0 0.4a 0.0A 0.00 4W 4.2 4001

070000020 052200 7005017 104.009. 040 0OM 023 0.40 0.0 02 4 0IA4040 .Ca1001 051000440 tfl1l..4.0) 000001 I.05 00400 000 022 040 1.00 022 000 am .0221 Ca~~~~~~~~~~~~~~~~~~~~~~~~~OMM Ax"

150014001.4..43001o.00 0.0 02 0380.4 031 034 .0440 22 .0AM

0504 00.000 @22500.41000014 000~~~~~ 0.0 040 04 04 mk0 0.0o oW Smm

bw. - -~~~~~~~~~~~~~~~~~2 -M -M -, -2 -2

0.021100000003 4.04 ~~~~~~~~~~~~~00 A02 On 00a 0 0 4m .0012 .0001111400004 (0000050000150t 031 022~~~~~~~O 02 021 040 02 W.0 00012 000A12000100, 00500w41 0.02~~~~~~~IA 034 04 022 02 3 000001 01

VO0 10000 000 10000 10o00 1000 O ISWO 000 00000 C100 I0 01,1 Oil 900 li 9.1*,11- El" ol 001 0W1 Vi Il li Il Vi IL TVV "I 004*146j

0o MO1 W.0 MO W0 0,0 000 90 00 000 CF Co C 0 09 09a09 0 09 09 09 0 09 9 0 a9 09 09 09 09 go 00 00 09 009,o Woo 000 Ioo coo 000 00 .0 ' 0 100 Ln 0 C0C0 C10 0 00 90 s W0 I0 I0 0 x C0 10 40 0 9 0 W0 T9 0669i0119)00i06lK0 W00 900 WC0 900 010 010 010 01.0 000 01 k Vi I0 V 01 I W, W1 91 . I II II V II 0 0 l i i i i I , S1 V ii 60 9600

140

Ci4 ito LIE coo oIC I WI a .o n too, tO, M,0 0D111 I90 100 090 1199 009 WIWI 0190 WM0 0001 T11 VIM0 25L r00 0901, 600 0g9 rie WM099.oL tC Ol V 0V 61 01 ol ot O 111 60 00 lO 00 IO 00 Il t V9 01 000 £0 0 V19 10 11 00 110 ht619600a

a 01 000s , 1 000 W0 iZs I 01 0K 000 0900 BCll SCil C101 OIOO ",K 00L MtO 09 0900 001 e000 VI"00 EnI W'"t i* I 00 LN. M11 10 1000 Sli B999ior S0 00C0 1 0 0 0 000 C C S C h 901 0 S009V0 0 V 00 i0 90 ISc Ol £o 1 0 6 90 00 1£ 00669109

ao W0 W,0 090 000 900 Is0 00 W 001 00 £000 99991 MO '19" 10*1 9010 9110,0M9 *00 toilE0wow09Lo Kok O1'191 0100 0191 £10 Cl009 099 0910 t"o 64ile006WI

0 00 00 000 a tOO 00 W W0CtO W 100 W00 C 1 00 00 0a tO 10 W1 0 O O£ C a £0a 0 00 C6 10 9 W0 l 6006091W 9 Ws N I .1 . to 1'Z M I GrI s 0 cm "C, ' s, "N" 'S" I tz W , M 0M EZ I Se r m S , M, wt" czll Iz`V -tpo1

0 10 90 01 01 W 00 01 10 01 20 060 900 1*0 010 090 Oa 000 100 110 60 010 -it 000 000 ta lcO 010 0*0 000 tttOS180- 0o

W. .0 W 0 W. Irl, Vol- 10. 1.0- I 9 I c I r. a, St 't S c r. 0 I a c IC --rt - Fr -ri jw -~~~~~~~~~~~~099 09O9600

0 00 000 00 W W0W0 00N00 00 a0 00. W000 0 0 00 T0 00 00 0 0 00 00 00 0 00 00a 00 00 00 00T 001 00 00r010

0 W0 000 W0 WC0 90 000 900 000 W 09 Co0 19 099 00 I, 11 000 00 090 90 010v 0W,901 00 it, 101 000 01 a 0

0 90 900 000 900 000 900 900 00 90 - 00 00 00 00 00 00 00 00 v 000 00 00 00 00 00o 00 00 00 00 disk0r.0109109

W.00 000 000 W C0 0 90 000 000 00 I 01 I i vI II iiC 0 C 01 a 01 T To 01 ai I T.1 TO wo 01 01 0C00 1 0l

0 000 000 00 00W0 90 00 900 00 900W. 101 Oe, 911 011 011 01 q oii tl Lot To, 191 000 wil FiL, Oil Ilt Oil T99 191 C0000

£0i- cli- Clt- no- no, I-I- ci c T c-a- c TV01I 199 W.9 900 *00` 00 lUO 0,. 000 000 099 06 tiO r1 096 909 099 00 000 co9 100619900 000 900 900 90 woa0 00 W00 00 00 90 .1 01 O91 T 01 1 *1 01 01 I I 00 1 01 91 v1 01 01 01 01 IIz 69 00110. -q.1104 610

* CUT 0 we 1 0 0 M 0 9 0 00 00 00 10 90 0£ O 19 00 i 00 10 10 00 0 0£ 90 90 0109900 880301£ 0 il90 0£e 1 010w " 10 £10 91, 01£c 1* 00m 00 09 01£ 0 c 000 ee co 900 69rS 06 , 00 0£ 0cm i 00980 T00 900 00 TIT 0 0 00 90 90 0 60 0* 0* 00 0 0 90*V00L9*K90 60 r010 Else £00 00 tI 000 00S00TI0T 9 1 10 00 00 11 90 9 0* 19 10 6* 00 £0C00 10 Ot 0 60 900 00 0* 190 10 00 000 00 oa 6

t N 90 10 00 (O 10 0 100 r0 MO I On O C OF* , 001 6-cl to, M,1 191 01 I Tt £0 itt Fc 0 0 2 0 60I01i309

010

W o 00 0W0 W10 W10 110 W 010 01 11 01£ 1- Ca9 60 601- 101- T1 110 0 09 Toa- 009 60 6C01 096 100 90 L a C1- 09 00a £09 W

O1" 6M 001 W.0 0W0 962 0- to WC00 Vl0 , 10 , 09 9600 9191 0001 0"I M0 0991 091 L00 009VI 0010 Ol00 0101 5991 001 *991, 1001 0101 010 go,0)9090090

191 00 £11 000 El0 £00 610c 900 -1 el 0 -00- pIt 01- 6au- III. 09- 001- 01- 010- 001 T- 0wt. 109, 10 09 IC 0 0- g1,- 0£1 -CO,*vs,000,

*1~ £01 9101 69 001 00 000 09 000 001 am1 9 0m Z 9901 01011 6011 66601 6010 S00 CIL0 9000 000 F19911 M01 0901 0901 0W0M 60 601 001 £111966901

5VV W. 0 I are Wa 'r, LI Cro' oln reLe TWO VKL FELL FM O'Ni I M rlt, Z M "'S S~~~~~~~~~~~~~~~~~~110 10*0000IsV 0 9. 10 90 10 10 0 0 1 00 19 11 1009d0 II0 1 :S 001k0m91 99 00 0 00 00 £11 £000 r6ut 001 I11 00 010*600060

6 S 009 T 0 W1 01 S 000 901 e WC9 0* 60 *a99 1 01C*9 Cs M *60 09 I 009 001 10 091C0M90 00 09 009 01 61 00 91 001 10 01900011

0 C0 W W 0 0 0 0 W 0 . W s 01 91 0 II CC 0z 0, 0 W. 01 0D 9 1 9s, 0 0 0I o 0 I690001C4a, 0S 90C6Et9 101009909990Z E t r vk ```CZ- 'CM61 tI -Afl 9C0C1S

p9100919UI o09.ooaoo M i59 n '9101900 uooo SIDI~oe 0090 I'0O0q rs 0 C1 C's-dk-Cl 1ue remu s COL u90WI Z wzz0z1099 W1110 ISM06PQ M00, -860 C161W9 BlteoQ M"S lavm VW = Cm

O, a -eg P.-~~~~~~~~~~~~~~~~~~~~~~gojo

0100Q =0M =0a nto two Ca0 MO, MD no iln oiWO imoV *10, 100 CFO itO to MO 01 OnSiOnpli

MWO o iWO SI WIO WI NV MO wI 0l00InKOo Om. = WI nOV Wo Me no a"mo1001i0a. 12IC n,4ot iwo70 01 No0 WC1 W10 aon I4 01 0 4l00un

Iw 0m- moo no Ito ml it i"o WIA n o o o

000V nol mo 00 no NV no ano no 1 Omooim w olm

mOO, O mOpr no 010 NoI WI Ion no * 0 0 0 1 1 0K= meR

MWO, ome m~~~~~n no no no ano no ` R30CP`4oml.iItao. scko WI WIo no nc no no 10 -nd0Ipa

-o no. Woo it WIO 0 no no no 4-00 w0301 dp.l0olb.

MWOO W=0 WI it NV WI 01 00 00l0itt4* sb .0Qr,OmOo=0=0=Om WI Wo tO WI 'CO iO 0

f00shi0 qlMWO mea, Men, w~~o Wo et no n`O no 11A,bm m50OV OK0. SOm Me In0 GOV 000 01 ml OAnlm*a0Kbplo.,

OCCO 00 000 ano it in on 10P0(p-- d C=0 nllMk WI no no WI WI ill "4ifSC0eqa

1- V,- 91, ff SV, w ZI, ano, no.,& .1 not no no no ,WIWnovtCo 000.iqOOOinomVIC,00 sor, 10

ae. .0m MO V m M m - n c o CWO 00.V V 0 - -'00, WO, V, we _ Me _ _ W. Co -4

, " " 1 C O n 1 " V , C a wZVO, PCO W. 123 EV, $VP "W O, 0l, WO, woo M 'm M IV u s 1. M M rm Nk M u " tu M I. 7L0a001Mo. m. Mo Mo Me no m0. MI Mi MOII 000, "il 00001 0010 Iil0 Wmflu 'm - " roil "1 010 M01 Mii 011 "C0 Wia V.1 IwOompObOafmWA.

SW.T-mo Ill Ito. il Ill. oo` 9Z' al,l M toi LOll at1 Mti Mot ilil 0M PSi vOC oil M00 M01 101 0001 0000 M01 ow ti vCi oM So. no. o n no n n"' no~ MCo no- Mn oc too 00 0QC M0 M0 VN 0VW M0 M0 M0 tO MO M0 M0 M0 i0 100 M0 Od P%J0qd-OkI WI SV 97i , hO n r M!l- arl, 00 iL r 0 i a0 Ia0 CaO 0 M ito 00 000 a Ca? Cal M1 MW Wu Ml M1 WIC to M iTZOa000

mWc cm nro, noat0 o- WI ml 00 00a- 10 00 00 CM 00 00 00 0 00 I 00 00 00 00 00 in 00 r0mo. "o. moV- W O . 0o o n o WI 0 l I "'t ' i iofl gi M 1 to W 0 to to l*lg ti r ii ea 00 iOt 00 t too % mmlot itM o o. lt lt nono. mo. IWO Mi L- 0 0 O O 0 iU 10 0 10 0 0 t I ii it 1KCa 0 ( ntSO KO 00 00 WI tOO ill SI 00 ml 0Oi~~~~~~~~l IVI 00 001 L"1 001 MO Ot ill 11 000 000 tOO vi 00? 101 la (t t II zo

sO,t it~ tO - WI 000, e, SOt, tot K-?` Du 101 rat t li ot to to m0 to Mi Cat Ca1 M0 M0 In 00 v"0 001 it a q w 1 bo onq

'00 lo~ to- 10C M0 W. ao WI ill WC in in o In Ca Cn on -o in Ca i n C im in re in in to to on Om b OJKimO00 no Me0 100 WI. M W W V 0 1 01 WV I -0 opt "1"11 tt iI 0 "It itt 011 Gt (11 C,t tot 001 lOt On, 604 10. Wo 001 dS tWo.u MOt lit lb. o.0 100 Wv SV to 11 00 Ct? 60? Ol 10 On 001 "00 "I0 101 fig 60 0 101 IS4 It? "01 I01 00 L oolbon

ilo. iOo. wo., K K, WI' ml 0 no Me Wo to to to M to M0 VI0 to to o 10 to` to M 0 1 to Cto to"ag4 10 ihOPooa

st ano too. K.t ilo. 00 iM0C0I 00 l INoI s 9 W n 00 n 05 Ot to 0M Ito 10 ill too 10 009 Mn 01- Til 6-l10I00.io iii.0SO n 0 l iO n i I m m. to it p 0 I 0 o COO 00 OK1 00 0 0 0 0 (0 69 it tO io0O h Ois%

0.

100 0 . 1 1 "V M . M W Ii 01 "6 Ill0 It at i 0,I Ce 0T 0I al l ie,00A 0519 00i000 050010 00il010 0000100~~~~~~~OCIA" -:VA 151

tp.OotIoIl O-VKt410.. 1540011(04 u.1. SIOAIP 10F100qyIOKU0AoI Wa.O040004jqu P53100141100011041 MqP USFMIIOU1i00A031 Unweid S

6 OWL6

Table 10UUIPA. Maco.contomlc lndlamtom

5 percoot govemo¶t fltl fomsign debt reduction whit Increased educaton b034s182mtt

dASoieviton rm baseibse. unkis otharwis tindicated)0as V*k.2.4C2roVb DifrO-- 11ooeesn C.orreot d Bae, Vakes

7 o1 2 3 4 S l 7 ae o O It I 2 3 4 2 6 7 I 9 tO

20,02w t23. Z09.7 20447 SIlO 2ItO 71. 7777 221. 2M647 2173 7982. M13" 2249. 2302 277737 21491 219. 21I70 t2 Z07 4.70 A* 412 61 2.4 2.77 SS? C34 U 7Os.. =ot o227 1014 1227I17224. ¶312 1734 73J413tOit 13242 144 12321 12465 122l 130A 1312, 1334 1356. 1379 ¶41A 147 .12 120 t47 IM 120 117 2.75 22 2.5kw0 g s lE 7M.7 7-4 77A. 720 MA2 109 9321 MA3 445.4 O 724 78M1 72.4 79tA O&2 14.7 MU0 637.7 44.4 SM77 'A1 20 0,4F 2.0 3.2 IN7 3.8 2.2t 420 I7

52248 20. 2059.1 22441 22770 21101 2143.4 2177 2172. 217 211 72. 713t1 22522 W2 2775.7 2344 21U10 2217.0 22521 MOO6I7wafs, 9110~~~~~~"' 54.7 944 05912 OMA9 B 282 M724 tOt9 ¶021 iO am" 9151 935.2 5232 810. 929l I20O tOl I44I1 7 12 i's 720 7.42 1.26 ¶20 1.2 120 2.11Pie mISt . 40.0~ ~~~~~~ 441 4h 42.4 43. 4.1 40 451 496 44 000 40.A 4t 6 24 433 1 45.0 45A 221 020 020 0am 0 0am .2 020 OM 0amPOfbmrt 2471~~~~~~M 92132 Z46.4 271G 274 272. 21A 242 22MA M.5 226 fl" 202 272. 77M. 778. 2414 20*S 20 M2. 2.47 a.2 73 021 0.2 014 S 1.24 ¶7 1.2 Ip.o.40in4 ~~~~ ~~~~31A 32.1 my7 33.4 29. 34 7 14 Ml. M0 M. 341 312 42 361 371 47 3 0.4a U201 40 1 47 .12 2.72 2t3 213 2.52 2.7 3233 2.7 3.73 XI7

OIgOdo&W03 743.3 7562 0274 779. 7MA 20.3 211.1 MA1 22A. M34 743.0 741 792 m M"8. 442 817.1 221.0 2419 420 02 .42 .01 .011 .02 001 207 0.14

0-` 6 -71.0 W.72 .70.9 .740 .772 82 .40 1 .2&. .42. -ta .730 .47A .71'A T.74 . .4 .451 .2. 4 3.7M . .047 .042 .0.4 .0.3 .0.0 .61 .017 .073 .0.7E0 Ogof.8 sflt 7223 737 My1 724 7704 MIS. 797 MA1 812 7301' 73716 749.4 72 7232 782. 791.0 MIS 012.2 .4.20 .020 .021 '0.1 .0.11 .020 mmO 00M 2.03

7205W 4*2 . 43 ..0 .A791 T73. J.79 I24 .4" C 2 .4. 402 . .401 .277 .70.7 .701 M.77 .202 ~.4 41 3213 3.11, 3.10 BAS 324 3.23 3.61 216 220

NM$ I -roh. 1&707 Its 10.0 19. 1613 11 "A1 I61 170D 17. 1. Its 74.0 72.7 III 141 70 761 170 177 020 020 OM2 022 0am OM2 0. am 022OA 0 04 00 0A 0.8 010 OA.0 M00 00 00 24 0.0 00 0.0 0.2 0D . 22 am2 0.22 0.2 0220 0AO O.2 0M on2

ft1al 7 my 2 a. I7 772 ?.A 72. 6.7 61 M6A 79M 72 27A 711 s2 70J 4 MA1 VA M1A 123 0M 05 0.82 00 0.24 001 0.6 0.7 0.77 053

Chph.-222 k2,215.7 21.4 702 X 72.1 12 731 13 34 122 A 13. 2MA 19.4 12.1 13 ItsA Its 32 141 127 a .12 0.1 M2 M2 0AS 5.2 0.S 0am OAS

24s0.422. 443 4MA. 416.5 4477 a" 43L7 `uCA 44. SM. 439. 433 482.1 437 4121 MA4 40 44la 442.7 OS OAS 0.2 720 'A 1 0 70 10

82. 40.4 471A 4. a3 4.7 431 420 421 0. 4Q0. 40.4 411 MA1 I9 49.0 45* 04A 49 020 0.2 0.2 0. 5 0A2 0.2 OM 0.2bat ~~~~ ~~~~ ~~~314 417. J17 .4 440 34 23 W, 321 M 221 73.2 73.1 361 572 371 24.2 24.2 45.0 1 73a 3.13 3.13 3.13 3.13 3.7 3.73 3.13 3.7 I .73596s.a 3172~~~~~~Vt M7 30.4 SOL JL2 352 Nu0 12.7 2003 227. 32 30871 367.7 3670 53 320 362 O 24. 20C2 300MC OM3 OM IA4 7.07 7.12 1,21 7. M72 Ba206240 0.4~~~~~~~~L 7.7 7.2 72 7A 715 7o 7.8 71 7. U 77 72 7 7.4 7s 72 71 71 7 0200 OM2 0.24 0.24) 0M OM.24 0.2 22 20

ft19 W8 2 41.7 a". 44.0 6"1 W3 a" WA1 47.2 671 s Su. 82 31 6.047.2 ul 4 S n 14 "I7 "4A 0. 3.13 .0.7 .4.3 .0.3 .3.3 .213 .3.3 .3.1 .13.7

4451405 417~~~~~~M M2A 17.2 17.2 774 771 77A8 19.2 MI17 L7 7.7 419 771 172 77.4 171 171 "JD 7Ml :02 520 020 02 0.2 020 am0 M 0.22

4 7 -a ao' la 1. I0 0 t 1.1 I. 77 S. 7.7 71 . I 10 7 10 7.7 7.7 7.1 77I 7.7 1 1 020 Ba 0.2 0 020 020 020 0OO 022 a,s LO2.n d264 2 20 00 OA OD 0.0 'A 0.0 00 M G 00 04 20 02 0.0LO LO 20 8.0 020 am2 0 M OM22 02 020 W SAW 0.22

Tow 24371644 707 MO1 243 2292 SI1 277 3502 SIA 1 217 212 210. 302 542, 279.4 20A 3W7A 177 7.5 0.2 611 OM 011 OAS0 0.2 am2 01 0S82nt 27.0~~~~~~~~M 020 417 221. 232 24.3 274 42U 31. 21. 32.0 411 248.7 232 MCI 671 50. 241 022 022 0,20 020 222 0.24 0.24 022 222

26214~ MI GA 0.0 00 0.0 04 0.0 00 20 20 0.0 OD GA 20 00 GA 20 0 0 0. Ba 0 OM22 W2 Ban 82 020 022 020

391 0. 27A 3412 51 201 73 MI 51 SI SI S no0 272 201 365.4 SI MA2 3M5e 37. 2 0 0.12 0. 0221 720 0.41 4 611 202 0.64OsesyO 7341~~~~~~~M 2141 274U 21". 2M I MU 02931 44 1 22 213. 216 22. 211 212 3321 SI 3403A M M0 0.7 020 0.2 220 02 020 0.4 0.4 018.

on422) 274. 214. Z17. 2112 217.2 232 2M3. 447 2211 21 21LI 219 217. 21 2 2121M 3241 MM3 SIA M. 07 0 .24a 0.2 22 020 O.3 0.9 01 Q`4B 0s

026 2010.6 3.1~~~~~~~& 4 A 3.7 31 40 02 4. 43 3.7 3.452 3.7 20 44 42 42 41 4 401 410, 40,1 4.0.0 .U0 4001 O 4u 4A .0.01 0 0

477A NW0. 4470 10411 7)2 5447 7036. 187. 2021 2202 4720 M MA40 20. 74 12L S IS 20 22. 1241a 2100. 222 220) 720 7SM 37 LOS 414 L.4 47 7.4 70.3508O%S-4es 20 20 00 00 GO 20 OA 0.0 00 a 00 t 0.0 OD0 LO 00 20 20 42 0. A0D 020 022 0 02 020 OM2 0.22 020 G

54023s42o 4711~~~~~~. 2247 447.0 04' 7o342 1247 UPS4 76742 69 2247 472 Tz 20.2 7a241 6240 743t 2 1454. 724.3 725 2227 DOW 'M3 7M 7.77 320 418 518 2.07 714 ML0g0242 124.0~~~~~~la i20ax 7240 623 SOLD 100.0 U012. MA 3 70L ". 724 724. 736" 124 MA0 20 1123.0"la 124. 72 0.2 0ou 020 020 02 0.2 0200 020 020

2-S9f2 M&A. 2781 2262 202 C6M. 7572 74,1 727.7 '24l1 n2 3840 2121 724. 053 7072 72771 127. 1279.7 1. 022 12 US 2`07 207 44 2S8 013 720 ''7om 255221 411 0.2 M4 MI7 80A 511 72 7701 In1 MA2 xi7 4 "a0.0 2 97 24. 011 76. 715.7 020 020 02 048 0.2 03 0.24 0., 0.12

rael ftbloh- ~ 4711 Wu2. 60. 124. 32l.2 127 7224. 7474 AN0A 2567 4722 2390 0401 , 129 2 S 1252. 7244.1 7261 21445 2102 8 7M 713 L77 3.2 -A8 518 6.4 7.74 O

OW-ft AND~~24. 3612 424 20.4 4 SM 8001 0142 12A 11242 5t MA6 3672 4481 e. W247 SO70 19.1 7=3. 7150.9 .0.01 006 0.12 020 014 0.3 0. 24O. 7.77 712 7

Owps..27 ~~~ ~~~~~04 1.7 1.6 I A 71 71 1A 1.4 .A 7A 00 1.7 I 7 A 1 A 7.4 A 'A4 I A1 7A 020 001 O2 Ba 0200 0.24M 020 0M2as Oooise.sO.) 9.3~~~G 42 O GA 20 a4 0 07 0.? 0.7 0. 0.7 4J 42 00 20 0. 0.7 0.7 0.7 0 0.72 .00 001 7 001 001 001 001 0.01 001

- 71~~~~~~~~~~? 72 7.3 7.4 l1 7. ?A 71 71 7.7 20 771 72 72 74 71i 72 72 71 7 424 472 4.7 .021ll 406 4007 .02 4.23 422 & .0.04 051.4400 770~~~~~~,l .1.3 7IM '1.7 I77 S.S IsI19 I7 6 20 101 1.7 773 7M C.3 72? 714 1.79 7.0 7.22 727 Iowa 0201 0o1 L.201 0.21 0.0 0.20 0201 0.240 0.207

lM0A. Bfrucurhl.Poveota nd Ocoe lo0.bq3io.s dIooUU pervessg0ve7000ws0foselgn debt odsct),swMi loesend educaticonlewosbmo,

lAhrnlse dealtlon. h-, b.eel-ne un0.4 Ottlrelee IdlgoetdlBs.Vo0. C.,,e4VW0e 0e~000. se.404 .. V0.0 2 5 A 7 2 1 1 4 3 4 A 8 7 8 A 10 I 2 3 A 8 1- 8 a 10

488'480'OdoA) 10.8 ~~~~~~~~~~12.It 1o0I0 13.3 53 143 10. 15.1 55.0 112 527 13.013. 3 .010 44 14 IL' I0. .500 011 .00 .001 401 .001 .01 .00A .0I .05S.gc30.,07%~~~~~~~l48.e.AJ ~~~~547 013 021 02 UI 8U. Ho 042 038 548 W.o 0135 54. U.00 A N I50 54 54 00 54 30 Om 401 0A .00 .00 401 401 A01

0,ro.~.010.shlbnV0.a..8.o.s.. 00.0 '8.0 42 412 412 MA. VT1 W'4 30. 3NA 505 4. 482 4AAA 12 002 317 384 U. 344 00 0.00M OM 0.0 ON 0.00 0.0 0.0 0.0 0.0.~~rr4h1mr00l0.004nr005m-pe42 ~~~~~A. 0W. W0.4 4A 40.1 420 40.4 U.3 31 32 U.8 5U, 504 478 400 42 4.8 03 PS. 2 000 0.0 .0 .00 .005 401 AM3 ..I .01 .031E ,Ih. n%t.Or.4 70 313 0MA 003 402 432 0. 41 a01 47A 01 410 ml2 MA3 40.4 40. 40. 44 02 473 .0l 0.0 m0 0.0 0.0 0Om 00 0am 0 000eJIO@,0b05181(%dwOb.lOo,2007 001 80. a1 003, 002 10 70A0MO 71.701 U 831& MA. a1 Oa 008 10.4 70. "IA 717 40.0 Om 0.0 000 0.0 0.. A.0 0AM 0 0000I188100044,o1000Ilsol..30 7~~~~~~~~., 1 71 71 7.5 7.1 2 7. 2 2 72 01 71 71 I., 71 ?1 12 72 12 72 .0 0.0 000 030 0.0 0 3 030 030 03055.1004000t.70.007,0122 54 -3 12 -0. I17 11. 12 112 151 10. I40 13t 12. 10.11 118 I 113 -1 11.1 0M 000 0M a 00 0.0 03 00 03 0.00M 0300.uO.485m..44..p. 00 02 03 02~~~~~~~~~~~~~~~A 0.4 02 02 02 02 02 02 02 00 02 02 0 A.2 02 02 02 & M0 03 00 030 0AM 00 OM 030 00M 0-0

14

52400548N0k00nr48...etS 08~~~~~~A 0A 0 08 0.8 0.8 02 03 0. as8 08 0. 08 02 084 t a. 0 08 08 0a 08 00 00 04 0M 000 030 AM0 0.0 000 am0PS.1%00p00,00.sl 402~~~~~~~~~~~M M0. 02 '0, a, M0. ml2 02 012 2" 25A 00o MA. 00.7 00.1 A0 as. 012 V.2 02 00 030 0.0 000 030 00 0.0 0.0 030 000

"Ov.)l(an no8 733 my. 12 3 140 51 142 542 714 13 T1.7 70. 70A 70. 143 145 72 742 74 .1 O,. 0.00 0.01 030 001 0.0, 0.1 A3t 00P4WS000000004 ~~~~~~~~~~~~~~Ma 0.7 M?. W,. 007 ml Mn, my W0. W, a, a, 007 8. 00. W, 00.7 . 007 00. a0m0 000 000 030 030 000 0.0 00 030POO.00.078447 ~~~~~~~~~~~~~~211 12 412 011 13 200 as2 007 002 00. 012 01 012 01.1 00 84 0. a, 002 0.0 .4.07 035 0.0 030 030 0.0 000 0M 030Ofmro(dbsas00 lo2 8M, W, 003 MA0 0 OAA 02 00.0 00.2 002 001 N I 80. 0A M MA. M? 00 .000 .00 0.01t 0A 030 000 0.0 000 030 030F -~~~~~~~~~~~~~~~~~~~~ . 1 3 A A M n I V 3A V A M A M A M 0N 0W .. G m5M a

020% 402~~~~~~~~~~~~~~~~~O a, A8 AT. II 18.0 002 003 'A 0 002 40 a. A.Y MO2 'A2 a' n 00. 1Is 0.0 M4 .0m3 AM3 .03 .2 Om0 AM .-0 am70.0.15m008070. 0.4 03 0.a 2 1 2 I 2 02A0.4 OA2 0.7 0. l3 12 04 A41M00 030 .30 030 030 00 030 030 0W

00007%48~~~~~~~, 100.4 100.0 ~~~~~~~~101 200 la 1324310 0A12 10. 1012 10.4 M,0 "A1 1VA. MA0 132 A34 100 525. ".12 1 18 Q3, 405 405l 031 40 331 am1 0

- - GAS -M - -,baSe aeOn .0.14'einwo 030~~~~~~~~o 0*o 024 0.0 000 00 0.M 0.0n. Q

'-. 0080At. GAS44u *~ 0.0 0.28, 0.27 0.0 .0304 .4A 41MO9.5 00050 We .42018 030 ~~~~~~~~~~~~0,00 8.30 am0 000 O." 410= Os 40,

00 4)0M* A 0.24 oil0 M4 027 0"0 0.3040 .0) CA

1018000089010100O4.0040000 0.01~~~~~~O 0.40 a 021S0.00 GM0 02 OMS GOMIN8f 0.8 010I 0.0 0.4 0.n 039 0. 0.12 4013T OMM .0.001

100b141a80i 0n04 000 01 0.00 0.58 021 0..0 47 OsM 4Osw.e.. bIM. a mcM cm AM cM Os QOs OsI

1248001 W 445 0.00~~~~~~~~~~~ 0.00 a.0 03 an5 0 flow 03013 0MID~~~. 4 0 000~~~~~~~~~~~~~ U, 0.00 0 O".3 2 0.00005 aOs00a02sWe8.00~~~~~~~~~~p. 027~~~ 02 0.3 0.47 0.0 m.3 0.00n GOs 0.'COk-OIoo hIOM am03 0JS 03, 034 0.0A 0.0 000504QN 4100

r0 h.wawo. 08.00@ Gcm 021 AM cM 021 042 0.5s BAC OsVt000041.101 cm~~~~~-W~ A am cm 0.0 AN 0M Os O OWSn-~~~Opsr..a*0.~~~~~~~ c~M cM cM cM cm cm QOM O'COK OsO

834~~~~~48)0y30 0.00 0.00 0.00 0.00 0.00~~~~~~~~~~~~~~~~~~~~~~~~A cM OMM OsM Os0

cm a m cm .0 c OsM OsP 030MI .sheav.I4.5418. 0.34 0= 030 cm 0.34 035 Os Os Os~~~~~~~~~~~~~~~~~~~~~~CA240S 4CO59.0~~~~~~OsWe5~~~~~~.~~28V50I~~~~~~~ cm cm 030 cm 030~~~~~~~~~~~~~0 cm Os. s0' Os

Figure 1Production Process

t ggregate production|

Rural production Urban production

NontradedTraded Nontraded goodgood good

Formal Ifrmal

Exports riv Public Nontradedgood

Domestic sales

118

Figure 2Employment and Skills

Labor force

Rural l +migration_ *S Urban skills [Skilledlunskilledl unskille acquisition

raded ~~~Informal] Fra

Nontraded unemployment]Urban production

Rural production

[Labor demand1

119

Figure 3Financial Assets and the Money Market

Household savings

Financial wealth

balances Domestic depositsdepositsJ

M onetary----------- ---b World interest ratef anddepositrate I

Balance of O-D EndogenousBalance of ------Credit to| | t J ~~~~Exogenouspayments ! government ! E n

120

Figure 4The Credit Market

Domestic Domestic funds Reservedeposits for bank lending requirements

World interest rate! Foreign borrowing Lending to |land deposit rate by commercial banks government '

. ( Premium 1 =------- ~--------------------

Demand Foreign borrowing ILoan rate for cret - by firms

. i Net worth.,

Working capital Firms' borrowing - Endogenous

and Investment (net of retained earnings) C Exogenous

121

Figure 5Balance of Payments

Firms ak Government [Exports Jrd

Tcotal Foreign[mots] Debt foreign deposits

service borrowing by households

Crrent Captal

Balance of payments (official reserves)

0 Monetary | ) Endogenous

base J Exogenous

122

Figure 6Public Sector

Credit from i (Commercial bank ' Foreigncentral bank! I loans i borrowing

to government 'by government

Financeable Interest paymentsTax \ Government deficit on debt

revenues ----------

Lump-sum transfers I Non-interest . | t expenditure

Aggregatedemand Private

disposable D Endogenousincome

L Exogenous

123

Figure 7Link IMMPA-Household Survey

Survey 1 Shock to IMMPA1

Aggregate data Growth rates of income and tin 6 IMMPA categories , consumption for 6 categories

Apply to each individualRural . in 6 categories Urban

poverty line (new absolute levels) poverty line

Rut a2pPoverty and distributional indicators tRurl 1 Urbanprice index, (short, medium and long term) price index

Comparison with baseline scenario

124

Figure 8Initial Distribution of Income, Based on Log-Normal Approximation

(obs denotes the number of households in each group)

4.0 Workers in Traded Agricultural Goods Sector 2.0 Workers in Non-traded Agricultural Goods Sector

3.5 1.81.6

3.01.4-

2.5 1.2 -

2.0- obs= 50 1.0 obs=514

1.5 f0.80.6

1.0 0.40.5~ ~ ~ ~ ~ ~ ~~~~~~~~~~~~~~~.

OS- 0.2

0.0 0.0 0.0 0.2 0.4 0.6 0.8 1.0 1.2 010 0.5 1.0 1.5 2.0 2.5

3.0 SWorkers in Informal Urban Sectcr 1.2 Workers k Formal Urban Sectr

2.5 - .0,I \t2.0- 8

1.5 * obs 170 .6 obs 114

0.5 '4. 42

0.2 01 1 *1. 40

0.0 2. 0 0.8 2.0 . 00 0.5 1.0 1.5 2.0 25 3.0 3.5

0.8 -Skilled Workers 0.4 Capitalists

0.7 -0.4

0.6 - 0.3-

0.5 - 0.3 - u

0.4 - cbs 1ia 0.2 - obb1400.3 -0.2 -

0.2 -0.1

0.1 .* 0.1

0.0 .- 0.0-0.0 1.0 2.0 3.0 4.0 5.0 0.0 1.0 2.0 3.0 4.0 5.0 6.0 7.0

Note: Vertical lines represent poverty line. It is 0.45 in rural sectors ard 0.52 in urban sectors.

125

Figure 9Initial Distribution of Consumption, Based on Log-Normal Approximation

(obs denotes the number of households in each group)

Wcrkers in Traded Agriailtural Goods Sector 2.5 Workes in Non-traded Agicultural Goods Sedor

4.5 2orer -nInora ra et Workers in Non-rnade Agiutrban GodsSeao

4.0

35 - 2.0 -

3.0-1.5

2 i -ecbs W k505 10 o5 Ci 5i

02. . 10 15 2. . .00.0 20 40 60 80 1.

1.5 P 1

1.0 0.5 -

0.5 %,

0.0 *0.0 * ___________________0.0 0. 0.4 0.6 0.8 1.0 a.0 0.5 1.0 1.5 2.0

45 Workers in Informal Urban Secto 1.4 Workers n Formal Urban Sectcr

4.0 - ~~~~~~~~~~~~~~~~~~~~1.2

1.03.0-

2.5 0. Ib*10 L obs 114

2.0 - .

1.5: 0. 1.0 1. 2.0 0. 0.5 1.0 1 20 215 3.0

1. Skilled Workers OL Capitalists0.9 0.4-

0.8 0. O4 -'

0.7 -. 3 v

0.6 -.

0.5 c~~~~~ bs 120 0.2 -obs 400.4-

0.3 0.

0.2 - 0.1-

0.1 0.1 I

0.0 0.O _ _ _ _ _ _ __O_ _ _ _ _

0.0 0.5 1.0 1.5 2.0 2.5 3.0 0.0 2.0 4.0 6.0 8.0 10.0

Note: Verfica lines represent poverty mne. it is.6~4 in rural sectors and 0.46 in urtan sectors.

126

Policy Research Working Paper Series

ContactTitle Author Date for paper

WPS3071 Survey Techniques to MeAasre an Rita Pelnikka June 203 H. SaoExplain Corruption Jakob Svensson 37698

WPS3072 Diversity Matters: The Economic Somik V. Lall June 2003 V. SoukhanovGeography of Industry Location in Jun Koo 35721india Sanjoy Chakravorty

WPS3073 Metronolitan Industrial Clusters: SanjoyC hakravort June 2003 V. SoukanovPatterns and Processes Jun Koo 35721

Somik V. Lall

WPS3074 The Gender Impact of Pension Estelle James June 2003 M. PonglumjeakReforrn; Cross-Country Analysis AleJandra Cox Edwards 31060

Rebeca Wong

WPS3075 Child Labor, Income Shocks, and Kathleen Beegle June 2003 E. de CastroAccess to Credit Rajeev H. Dehejia 89121

Roberta Gatti

WPS30r76 Trade Re,'orm in Vin " Ietnam: Pi-ilippe 'uIFret June 2U003 K. ToIrrilinsonOpportunities with Emerging Challenges 39763

WPS3077 Do More Transparent Governments Roumeen Islam June 2003 R. IslamGovern Better? 32628

WPS3078 Regional Integration in East Asia: Eisuke Sakakibara June 2003 S. YusufCha'.Oenges and Opportunit lUes- Sharon vi amakawa 82339Part l: History and Institutions

WPS3079 Regional Integration in East Asia: Eisuke Sakakibara June 2003 S. YusufChallenges and Opportunities- Sharon Yamakawa 82339Part II: Trade, Finance, and Integration

WPS3080 Can Fiscal Rules He'p Reduce G uil!ermo Pe rn;V un 203 R. 'zquierdoMacroeconomic Volatility in the Latin 84161America and Caribbean Region?

WPS3081 The Anatomy of a Multiple Crisis: Guillermo Perry June 2003 R. lzquierdoWhy was Argentina Special and Luis Serven 84161What Can We Learn from It?

WPS3082 Financial Dollarization and Central Kevin Cowan June 2003 0. DoBank Credibility Quy-Toan Do 34813

WPS3083 Mine Closure and its Impact on the Michael Haney June 2003 L. MarquezCommunity: Five Years after Mine Maria Shkaratan 36578Closure in Romania, Russia, and Ukraine

WPS3084 Major Trade Trends in East Asia: Francis Ng June 2003 P. FlewittWhat are their Implications for Alexander Yeats 32724Regional Cooperation and Growth?

WPS3085 Export Profiles of Small Landlocked Francis Ng june 2003 P. FiewittCountries: A Case Study Focusing Alexander Yeats 32724nn their !mnlicatinns fnr I otho

WPS3086 Intertemporal Excess Burden, Derek Hung Chiat Chen June 2003 D. ChenBequest Motives, and the Budget Deficit 81602

Policy Research Working Paper Series

ContactTitle Author Date for paper

WPS3087 Gender, Generations, and Nonfarm M. Shahe Emran June 2003 P. KokilaParticipation Misuzu Otsuka 33716

Forhad Shilpi

WPS3088 U.S. Contingent Protection against Julio J. Nogu6s June 2003 P. FlewittHoney Imports: Development Aspects 32724and the Doha Round

WPS3089 The "Glass of Milk" Subsidy Program David Stifel June 2003 H. Sladovichand Malnutrition in Peru Harold Alderman 37698

WPS3090 The Cotonou Agreement and its Manuel de la Rocha June 2003 F. SyImplications for the Regional Trade 89750Agenda in Eastern and Southern Africa

WPS3091 Labor Market Policies and Pierre-Richard Agenor July 2003 M. GosiengfiaoUnemployment in Morocco: Karim El Aynaoui 33363A Quantitative Analysis