PoL icY RESEARCH WORKING PAPER 28 8 8 A Poverty Analysis Macroeconomic Simulator (PAMS) Linking Household Surveys with Macro-Models Luiz A. Pereira da Silva B. Essama-Nssah Issouf Samake The World Bank Development Economics Office of the Senior Vice President and Chief Economist and Poverty Reduction and Economic Management Network Poverty Reduction Group September 2002 [ PoiLic RsESFA CHI WORKIN INC A iP- R 2888 Abstract The Povcrty Analysis Mlacrocconomic Simulator (PAN'IS) A moidel that uses the labor model resuilts for each is a miodel that links standard clhouschold SUIrveys Withl labor catCgory to sililtilate the inlcomile growth for each macro frameworks. It allOWS users to assess the effect of individual inside its own group, asstuiied to be the iniacroccoinoimlic policies-iii particulair, those associated average of its group. After projecting incihVidlial inconies, with Poverty Reduction1 Strategies papers-on sectoral PAl\S calculates thle incidenice of poverty anci the inter- employment and inlcoIlle, the iclcidceICC of poverty, and group ineqUality. i ncomlle distributioll. PAN IS cani produce historical or couLnterfactuial I'AMS (in Excel) has three interconnected simtilationis of: components: o Alternative growth scenarios wvith different o A stanidard aggregate macro-framework that cani be assuL11ption1s for inflation, fiscal, and curreirt account taken from any macro-consistenicy moclel (for example, balances. These simulations allow test tradeoffs within a RNSNSM-X, 123) to project GDP, narioi1al accouLnts, the miacro stabilization program. national budget, the BoP, price levels, and so on, in Differenit combinations of sectoral growth aggregate consistenit accon nts. (agricultural or industrial, tradable or nonltr-adable goods * A labor market model breaking down labor sectors), within a given aggregate GDP growth rare. catecgories by skill level and econiomiiic sectors wvhose Tax and hudgetary tranisfer policies. production total is consistenut wiri rthat of the macro F1or examplc, PAMS will simulate a baseline macro- framework. IndciividuJals frolm1 the houLselhold SUrVeCS are sccnario for Burkina Faso corresponding to an existing grouped in representative grOups of households defined IMF/World Bank-supported program and introduce by the labor category of the head of the household. For changes in tax, fiscal, and sectoral growth policies to each labor category, labor demand dcpeiids On sectoral reduce povertv and ineCILIalitV more effectively than the output and real \\igcs Wage income levels by economiiic base sceniario. So, the authors argLIe that there are several sector and labor- category can thus be detetrminiied. In possible 'equilibria" in termlis of poverty and inequlality addition, different incoime tax rares anci diffetrent levels within the samie macro framework. of budgetary' transfers across labor categories can bc added to wvage income. This paper-a joint product of the Office of the SCIeior Vice P'residenit and Clhief Economliist, Development Economiiics, and the Poverty Reduction Group, PoVcrty Reduction and Economic Management Ncrvork-is part of a larger effort in the liank to provide better tools to evaluate the poverty inipact of econoimic policies. Copies of the paper are a\ailable free from the World Bank, IX IX H Street NW, Washington, [)C 20433. Please contact R{otila Yazigi, room M C4-328, telepholice 202- 473-71 76, fax 202-522-l I 58, emiail addr-ess r-azigi((c\ worldb)anlk.org. Police Research Working Papers are also posted on tIlc Web at http://econ w\orldcbaink.org. The authors may bc contacted at Ipeiradasiva(o \vcrldhaiik.org, hessamanssali() worldhan k.org, or isamake(o worldbai k.org. SeptcmibCr 20f2. (66 pages) The PoivcY Research Working P'a per Series disseLimiimates the finclings of iio-k iii pirogiess to encolur.iizg the exc-bPanige of idles/ a Pout dei clopmient issies. An objective (of the series is to vt t7he iWndiidNs oult quickl/. ueinii i/ tbe piresenitationis ire less thanl jul// polished1. 77Je paplrs car1n thv be iiies *i tbe aiutb,ors anu shouild he cited accordimkl),. Ihe I/Indiis. il terpretatl(iniS nid Co(,llclusilns expressed in this piper14 ,,ie e,itirelv those if the authours. heY dJ .1, nii necess.ril)v r he vi jew o/ te,, X,1 n,rld Ink, its Executive Directors, or the CeOUIntries thoe represetnt. ProdUced bv the Rescar-clh Advisorv Staff A Poverty Analysis Macroeconomic Simulator (PAMS) Linking household surveys with macro-models1 The World Bank, 1818 H street NW, Washington DC, USA Luiz A. Pereira da Silva, B. Essama-Nssah and Issouf Samak6 DECVP and PRMPR Loereiradasilva aworldbank.ora, Bessamanssah(Mworldbank.orp, isamaketDworldbank.orm Key Words: Macroeconomic model, Poverty, Distribution, Labor market, Evaluation, Economic Policies JEL Classification Numbers: C51, D31, E61 Hong-Ghl Min contributed to this paper writing Annex 3 (Labor Demand Elasticities). Alya Husain wrote Annex I (Policy Issues for PRSPs). We are grateful for the comments and support received from N. Stem, F. Bourguignon, P. Le Houerou, P. Collier, J. Page, S. Devarajan, and P.R. Agenor. A Poverty Analysis Macroeconomic Simulator- PAMS DRAFT Contents 1. Introduction and Motivation pg. 3 1.1. Objective of the PAMS Package pg. 3 1.2. Summary of the features of the Package pg. 4 1.3. Policy simulations addressed by the Package pg. 6 2. The Main Characteristics of the Poverty Analysis and Macroeconomic Simulator pg. 8 2.1. General features pg. 8 2.2. Sources of Tools and Data pg. 9 3. The Structure of the Labor and Wage-income Module pg. 12 3.1. Production pg. 12 3.2. Labor Market, Employment and Migration pg. 14 3.3. Prices pg. 20 3.2. Income and Expenditures of Representative Households (RHs) pg. 22 4. The Household Survey Simulator of the PAMS pg. 24 4.1. Income Distribution between RHs pg. 24 4.2. Projecting Poverty Headcount pg. 25 4.3. Social Indicators and Solving for IDGs pg. 26 5. Conclusions pg. 28 6. References pg. 29 Annexes Annex 1: Areas of Policy Change Most Commonly Examined in a Sample of PRSPs and l-PRSPs pg. 31 Annex 2: Extracting Relevant Information about Socioeconomic Groups from a Household Survey pg. 33 Annex 3: Labor Demand Elasticities for PAMS pg. 39 Annex 4: Linking PAMS with macro-consistency frameworks, Procedure for the RMSM-X pg. 45 Annex 5: Household Survey Simulator (HHSS) pg. 49 Annex 6: Application to Burkina Faso pg. 55 Page 2 A Poverty Analysis Macroeconomic Simulator - PAMS DRAFT 1. Introduction and Motivation 1.1. Objective of the PAMS package The Poverty Analysis Macroeconomic Simulator (PAMS) is a three-layer package of five simple, inter- connected tools operating in an Eviews - ExcelTm environment. Its main objective is to help economists / analysts conduct historical and counterfactual dynamic simulations i.e. project over a chosen time period the poverty and distributional effects of macro and structural policies on representative households (RHs) or socio-economic groups within a developing economy. The three-layer approach allows simulations of scenarios; comprising the macroeconomic framework chosen by a country, the employment situation and a projection of the country's distribution of income and poverty levels (through a top-down approach, i.e. from the macro to the micro levels). The simulation process consists in projecting the mean income of each of the several RHs of the economy, resulting from changes in the macro situation -either because of policies or shocks--, assuming that there are no effects to the intra-group distribution of income or expenditure. The package is designed as a "shell" that can host data from any country. The minimum requirement is a macro-consistency framework (for example a RMSM-X) and a household survey linked together as described in Annex 4. PAMS extracts information from the country's Household survey (HHS) and stores it in a particular format (described in Annex 5). Its operating principles are very similar to those of most spreadsheet based tools. The package uses five components that are inter-connected (see Figures 1 (a) and 1(b) below). * A maicro-consistency accounting framework and/or any macroeconomic model * A Labor and Wage-income module * A Simulator of Poverty and Distribution * A Household Survey (HHS) * A procedure to extract data from the HHS in a specific format (e.g., broken down by RA) PAMS is a simple tool to answer some of the questions about the distribution, poverty and social effects of Structural Adjustment Programs as well as of 'globalization". These questions became a practical operational objective for multilateral and other aid agencies2. Poverty reduction and pro-poor growth 2The current intemational debate on debt relief for highly-indebted, low-income countries: has led to associate the goal of sustained poverty reduction -now the main objective of adjustment programs-with demands for specific monitoring indicators relating poverty to other macroeconomic variables. In practice, on Dec. 22, 1999 the IMF and the World Bank endorsed the elaboration of a Poverty Reduction Strategy Paper (PRSP) as the central mechanism for providing concessional lending to low-income countries. One of the objectives of the PRSPs is to provide a Page 4 A Poverty Analysis Macroeconomic Simulator - PAMS DRAFT strategies require policy choices to be evaluated ex ante (and monitored ex post) for their impact on poverty and distribution. But few quantitative methods are presently available to meet that need. This paper proposes a method in between simple aggregate approaches and more sophisticated models3. PAMS keeps the simplicity of macroeconomic consistency frameworks used in many public and private agencies (e.g., RMSM-Xs or other country-based macro-consistency models). At the same time it goes beyond a poverty-distribution analysis conducted with aggregate relations between the economy's mean income (GDP per capita) and poverty-distribution-social indicators levels based on cross-section regressions4. But it stops short of being a fully disaggregated macro-econometric or CGE framewrok. 1.2. Summary of the features of the package PAMS simulates the income changes of various RHs for any given change in output growth disaggregated by sector. The insight (the 'technique") upon which PAMS built is the basic principle of decomposition in Bourguignon [2002], i.e. the change in poverty can be decomposed into two parts: the change related to the uniform growth of income and the change that is due to changes in relative incomes. Predicting the consequences of a policy affecting aggregate output growth on poverty can be done with this sort of technique, under the assumption that the policy under scrutiny will be distribution neutral or conversely assuming a specific quantifiable form for the distributional change. The solution that is proposed extends this relationship between macroeconomic outcomes (e.g., GDP growth, consumer price , inflation, employment) and the income of various groups in the economy, by breaking it down to various socio-economic groups and economic sectors in the same economy. The solution is a distributional dynamic process between several "typical" socio-economic groups using the RH hypothesis. Each RH is employed in a different economic sector. Hence, it is necessary to disaggregate the production side of the economy. In addition an explicit labor market is needed that reflects the skill composition of the labor force, the dichotomy between rural and urban areas, the effect of sectoral output growth and of real wages on the demand for labor. In a nutshell, the PAMS package: o takes a macro-framework from any macro-consistency package (RMSM-X, 123 or a Government model); country-owned, medium-term framework to reduce poverty and generate more rapid growth, with assistance from bilateral donors and multilateral institutions. The challenge now consists in providing PRSPs with the proper set of quantitative instruments enabling them to achieve their goal. See Bourguignon, Pereira da Silva and Stem [2002] 4 See D. Chen and A. Storozhuk [2002] Page 5 A Poverty Analysis Macroeconomic Simulator - PAMS DRAFT * takes the initial poverty headcounts and the income distribution from the household survey (regrouping individual observations into representative groups RHs defined by the labor category of the head of household); * disaggregates production into economic sectors to match the labor categories created from the household survey; each economic sector employs one labor category (one RH) only; * simulates labor demand and supply in a disaggregated labor market (with options for accepting or rejecting wage flexibility in specific segments of the labor market); hence determines wage income for each RH; * endogeneizes the price level (production price only) through a mark-up on wages, hence can project a poverty line accordingly; * simulates the effect of applying different (average) income tax rates across labor categories; * simulates the effect of applying different budgetary transfers across labor categories, consistent with the macro envelope for current expenditures given by the macro framework; * calculates income growth for each labor category; * feeds these growth rates into the household survey broken down by representative agents of each labor category; * simulates the new poverty headcount and the new level of inter-group inequality (Gini) There are two caveats for the approach. First, PAMS uses the macro framework of the macro model that runs on top of it. In that sense, it will inherit the strengths and weaknesses of that model. If the model is a RMSM-X, there will be no relative price effects on the production side of the economy (with the exception of real exports and imports reacting to changes in the real exchange rate RER). Moreover, using an aggregrated fixed coefficient production function5 eliminates from the discussion any substitution effect between factors of production coming from changes in relative factor costs. 5 One of the theoretical underpinning of the macro-consistency models used by the World Bank and the IMF is the Harrod-Domar hypothesis of a linear and stable long-term relation between the rate of growth of output and the investment-to-GDP ratio. The origin of the ICOR is Domar's celebrated 1946 paper, but a very similar approach can also be found in the central planning economic literature, largely inspired by an engineering approach to economics. Domars growth story posits a fixed relationship between growth and the share of (net not gross) investment to GDP (I/). Y, = oK,-, AY, = crAK,-,, =:- G7 AK-, = a7 I,, Y-I Y,-I YI-I Assuming that output (Y) is a fixed proportion of the stock of capital (K), that there is also no depreciation, a first difference transformation divided by output yields Domar's relation later also stated by Harrod. The a was also labeled by economists the inverse of the Incremental Capital-Output Ratio (ICOR), which measured the ratio of required investment to desired growth. A country with an investment-to-GDP ratio of 10% and an ICOR of say 4 would grow at 2.5%. In order to achieve higher growth, additional investment (hence more domestic savings) would have to be mobilized. As pointed out by Easterly, the Harrod-Domar story was not meant to be a relationship for the long-run but rather for short-term output changes in developed countries. Nevertheless, for a variety of reasons - lively described by Easterly [1997]- pertaining to the political economy of the Cold War and the directions taken by the High Development Theory of the 1950s, the ICOR remained for 50 years at the center of the design of development assistance. Page 6 A Poverty Analysis Macroeconomic Simulator - PAMS DRAFT The second caveat comes from the assumption (a single representative household or RH) used to determine income (wages, transfers). The simulations assume that the mean income growth of each RH affects homogenously all households in that particular group (e.g. there is no change in the intra-group distribution of income, e.g., no individual heterogeneity). Moreover there are no changes in the demographic composition of each of the RHs. For example, there is no endogenous shift between workers from on RH to another for those households that could "migrate" from one group to another given their characteristics and the incentives provided by relative income growth rates. These two caveats, however, can be partially "corrected" by the end-user of the PAMS. The flexibility of the Eviews-Excel environment allows precisely to construct simulations that do not need to be a simple "mechanical" top-to-bottom exercise. Some exogenous "additional" assumptions related to the "supply" side of the model can play a role in a carefully designed simulation. 1.3. Policy simulations that can be addressed by the new package Broadly speaking, based on an intemal survey conducted at the World Bank on Poverty Reduction Strategy Papers (PRSPs)6, the main policy issues which -according to the survey-- need to be evaluated --in their poverty and distribution dimensions-- are as follows: o What is the poverty-impact of specific changes in Dublic spending? How can changes in the delivery of public services, especially for health and education affect the poor? o What is the poverty-impact of specific chanaes in taxation? How can the financial and administrative burden of taxation on poor people be reduced? o What is the poverty-impact of improving Dublic expenditure targeting? How can public expenditure and revenue be better monitored and improved? o What is the poverty impact of structural reforms such as trade policy, privatization, agricultural liberalization and price decontrol? How could policy sequence these reforms? o What is the poverty impact of changes in the macro framework such as the fiscal, inflation and exchange rate targets? How can policy best deal with the possible trade-offs between several objectives? 6 The sample consisted of 4 full PRSPs (100% of actual, Uganda, Burkina Faso, Tanzania, and Mauritania) and 13 Interim or l-PRSPs (40% of actual, Yemen, Chad, Ghana, Cameroon, Kenya, Zambia, Rwanda, Cambodia, Vietnam, Bolivia, Honduras, Albania, and Georgia). The objective of the exercise was to identify in the sample what were the most common policies and instruments used for poverty reduction. The macroeconomic policy measures included monetary, fiscal, and exchange rate policies. The structural reform measures encompassed institutional changes (including anti-corruption, decentralization, tax administration, and budgetary reform), sectoral reform policies such as privatization, changes in tax rates, and expenditure increases/decreases in specific sectors. Page 7 A Poverty Analysis Macroeconomic Simulator - PAMS DRAFT • What is the poverty impact of terms of exogenous shocks such as trade shocks, capital flows volatility, changes in foreign aid and foreign payment crises? How can policy mitigate these effects? * Finally, what is the poverty impact of the quality of governance in its relation to investment and to growth (through the effect on the perceptions by private investors of the stability of the business environment in which they will operate, i.e. the 'investment climate"). What measures, policies can improve governance and productivity? Despite PAMS' simolicitv. there are some interestina macro and (some micro) Dolicy issues that can be addressed within this framework. PAMS can address some (but not all) of the issues listed above. The package can provide quantified simulations for the following policy scenarios: * altemative scenarios for GDP growth (policy-driven or external shock), including different combinations of inflation, fiscal and current account deficits to achieve higher poverty reduction targets; • altemative scenarios for pro-poor growth strategies emphasizing sectoral growth (agricultural or industrial) tradable or non-tradable (within a given GDP growth rate); . applying different rates of taxation to income by group (within the macro-consistent budget constraint); • applying different levels of social (budgetary) transfers to different groups (within the macro- consistent budget constraint). The paper is organized as follows. Section 2 describes the main features of the PAMS. Then, in Section 3, the main analytical relations of the Labor and Wage-income module are discussed. Section 4 explains the operation of the Simulator for the HHS. Section 5 summarizes some policy simulations based on the case of Burkina Faso. Finally in Section 6 we provide concluding remarks. Page 8 A Poverty Analysis Macroeconomic Simulator - PAMS DRAFT 2. The Main Characteristics of the Poverty Analysis Macroeconomic Simulator (PAMS) 2.1. General Features PAMS comprises (1) A base year household survey or HHS; (2) A macro-consistency accounting framework or a macro model (e.g. a RMSM-X); (3) A labor market model; (4) A household survey simulator or HHSS; and (5) a procedure to extract household data in a specific format (to construct the RHs) from the base year HHS. The first four of these five components are Excel worksheets. The fifth is an Eviews procedure that extracts data from the HHS, and stores it in an Excel HHS database in a specific format. This procedure can also be implemented using other software (e.g., SPSS, STATA, etc.). One of the features -by design- of the package is that each component can operate independently of the others. Alternatively, it can receive inputs from the others and simulate the impact of policies and shocks in a consistent way. The macro-consistency accounting framework or macro model (e.g., a RMSM-X or any other macro model available and used by the country) is the component of the package that provides macro consistency to the PAMS. This first layer, the macro model could be a general equilibrium model as well, or even a more sophisticated macro-econometric model whose coefficients and relationships are estimated with the country's time series data. This component gives national accounts consistency, in real and nominal terms (price consistency) and ensure that economic agents' budget constraints are respected at an aggregate level. The base year household survey or HHS is the component of the package that provides the information about initial levels of income and expenditure by economic sector of employment, skill levels, location (urban or rural) and degree of formality. It breaks down the total labor force into the categories that are needed to simulate the functioning of the labor market. Finally, the average wage and non-wage income of workers in each RH group will come from the latest available (and reliable) household survey. The labor market model is the component of the package that simulates the labor market linked to the consistency macro-economic framework (labor demand and supply functions can be modeled and elasticities can be estimated econometrically with country-specific time-series). First, the module breaks down the economy into two basic components: rural and urban. Then within each component, we distinguish the formal from an informal sector. Within each one of this sectors, PAMS defines sub-sectors Page 9 A Poverty Analysis Macroeconomic Simulator - PAMS DRAFT producing tradable are distinguished from non-tradable ones. This breakdown allows one to link each sub-sector of the production side of PAMS to each component of the segmented labor market. Labor supply is driven by demographic considerations and exogenous migrations of labor and skill categories. Labor demand is broken down by economic sector, skill level and location -rural/urban- and dependent upon the relevant sectoral demand (output growth) as well as real wages. Hence, the new module determines wage income broken down by socio-economic categories, skill levels and location (rural/urban). The module also features a sub-section on taxes, transfers and social expenditures (consistent with the macro model and the Govemment's budget). For each of the country's socio-economic categories (e.g., along the lines of a macro-consistent incidence analysis) it will be able to make average transfers or average taxation of that specific RH with a specific average tax or transfer instrument. It is also able to simulate the cost of attaining certain socio-economic goals, such as the International Development Goals (IDGs)7 with their 2015 targets, in a normative solving mode; and calculate which goals can be achieved given the country's macroeconomic constraints. The Household Survey Simulator (HHSS) is the component of the package that simulates/projects the effect of the labor market and the macro-consistent framework using the intitial information from the HHS. Since we have a starting level of income for each RH and projected levels of income after taxes and transfers by labor category (by RH), we are able with the Simulator to apply the projected average growth rate for each RH to all the households or individuals that belong to that same RH. Therefore, we can calculate income distribution indicators (e.g., Gini). With specific assumptions regarding the initial and projected poverty lines and assuming no change in the intra-group distribution of income, we can project absolute levels of poverty head counts. 2.2. Sources of Tools and Data A significant number of household surveys can be found at the World Bank and the relevant statistical units in Government. For example, there are relevant Websites such as the Poverty Monitoring Database, HTTP://NWW.WORLDBANK.ORG/POVERTY/DATA/POVMON..HTM that has six Main Components: * Household Surveys: 124 countries, classified by country, year or region. * News on upcoming surveys, studies and poverty assessments. 7 The Intemational Development Goals (IDGs) are targets that help frame the World Bank's business strategy and have been extensively discussed by the intemational community of donors. They are multidimensional benchmarks (income poverty, education, health, gender and environment). Their role in each country --and the capacity of that country to achieve them- requires careful assessment of the countries economic, demographic and institutional characteristcs. Costing should take into account these characteristics Page 10 A Poverty Analysis Macroeconomic Simulator - PAMS DRAFT o Social Indicators o Summaries of all poverty assessment by WB since 1993 o Basic Information on participatory poverty assessments by WB and other institutions. o Links to Other Relevant Sites RMSM-Xs. Can also be found as generic 'shells" that need to be calibrated specifically for each country case. The World Bank's DECDG site features a special menu area where typical RMSM-Xs, user guides and instructions can be downloaded. Alternatively, many Government agencies operate RMSM-Xs and/or other macro-models that can be used to ensure consistency. The labor market model is also a generic 'shell" that can be adapted to each new country case. It operates in a standard Excel™ worksheet composed of several separate spreadsheets (see Annex 4). There are several possibilities described in Annex 4 for connections with other macro-models and macro consistency frameworks. The new module can be hooked to the RMSM-X (Real economy, RX and Debt module, DM) but there could be other ways to generate the aggregate level of output as the starting point for the Labor and Wage-Income module. Finally, there is a need for a careful calibration of elasticities in the labor market model. This is described in Annex 3. Figure 1(a): Main Linkages of the PAMS using the RMSM-X macro framework RMSM-X Module or Debt Module Any other < Financing Flows AF Macro-consistency Framework, projecting Output AY growth and production Household technology a Survey Budget and Public Finance Breakdown by RA Labor and Wages SI,§MULATOR * - v 7 /< From Income .GroWth by. Labor and Poverty Module RA & Poverty Line Breakdown of Production by sector Simulates Poverty Labor supply and demand by RA category -Headcount and Wage Income, Taxes and Transfers by RA Dist-ibution intergroups Page 11 A Poverty Analysis Macroeconomic Simulator - PAMS DRAFT Figure 1(b): Diagram explaining the functioning of the PAMS Layer 1: Macroeconomic consistency framework National Balance of t Exteral Accounts Payments Resources Aggregate _Layer 2: Disaggregated Production, Labor Market and \ransfers Disaggregated Demand and Supply Disaggregated Production: of Labor (Unemployt): Transfers: * Agriculture * Agriculture * Social Expend. Tradable/Ntrad. By skill Educ., Health * Urban L (unskilled) . Taxation (Manufacturing, * Urban By group Services) l / ^ (Manufacturing, Tradable/Ntrad. I /Services) lI . z | ~~~Rkillarltl Inckilbl Tr| [PMinimum Wage a nd c lnntGie ategory | n PWages by category Layer 3: H ousehold Su sntapi v house ds l~~~~~~~~~~~~~~~~~~A PI Rur Etc Urban, ~~~~Povertv and Ineoualitv Indicators (Headcount, Gini. etc.) Page 12 A Poverty Analysis Macroeconomic Simulator- PAMS DRAFT 3. The Structure of the Labor Market model 3.1. Production In order to determine income by RH, one way is to match each RH group with a specific sector of production (i.e. like in Agenor and alii [2001]). PAMS distinguishes urban and rural production, formal and informal and tradable and non-tradable goods production. One reason for that is to argue that the production technology (use of labor and capital, mix of skills) is very different between these sectors. That, in turn, makes labor demand of each of these sectors different. Hence wages will also be significantly different. These differences produce the heterogeneity in the pattern of the overall income distribution and estimates of poverty. Figure 2: Production breakdown We use the following '-Y,- | *from -. | production breakdown. Gross LMac6.o- consistent_t FCJ--Domestic product (GDP) or Y is taken from the macro- consistency framework and is Rural Production Urban Production therefore exogenous. Then Y / \OO\ is broken down between rural Non-Tradable (2) Formal (YRUR) and urban (YuRB). Private Rural GDP is divided between 2 sectors (in parentheses, we Informal (3) * .X,E;from, bft \ put numbers for each sector): NiaX fro- aorditanro- Nnra! ( (1) the production of cash tc6nsi§tdnt F'....-> .... ---~ ''* { Non-Tradable (6) crops XRUR (tradable goods for exports) and (2) subsistence agriculture (DRUR). Urban GDP is divided between a formal and an (3) informal sectors. The formal sector includes the private and the (4) public sectors. Finally, the private sector is divided between (5) tradable goods (exports) and (6) non-tradable domestic goods (i.e. the sector number (3), the informal urban sector is also a non-tradable goods sector and also private). There are in total, 6 sectors and assuming all prices normalized to one, the accounting framework becomes: Y YRUR + YURB (XRUR + DRUR ) + [(XURB,PRIV + YURB,PRIV ) + YURB,PUB + DURB Page 13 A Poverty Analysis Macroeconomic Simulator - PAMS DRAFT To keep the simplicity of (and the linkage with) the macro consistency framework, the aggregate production of this economy (total GDP or Y) is derived from there and residual sectors such as the informal urban sector (number (3)) and the rural subsistence agricultural sector (sector number (2)). will ensure the overall consistency. Similar linkages with other macro models can be envisaged as well (e.g. macro-econometric models or CGEs). The expDirt sector of the economy is divided between agricultural exports and non-agricultural (urban) exports. For both, the level of production is exogenous8, dependent upon foreign demand (Y*) and the respectivea real exchange rates for each sector based on the relevant domestic and foreign prices. X = (XRUR + XUROXR1V) XRUR =C Y, EPRUR PRUR,D XuPM, = x Y EURB, PUR8,Y In the rural economy, there are several options to determine output. One is to take agricultural production as given by the RMSM-X. Another is to model rural production separately. Under the latter option, the simplest specification is to calculate YRUR using a constant elasticity (JRUR ) of output to rural labor. In a more complicated specification, there could be complementarity between factors of production and infrastructure (public investment) such as roads, etc. In such a case, rural 'production technology' depends also on public investment (IGR ) in infrastructure in rural areas, measured on a per capita basis. The reason could be that a minimal level of infrastructure (say rural feeder roads) is necessary to make non-subsistence agricultural production profitable. However, for public investment in rural areas to be effective for development, a minimum level is required, below which retums are zero. The elasticity ( RUR ) is positive. One of the specifications below can be chosen. YRUR R RUR yRU *- rLRUR YRUR -RUR RUR Y A Y (I G 'RIJA L6RUR RUR * RUR k RUR J RUR 8 In the version of the PAMS linked to the RMSM-X, exports (tradable goods sectors) are determined in the Trade sheet of the RMSM-X model. The functional form of export determination, however, follows a traditional demand specification -using the small country assumption-- dependent upon foreign demand and the real exchange rate. Page 14 A Poverty Analysis Macroeconomic Simulator - PAMS DRAFT Once XRUR and YRUR are determined, non-tradable rural output or subsistence agricultural output (DRUR) can be calculated as a residual. In the urban economv, the breakdown of production is the following: Total urban GDP (YURB) production is calculated, with the production of tradable (export) goods defined earlier as XURB,PRrv and the public sector product being exogenous and fixed. YURB,PUB = YURB,PUB- Y -YRUR -XURB,PRIY YURB,PUB (YURB,PRIV + DURB) Similar options exist for the urban non-tradable and formal private GDP (YURB,PRIV) regarding the choice of a production function. One solution is to use a "private urban" incremental capital output ratio (ICOR). All private investment in the economy (I = lIURB,PRIV) takes place in the formal private urban economy where there is all the private capital stock (K). The growth rate of output in the urban economy is therefore given by a fixed-coefficient relation to the ratio of investment to output. Other options would include modeling a specific functional form for private urban investment. AYURB,PPIV /URB,PRJ, I y~~~ 0URB- YURB,PRIV,-I U URB,PRJV,-I Hence, after determining the 5 sectors, it is possible to determine the output of the 6h, the residual of urban output i.e. the informal non-tradable goods. Calculating total GDP minus total agricultural output minus the urban private production of tradable goods minus the public sector, minus the private production of non-tradable goods gives the production of the informal non-tradable good sector DURB as the residual. Where in the labor market model could relative price affect supply decisions, i.e. resource allocation? It is easy to show that these effects could be introduced in the rural and urban formal private sectors. 3.2. Labor Market, Employment and Migration Employment determination in the labor market model follows the breakdown of the economy into its real components, with the additional dimension of the two types of labor (skilled and unskilled). The departure point for modeling labor supply and demand is the breakdown of production. Paae 15 A Poverty Analysis Macroeconomic Simulator - PAMS DRAFT Figure 3: L.abor Demand breakdown The total labor force is a iEL'abo Population Supply fraction of total population. Total Labor Supply There are two labor Demand categories, skilled and unskilled only. Each sector Rural Production . 3 * UrbanProduction on the production side, is Foral assumed to hire only one _ - _ / V,% \type of labor, skilled or Private _ \ unskilled. There is no production process in this // \simple model that employs both categories of labor and .______________________________________________ there is no substitution between labor in two different sectors except for a possibility of exogenous migration that follows a Harris- Todaro-like process. Hence employment is divided between unskilled labor employed in the rural economy (in both the export and the subsistence sectors), skilled labor employed in the formal export sector of the urban economy; unskilled labor employed in the non-tradable sector of the urban economy; unskilled labor employed in the informal sector and public sector employees (which are assumed to be skilled labor only). Writing (as below) subscripts RUR and URB for the sectoral origin of demand, the superscript D for Demand and superscripts UNSK and SK for unskilled and skilled labor respectively; and subscripts X, for tradable, D for domestic informal, non-tradable, Y for domestic formal, non-tradable, and G for public sector, we can decompose labor demand into its components according to the production side of the model. L-LD +LD -{D,UNSK + D,UNVSK ) + tD,SK + D,UNSK +D,MNK \ D 1 L'= LDR + LuR =. (LRu J45x + + + L 4~K+ Lf,K)LG Y RJ R RUR , RUR,D J -URB,X URB,Y URB,D J + URB,G The rigidity of this representation of the labor market can be amended in a couple of ways. First, while each sector hires only one type of labor as depicted in Figure 3, and labor categories are 'pre-assigned" to the relevant sector, the starting wage rates in different sectors for the same level of skills are different, allowing for a differentiation of average wage incomes across sectors. Second, there is migration between the rural and urban economies, and between skills categories. The process is not modeled explicitly but left to the judgment of the end-user of the PAMS package. Finally and third, as we shall see below, unemployment will affect real wage rates differently and introduce more differentiation between the wage income of the various categories of labor in the model. a) Employment, Migration in the Rural Economy. Page 16 A Poverty Analysis Macroeconomic Simulator - PAMS DRAFT The rural economy produces (YRUR) tradable (exports, cash crops) and non-tradable goods but employs only unskilled workers. Employment in the rural economy follows the Lewis tradition of a situation of "unlimited supply" of unskilled workers growing with the population growth rate I(POP). However, migration flows from the rural to the urban economy need to be subtracted. LUSK = LRUR K .[1 + 77(POP) - MIGR] Labor demand in the rural economy depends positively on both components of rural output (YRUR = XRUR + DRUR) with an elasticity Of 0ORUR (which could take a unitary value hence making a labor demand per unit of output a function of the real wage) and negatively on the real wage rate with an elasticity aRUR* Since the rural sector comprises only unskilled workers, workers will opt for being employed first in whichever sector offers a higher real wage. The export (cash crop) sector has higher (real) wages WRURX than in those in the subsistence agricultural sector UNSK . The basis for real wage setting is a fixed minimum sectoral UNSK subsistence wage WRUR,X that adjusts if wage are assumed to be flexible (see below). The nominal (product) wage is the product of the real wage by the sectoral producer price. D,UNSK - uNsK -aRUR,X LRUR,X = RUR,X RURRu' * RUR,X w UNSK UNSK RUR,X = PRURWRUR,X UNSK WUNSK WRUR ,X =L+ JWRUR,X-1 or w flex & URUNSK > O > WRUR ,X adjust The supply of unskilled labor in the informal subsistence agricultural sector is the residual of labor supply minus labor employed in the export sector. A similar wage determination mechanism is introduced in the subsistence agricultural sector. However, there is a higher probability there that the real wage rate is flexible and adjusts more rapidly to clear the market following a wage-curve specification. Alternatively, the model can also feature other types of specifications for the rural informal sector: either assuming real wage rigidity or an instantaneous wage clearing specification. Page 17 A Poverty Analysis Macroeconomic Simulator - PAMS DRAFT Ls=UNSK Ls UNSK -LoUNSK RUR,D RUR RUR,X L D,UNSK _ D RIUR,D.WUNSK -aRUR,D RUR,D )RUR,D RUR WRUR,D 'UNSK US WRUR D = + WRUR,X-l or w flex & UR7URD > 0 =' WRURD adjust MIGR = MIGR w UNSK _ UNSK RUR,D PRUR WRUR,D Migration from the rural to the urban economy follows the Harris-Todaro tradition. Unskilled labor moves to town -at no cost and at the rate MIGR-- attracted by the (expected) wage differential between the rural and the urban economies (which traditionally depends on the perceived probability of getting an unskilled job in the urban economy). b) Employment, Upgrading Skills in the Urban Economy. The Urban economy is divided between a formal and an (3) informal sectors. The formal sector includes the private and the (4) public sectors that employs only skilled labor. The private sector is divided between (5) tradable goods (exports) employing skilled labor and (6) non-tradable domestic goods employing unskilled labor. The sector number (3), the informal urban sector is also a non-tradable goods sector and employs only unskilled workers. Labor demand in the public sector is exogenous. Given the advantages (fringe benefits and perks) associated with public sector employment, workers will opt to be employed first in the public sector. The civil service employs a fixed number of skilled workers only that are subtracted from the urban labor supply. L D L D,SK URB,G URB,G The rest of the urban sector is the private sector. There, we specify labor demand for the formal sector and then leave employment in the informal sector as a residual. The other simplification that we make is to assign unskilled labor only to the formal non-tradable goods sub-sector and skilled labor to the tradable goods sub-sector. Hence, labor demand for unskilled (respectively skilled) workers in the two formal sub- sectors oiF the urban economy depends positively on the components YURB,PRIv and XURB of urban output Page 18 A Poverty Analysis Macroeconomic Simulator - PAMS DRAFT (YURB) with their respective elasticities (wuu Y and wu,,,I) and negatively on the real wage rate with an elasticity auAK (respectively a-j)- D,UNWSK _UNSK yWuu.Y UNSK -uB LURB Y = URB,PRIV U1IB,PRIVWURB,Y D,SK SK Xa sw SK -asK LURB Y = KURB,PRV URS WURB, Y The supply of unskilled (respectively skilled) labor in the urban economy can be modeled in several ways. In the simplest, current specification it grows with the rate of population growth, the rate of migration from rural areas (for unskilled labor only) and the rate of upgrading unskilled workers. In more sophisticated models, it can also depends (as in the two specifications LI and L2 below, on the wage premium (for skilled labor). We also assume that there is no 'skilled unemployment' in the urban economy. Skilled workers that can not find a job at the prevailing wage do not 'downgrade" to the unskilled segment of the labor market. They rather stay idle (voluntary unemployment) waiting for job opportunities. Altematively, one can also introduce an equation for 'emigration" where unemployed skilled workers will leave the country and find job in foreign labor markets. In such a case, the supply of skilled workers would be reduced by a rate of emigration EMIGR that would depend on skilled unemployment and the difference between expected wage abroad and prevailing wage for skilled labor in the domestic economy. AesK =L1 ( \, sK)X7R(o)pRElR ,SKN L5 USK )' UNSK )S7URBJ (POP), IGRUPGR, EMG LsURB, -I LWU", -I WURB2 UPGR = UPGR~ S EMIGR UPGR]G AJIUNSK =L2 I UNSK UNSK 77URB (POP), MIGR, LURB, -I I..WURB,-1J WUJpB UPGR = UPGR EMIGR =EMIGR SKUNSK _ URB__ _ V D,SK,UNSK URB The determination of wage rates in the urban economy for unskilled workers is as follows: Page 19 A Poverty Analysis Macroeconomic Simulator - PAMS DRAFT The nominal wage rate in the public sector is exogenous. It is also possible to add a specific condition where public sector wages for skilled workers are set to be above (or below) comparable private sector wage rates. WG = WJ WG = PURB WG > PuRaBwUR Unskilled workers will choose to work first in the formal urban sector, at its given wage rate (assumed to be always higher than that of the informal urban sector). Unskilled workers will then tum to the informal urban sector job market. We assume that there are frictions in the formal urban labor market and that adjustments there can be more or less sluggish, thus generating involuntary urban unemployment. c) Wage determination for unskilled workers The real wage determination for unskilled workers depends on the following considerations in both the rural and urban sectors. There is a 'minimum historical subsistence" wage level for unskilled labor that is set by institutional arrangements (e.g., unions bargaining power or a benevolent Govemment or both). Two forces pull in different directions. On the one hand, unions push for a regular increase in the "minimum historical subsistence" wage level for unskilled labor in urban areas. On the other, the flux of migrant workers tend to increase the supply of unskilled workers and hence to depress the real wage (by increasing unemployment). For each of the 'sectors" of the economy, and for unskilled labor, UNSK { UNSK \-6 1 UNSK WSECTOR iECro, qUVSEC0 E-SC0 If k SECTOR =0, the wage for unskilled labor is fixed at its historical subsistence level, WUENR plus whatever the increase E obtained by unions. If x SECTOR =1, the wage level is related to the level of unemployment through a wage-curve type of relation (Blanchflower and Oswald [19941). Alternatively, one could use a specification where it is the chanae in the wage rate that is negatively related to unemployment (e.g., a Phillips-curve type of relation). Then, the nominal wage becomes: WSECTOR PURB WSECTOR Page 20 A Poverty Analysis Macroeconomic Simulator - PAMS DRAFT Now the wage of the residual informal sector has to be determined. The sector is a residual for both production and employment. We opt here for the same type of adjustment. Here too, a market clearing wage can be used. Finally the wage rate in the urban economy for skilled workers follows efficiency wage considerations to create incentives for skilled workers to remain in the domestic economy and avoid shirking. Hence, employers are prepared to pay a premium for skills over unskilled labor wage rates. One of the reasons is that skilled labor is a closer substitute for capital. However, in our simple framework, the determination of the premium can not be based on the possibility of substituting skilled labor by capital. Nevertheless, upgrading unskilled labor would aim precisely at making it more substitutable to capital. Hence, the user of the PAMS framework has to rely on the information of the HHS to proxy the premiums between labor categories.9 d) Skills acquisition and upgrading of labor Skills acquisition depends on expenditures -private and public-in education (see below). But this can only occur in urban areas (e.g., there is no skills upgrading in the rural economy). Skills acquisition allows unskilled labor to join the skilled labor category in the urban economy. In order to acount for structural changes in the economy, coming from changes in the composition of the labor force (skills), its allocation across sectors (sectoral labor demand) and the relative shifts in the structure of production, PAMS re-weights the number of households belonging to each RH from the original sample to reflect the sectoral structure of production and employment in the simulated scenario. Notice finally that this framework is simple and assumes no substitution between labor categories and sectors other than the ones that can be exogenously inserted into the simulation. Other options for the macro and labor models are possible (see for example for South Africa, Fallon and Pereira da Silva [1994]) where a specific two-level nested CES production function allows substitution factors of production and determines factor prices (real wages, price of capital) accordingly. One way to proxy the workings of a production function Y = f (LUSK,LLSK, KD with three inputs (unskilled labor, and skilled labor substitutable with capital), is to consider that the premium that employers would pay is at minimum equivalent to the rental of one unit of capital. The rental of one unit of capital can be proxied by an opportunity cost such as the rate of profit (e.g., the economy's profits (PROF) or the retums on the stock of capital). However in the absence of information on the stock of capital (K), we proxy rate of profit by assuming that it is equivalent to the domestic lending interest rate r plus a commercial risk premium (p). Therefore, wK,= (I+ r + P)WuRN.Y (r+ p) (PROF) = i Page 21 A Poverty Analysis Macroeconomic Simulator - PAMS DRAFT 3.3. Prices The Labor and Wage-Income module is run under the umbrella of the macro-consistency framework mentionecl above. The general (GDP) price level of the macro-framework applies to the aggregate production. However, there are tWo endogenous determinations of changes in the price indices allowed once an initial price level is chosen. Export prices are exogenous (following the traditional small country assumption). If E is the nominal exchange rate, and p* the foreign currency price of exports, Px = E.p*. In the rural areas, the price index is the weighted average of cash-crop prices and the prices of subsistence agriculture. The weights are their contributions to agricultural GDP. The change in the price index in the subsistence agriculture sector is a mark-up over cost components which include a weighted average of the minimum subsistence wage for the informal rural sector and formal rural wage cost. , XRUR DRUR PRUR = PRUR Y + PRUR,D RUR RUR PRUR,D = ((RUR,D91)1[ AwRUR,D +02AWRUR,Y) oJ + 02 =1 In urban areas, we assume an identical procedure for determining price indices. XURB (Y - YRUR - XURB) PUR8 ='PURB + PURB,Y (Y YRUR ) (Y YRUR ) PURB,D = (PURB,D-1)[1 + (A WUM,D + P 81 + 2 =P1 Therefore, in nominal terms, GDP (pyYy) can be expressed as the sum of nominal agricultural production (PRURYRUR), nominal exports (expressed in local currency, pxYx) and the nominal value of domestic goods produced at the given domestic price (PDYD) of non-tradable. Alternatively, the aggregate price level of the macro consistency framework (say the RMSM-X or another) can be used and one of the sectoral price levels would adjust residually. PY P.RURYRUR +PXX + pDD PXX PRUR,XXRUR + PURB,XXURB PDD =PRUR,DDRUR + PURB,YYURB + PURB,DDURB Page 22 A Poverty Analysis Macroeconomic Simulator - PAMS DRAFT These two RUR and URB price levels will be used to project the poverty lines in the rural and urban areas. 3.4 Income and Expenditures of Representative Households (RHs)'° a) Representative Households, Wage Income and Profits We have now determined nominal wage income and employment (hence sectoral wage income) for the following categories of workers which constitute our i - representative households for this economy (i = 1 to 6). Rural unskilled workers of the tradable goods sector: WRUSK Lx UNSK X i = 1 Rural unskilled workers of the non-tradable goods sector: WRURI/ D.Lx,UN iK = 2 Urban unskilled workers in the non-tradable formal private sector: W.UNSK LUNsK i = 3 URB,Y IUBY Urban unskilled workers in the non-tradable informal private sector: WUNSK LUUSK i = 4 Urban skilled workers in the tradable sector: WSK X.LSK i =5 Urban civil servants (skilled): W .LURB,G, i = 6 In addition, there is seventh non-working group that receives income in this economy. Capitalists and rentiers get non-wage income or profits PROF = prY - E Wi Li , which is the difference between all income generated in the economy and wage income. We assume that there are no financial assets in this economy held by non-capitalist groups. There are, however, many ways in which this assumption can be relaxed if the distribution of financial assets is known (say from a detailed household survey). For example, the interest revenue from the macro-consistency framework could be split between various groups. 10 In Annex 5, we depart from the rule explained so far of 6 groups of RHs plus a 7^ group of 'Rentiers', by adding from the Burkina Faso case, used as an example, 'Self-Employed", 'Unemployed" and a small category of skilled workers in the Urban Non-Tradable Urban" sector. Page 23 A Poverty Analysis Macroeconomic Simulator - PAMS DRAFT Therefore, for this economy, the distribution of income by group is known and one can draw indicators of inter-group income inequality. b) Disposable Income, Taxes and Transfers Each RH pays income tax at a category-specific average rate Xr over its respective gross income. Each RH also receives lump sum budgetary transfers T from the Government's budget. The Govemment initially is not capable of targeting the transfers T and therefore simply provides them on a per capita basis. Hence, disposable income is composed of wage income (or profits) plus social transfers from the budget to a specific labor category, minus taxes paid by that specific category of labor. DINC, = (li - ri )(Wj Lj + PROF)+ TL 7 DINC = DfNC, ,=, For consistency purposes, the sum of income taxes paid should be equal to the Govemment's budget's total incorne tax (and checked against the figures that appear in the macro-consistency framework such as the RMISM-X). A similar consistency check has also to be done for the total disposable income. The breakdown of total income between its wage and profit components should also be consistent with national account identities in the macro framework. c) Expenditures, public and private Each RH has also a structure of expenditures once its disposable income is determined. In particular, we are interested by the complementarities between its expenditures on specific items such as education, health, social services and the same expenditures by the public sector. We assume that the social outcome on these social sectors will depend from both private and public spending. The sums of both private and public expenditures on each specific item (such as education, health, etc.) should respect private and public budget constraints, included in both the household survey data, the budget data and the RMSM-X consistency framework. Page 24 A Poverty Analysis Macroeconomic Simulator - PAMS DRAFT J DINCG y/42/z g Income Income I RH, to shifting the entire distribution ______ =_ c-ts a atyo , aof DINCs to the right or to the left, depending on the macro-result of the policy or the shock. For example, in Figure 4, if the distribution of income for all households inside a specific RH can be proxied by a normal density function, what the simulator does is to re-calculate for the whole economy (the 6 RH) the resulting changes in the inter-group poverty and income distribution, after changing the DINC of each household in each RH with its proper growth rate. Page 25 A Poverty,Analysis Macroeconomic Simulator - PAMS DRAFT 4.1. Income Distribution between RHs Once we have established the disposable income for each of the 6 representative groups of this economy (and as a residual, the income of the "capitalists" and/or 'rentiers"), it is easy to calculate a Gini coefficient measuring (disposable) income inequality. The Gini provides a measure of the distance between the perfect equality curve (Gini=O) where each group receives a share of income exactly proportional to its population size and the Lorenz curve obtained by the actual cumulative incomes of each group. Gini = Z JNC-DfNCJ, i,J=1,2,3,4,5,6 n=7 2.n 2 DINfC, i j = E DINCi DINC. = i n In addition, one can also calculate a Gini for Rural and Urban areas separately. Also, note that the above inter-group Gini may be computed as twice the covariance of the mean income of each group and the group's relative rank, divided by the overall mean income" 4.2. Projecting Poverty Headcount PAMS is able to measure the effects of changes in policies and macro variables on the disposable incomes of each of our RH groups. Then, the framework uses household survey data to estimate the changes on the poverty headcount and the poverty gap. Let us assume that we can define Poverty lines for the rural and urban sectors. ZRUR =Z(PRUR) ZURB = (PURB ) The labor market model generates growth rates of disposable income for each of our 6 groups and the group of capitalists and rentiers (6+1=7 groups). There is also information on household incomes or expenditures from a standard household survey (HHS). The PO and P1 indices are projected by linking See Yitzhaki and Lerman (1991: 322) for details. Page 26 A Poverty Analysis Macroeconomic Simulator - PAMS DRAFT the results of the PAMS projected mean income of each of the 7 income groups to the income/expenditure levels of each household in the whole sample of the HHS12. The following are the key assumptions for the exercise: The head of each household in the HHS belongs to one of the labor categories of PAMS. The HHS data provides the specific number of household members. PAMS assumes (as explained above) that the overall income or expenditure of each household grows by the same growth rate as the mean net income (minus taxes plus transfers) of the category to which it belongs. Thus the assumption here is that the distribution of income within each of our 7 groups is unchanged. Once this is done for a given year, a new poverty head count (PO) and poverty gap (P1) can be calculated counting all households in each of the 7 categories of RA, assuming a new projected level (nominal income-based) poverty line for rural and urban areas. By comparing the ex-post and ex-ante P0 and P1, an analysis of the impact of the shock or the policy change can be conducted. 4.3 Social Indicators and Solving for IDGs PAMS can also feature as additional options new modules that can be constructed aiming to assess how the composition of expenditures (both public and private) affect macroeconomic outcomes. The underlying assumption is that there is a long term effect between skills accumulation and total factor productivity, for example a la endogenous growth. It is possible to model this 'micro" to "macro" linkage through -for example-- the RMSM-X ICORs on the production side of PAMS and an 'implicit" production function for social services. Recall that -in one of the possible specifications-- the technology of production for rural output, depends on the (public) investment in rural infrastructure. Similarly, it can also be assumed that the skill composition in the economy could affect the urban ICOR CaURB. The effect of skills is to increase productivity, i.e. to reduce the need for a higher share of fixed investment. How would PAMS treat the accumulation of skills? First, there are public (exogenous) policies that favor skills upgrading. Spending in Education for example. But the "behavior" from unskilled workers can be also taken into account (also "exogenously" but with a rationale): they would spend a greater fraction of their disposable income into skills acquisition because of the income (wage) returns to education. There 12 On additional feature: in order to acount for structural changes in the economy, coming from changes in the composition of the labor force (skills), its allocation across sectors (sectoral labor demand) and the relative shifts in the structure of production, PAMS re-weights the number of households belonging to each RH from the original initial sample to reflect the sectoral structure of production and employment in the simulated scenario. Page 27 A Poverty Analysis Macroeconomic Simulator - PAMS DRAFT is finally an assessment of the efficiency of social expenditures on social items. Let us take the example of Education (Primary) enrollment below. Education: popi ==" EDU.pop EU IC EDuG KEDUDic EU -- __G _ EDU *POP XG - ,PRV i - G,i POP PRI, POP; I EDU = x C W +u L. (exp (O~PRJViXRV,J Y'G,i JG (Df i .IG,i ~ XRIVj DLN,~ + (OEDU ED + bEDU. EDU' D EDU = DEDU 2EDU] IDG ~g[E Ag i GAPEDU = IDGEDU = J-I (iDU XEDU It is possible to define the incidence of (private and public) expenditures on say, an education goal such as 'enrollment ratio in primary schools", for each of our RH groups. One has to assume the proportion of children that falls into the primary school category for each of our (i) groups (or in the current version, as an aggregate). Then per capita expenditures are calculated. The "enrollment" ratio is defined as a function of two arguments: Income per capita level of the labor category, and A logit function, normalized to yield results comprised between 0 and 1, is defined to determine the "enrollment" ratio, for each labor category. The logit function posits that the 'enrollment' ratio is the joint product of complementary public and private expenditures on education. The estimation of the parameters qPE'gJ and OGiu for each group will tell the degree of complementarity between the public and the private expenditures. The model can be solved in two modes: * Positive mode: PAMS computes the enrollment rates resulting from the levels of (private and public) per capita expenditures on a specific social item. * Normative mode: PAMS computes the levels of public per capita expenditures on a specific social itern that is needed to achieve the desired level of an IDG, given an assumed level of private per capita expenditures. Page 28 A Poverty Analysis Macroeconomic Simulator - PAMS DRAFT 5. Conclusions This paper provides a general simple procedure for linking simple macro models (and particularly the macro-consistency frameworks such as the RMSM-X or the Financial Programming frameworks) to Household Surveys. The method used in the PAMS sees growth as distributional dynamic process across several "typical" socio-economic groups or RHs. The framework determines the income of each group using a Representative Household (RH) hypothesis. It disaggregates production and "link" sectors with an explicit labor market broken down by labor categories. It simulates the top-bottom effects of macro economic policies and shocks on poverty and distribution using country-specific household survey (HHS). But PAMS rests on a set of assumptions (and limitations) that have to be kept in mind by the user. The framework breaks down production from a pre-existing macro model. If -as it is the case with the RMSM- X-production comes from a fixed coefficient production function, it will remain so in the PAMS. Similarly, there is only partial (rural/urban) endogeneity of the changes in price indices. The framework -in general-continues to assume that there are limited relative price effects (except for the effect of the real exchange rate on exports and imports) affecting resource allocation in the economy. The framework also assumes limited substitution between categories of labor, except for possible migration between rural and urban areas for unskilled workers. The projected levels of poverty and inequality derive from the assumption that there are no changes in the distribution of income within each representative group, once the mean income of the economy is projected. Finally, to see in practice how PAMS performs, we refer the reader to Annex 6 where we simulate a baseline macro scenario for Burkina Faso corresponding to existing poverty-reduction macro programs (a PRGF and PRSC) with the IMF and the World Bank. We introduce -within the assumptions of these existing programs-marginal changes of tax, fiscal and sectoral growth policies to reduce further poverty and the level of inequality vis-a-vis the base case. Hence, we make the case for the existence of several possible 'equilibria" in terms of poverty and inequality within the same macro-stabilization framework. Page 29 A Poverty Analysis Macroeconomic Simulator - PAMS DRAFT 6. References Agenor FP.R., A. Izquierdo and H. Fofack [2001], IMMPA: A Quantitative Macroeconomic Framework for the Analysis of Poverty Reduction Strategies, The World Bank, Mimeo, March 2001. Blanchflower D. and Oswald A. [1994], The Wage Curve, MIT Press Bourguignon, F. [2002], The growth elasticity of poverty reduction: explaining heterogeneity across countries and time periods, in T. Eicher and S. Tumovsky (eds), Growth and Inequality, MIT Press (forthcoming) Bourguignon, F., de Melo, J., Morrisson C., (1991], Poverty and Income Distribution during Adjustment: Issues and Evidence from the OECD Project, World Development 19(1), 1485-1508. Bourguignon, Pereira da Silva and Stem [2002], Evaluating the Poverty Impact of Economic Policies: Some Analytical Challenges, The World Bank (paper presented at the March 14-15 IMF Conference on Macroeconomic Policies for Poverty Reduction) D. Chen and A. Storozhuk [2002]. RMSM-X+P A Minimal Poverty Module for RMSM-X, The World Bank Decaluwe B., Patry A., Savard L., Thorbecke E., [1999], Poverty Analysis within a General Equilibrium Framework, CREFA Working Paper 9909, University of Laval. Devarajan S., Easterly W., Go D., Petersen C., Pizzati L., Scott C., Serven L. [2000], A Macroeconomic Framework for Poverty Reduction Strategy Papers, The World Bank, Mimeo. Easterly W. [1997], The Ghost of Financing Gap, Policy Research paper No. 1807, The World Bank, 1997. Fallon P. and Pereira da Silva L. [1996], South Africa: Economic Performance and Policies, Southem Africa Department Discussion paper No. 7, 1996. Hamerrnesh, Daniel S. [1993] Labor Demand, Princeton, N.J. : Princeton University Press. Ravallion M., [2001], Growth Inequality and Poverty: Looking beyond Averages, The World Bank, 2001 Robilliard A.S., Bourguignon F., and Robinson S. [2001], Crisis and Income Distribution, A Micro-Macro Model for Indonesia, The World Bank, mimeo. RMSM-X [1994], User's Guide, The World Bank Yitzhaki Shlomo, and Lerman Robert I. (1991] Income Stratification and Income Inequality. Review of Income and Wealth, Series 37, No. 3 (September):313-329. Page 30 PAMS - Annex I Annex 113: Policy Questions Areas of Policy Change Most Commonly Examined in a Sample of PRSPs and l-PRSPs A survey was undertaken to assess the policy content of a sample of PRSPs and l-PRSPs. The objective of the exercise was to test in the sample what were the most common policies and instruments used for poverty reduction. The sample consisted of 4 full PRSPs (100% of actual)'4 and 13 I-PRSPs (40% of actual)15. Countries from MENA (1 I-PRSPs), AFR (4 PRSPs, 6 I-PRSPs), EAP (2 I-PRSPs), LAC (2 I- PRSPs), and ECA (2 I-PRSPs) were included. To tabulate the policy content of existing PRSPs/l-PRSPs, the following methodology was used. Policy measures in each sample PRSP/I-PRSP were catalogued based on a list of specific macroeconomic and structural reform measures. The macroeconomic policy measures included monetary, fiscal, and exchange rate policies. The structural reform measures encompassed institutional changes (including anti-corruption, decentralization, tax administration, and budgetary reform), sectoral reform policies such as privatization, changes in tax rates, and expenditure increases/decreases in specific sectors. When a sample strategy advocated a specified policy change, a "hit" was generated in the table. The number of "hits" per policy was added up, and the percentage of l-PRSPs and PRSPs which cited that particular policy was calculated. Some trends are apparent in the aggregation of the data. The most commonly advocated poverty reduction policies are expenditure increases in social sector spending, including primary health, education, and water & sanitation. Almost all of the sample strategies mention anti-corruption measures as well. Other institutional reform measures such as decentralization, civil service reform, and budgetary reform are also commonly cited. Although most policy measures advocate for some type of increase in expenditure, there is little mention of changes in macroeconomic policy targets to fund this increase. Instead, according to many of the strategies, govemments will fund the increase in expenditure through improvements in tax administration and changes in the tax rates. Every strategy in the survey promotes some increase in social sector spending.'6 Areas of focus are primary health care, primary education, other education activities (adult training is frequently mentioned), and water and sanitation. All PRSPs and l-PRSPs mention increasing primary health and education expenditures. Most strategies also call for more investment in water and sanitation facilities and non-primary health and education Govemments also advocate for a reexamination of cost-recovery in social sectors in a number of strategies (38% of I-PRSPs, 50% of PRSPs). Institutional reform measures to improve government efficiency and transparency are stressed in virtually all of the strategies. All PRSPs and almost all l-PRSPs (92%) promote some sort of anti-corruption policy. Decentralization of govemment activities is cited in virtually all the sample strategies (92% of l-PRSPs and 100% of PRSPs). Many strategies also mention civil service and budgetary reform. 13 This Annex was written by AIya Husain (PRMPR) 14 Uganda, Burkina Faso, Tanzania, and Mauritania '5 Yemen, Chad, Ghana, Cameroon, Kenya, Zambia, Rwanda, Cambodia, Vietnam, Bolivia, Honduras, Albania, and Georgia 16 In a few l-PRSP documents, this was implied rather than explicitly stated. Page -31 PAMS - Annex 1 As noted earlier, very few strategies advocate macroeconomic policy changes (from previous IMF programs). The most commonly cited macroeconomic measure is a change in the fiscal target (23% of I-PRSPs, 50% of PRSPs). Although we have a very small sample, it may be of some significance that half of the PRSPs mention relaxation of the fiscal targets. As we move into the stage where more countries are preparing PRSPs with fully articulated public expenditure programs, the macroeconomic targets may be revisited. Structural reform measures including trade reform, privatization, financial sector reform, and agricultural sector reform are frequently cited in the sample strategies. Policies advocated include customs reform, lines of credit for small and medium enterprises, privatization of utilities, and reform of the regulatory system. Among the most commonly cited agricultural policies are land reform and investments in rural infrastructure. Improvements in tax administration as well as changes in tax rates and tax composition are commonly mentioned as a vehicle for increasing state revenues, especially in l-PRSPs. The resulting revenues are seen as crucial for financing the public expenditure program. Very few strategies promote any decreases in expenditures. The Uganda PRSP is the only document which mentions that the govemment will decrease military expenditures, and few strategies also cite a decrease in specific subsidies. Most common among expenditure reductions are cuts in civil service employment taken in the context of a civil service reform. Because of the absence of a significant sample of full PRSPs, the findings should be seen only as indicative. In most l-PRSPs, it was ongoing policies as opposed to fully articulated strategies which were assessed. As more PRSPs are developed, some of these conclusions, particularly those pertaining to the macroeconomic and fiscal areas could change. Page -32 PAMS - Annex 2 Annex 2: Extracting Relevant Information about Socioeconomic Groups from a Household Survey (HHS) The analysis of the distributional impact of shocks and policies requires that information be organized according to the relevant socioeconomic groups that make up the society under consideration. The purpose of this annex is to explain in details the type of data required and how they could be extracted form a household survey. Some of the procedures involved will be illustrated in the context of the Burkina Faso 1998 Priority Survey. I. Data Requirements Fundamentally, the impact of shocks and policies on the living standard of individuals or household depends on their participation in the socioeconomic activity. The reward they get from this participation depends in turn on the source of their livelihood and how they allocate their resources. Thus employment status, and the level of earnings and expenditures are the key variables that must be combined with demographical information to organize the data in the desired structure. Where to find this information in a household survey depends on the structure of the questionnaire. The 1998 Burkina Faso Priority Household Survey is Organized as follows: o Section 0 provides information on the head of household such as her/his ethnic group or nationality, language, or religion. o Section 1 contains information on the demographics of each household. For each household member, this section collects information on (1) the relation with the head of household, (2) age, (3) marital status, and (4) gender. o Section 2 collects health information a Section 3 contains information on education for members who are at least six years old. o Section 4 collect information about literacy and migration for all members of the household who are at least 10 years old. o Section 5 concems the employment of all members of the household who are at least 10 years old. It gathers information on the principal occupation, the secondary occupation and previous employment. This is where one finds information on whether or not the household member is currently (i.e. at the time of the survey) employed, the type of employment (permanent or temporary position), the sector of employment, the skill profile, earnings and other compensations. o Section 6 is on housing a Section 7 is on agriculture and livestock o Section 8 is on non-agricultural activities o Section 9 collects information on access to basic social services. o Section 10 provides information on household expenditure. o Section 11 contains information revenues classified by source: Salaries, agricultural income, non-agricultural income and transfers. a Section 12 has information on household assets. o Section 13 provides anthropometrical data for children between 6 and 59 months of age. The pilot version of PAMS is based essentially on information from Sections 0,1,5 and 10. Information on the revenue Section was deemed unreliable. We now move to explanation of how the data was processed to conform to the needs of PAMS. Page -33 PAMS - Annex 2 II. Data Processing Data processing is conducted into two basic steps. The first one is to define the following basic categorical variables. . Working = 1 if the individual in question worked during the year of the survey, 0 otherwise. * Rural=l if individual resides in rural area, 0 otherwise. * Public=1 if individual works in the public sector, 0 otherwise. * Private=1 if individual works in the private formal sector, 0 otherwise. * Tradable=1 if individual works in a sector producing tradable goods, 0 otherwise. * Skilled=1 if individual is considered skilled, 0 otherwise. The second step builds a global categorical variable by aggregating the basic variables defined above. In fact each final socioeconomic group is a subset of the cross-product of the elemental categorical variables. Example the group called Rural Nontradable Unskilled is defined as follows: Rural =1 and Tradable=0 and Skilled=0. The following is an EViews Subroutine that defines the basic variables and does the aggregation: 'BFALPCLASS.PRG: A global subroutine called by BFALPGROUPS to restructure the 1998 Household Priority Survey for Burkina Faso according to the categories of the LP Module of RMSM-X-LP. 'B.Essama-Nssah, PRMPR (World Bank) November 17, 2001 SUBROUTINE BFALPCLASS 'Make sure to start with a full sample Smpl Wall Series LPSEG 'Socio-Economic Groups of LP Series RURAL' =1 if G4=2 and 0 otherwise Series WRKNG 'Dummy indicator of Economic Activity based on GSE <>7 and 9 Series PUBLIC 'Dummy indicator of Public Sector based on GSE=1 Series PRIVATEF' Dummy indicator of Formal Private Sector based on PUBLIC=0 and GSE=2 Series TRDBL 'Dummy indicator of whether individual works in a tradable sector based on GSE=5 or S51Q9>=10 and <=14 Series SKLD 'Dummy indicator of skill level mainly for the private formal sector based on S51Q10=1-5; 6-8;9-10 'Recode Rural Sector Smpl if g4=2 rural=l Smpl if rural<>1 rural=0 'Assign Economic Status (working or not) Smpl if S51Q2=1 wrkng=l Smpl if wrkng<>1 wrkng=0 'Create Public Sector Smpl if gse=l Public=1 Smpl if Public<>1 Public=O 'Create Formal Private Sector based on gse Smpl if gse=2 Privatef=1 Smpl if Privatef<>1 Page -34 PAMS - Annex 2 Privatef=O 'Define Sectors Producing Tradable Goods based on GSE and S51Q9: cotton and Mining Smpl if (gse=5 or (s51q9>=10 and s51q9<=14)) Trdbl =1 Smpl @all if trdbl<>1 Trdbl=O 'Assign Skill Levels Smpl if s5lqlO<=5 skld=1 Smpl if (s51q10=6 or s5lqlO=7 or s51q10=8) skld=O Smpl if s5lqlO>=9 skld=2 'Build LPSEG 'Rural Nontradable (Assumed Unskilled) Smpl if ((rural=1 and wrkng=1 and gse>=2) and (trdbl=O and skId<=2)) Lpseg=1 'Rural Tradable (Also Asummed Mostly Unskilled) Smpl if ( (Ipseg=na and rural=1 and wrkng=1) and (trdbl=1 and skid<=2)) Lpseg=2 'Public Sector Smpl if (Ipseg=na and Public=1) Lpseg=3 'Urban Tradable Skilled (formal Private Sector) Smpl if ( (Ipseg=na and wrkng=1 and rural=O and Public=O) and (privatef=1 and trdbl=1 and skld=1)) Ipseg=4 'Urban Nontradable Skilled (Formal Private sector) Smpl if (( Ipseg=na and wrkng=1 and rural=O and Public=0 ) and (privatef=1 and trdbl=O and skld=l)) Lpseg=5 'Urban Nontradable Unskilled Smpl if (( Ipseg=na and wrkng=1 and rural=O and public=O) and (privatef=1 and trdbl=O) and (skld=O or skld=2)) Lpseg=6 'Urban Informal Economy (all assumed unskilled) Smpl if ((Ipseg=na and wrkng=1 and rural=O and public=O) and (gse=4 or gse=5 or gse=6 or gse=8)) Lpseg=7 'Self-Employed Smpl @all if ((Ipseg=na and wrkng=1 and rural=O and public=O ) and (gse=3 and skid<=2)) Lpseg=8 'Rentiers Smpl if ( Ipseg=na and S51 Q2=6) Ipseg=9 'Not working 'smpl if (Ipseg=na and wrkng=O and s51q2<>6) smpl if (Ipseg=na and (wrkng=O or gse=7 or gse=9)) Ipseg=1 0 'Back to full sample Page -35 PAMS - Annex 2 Smpl @all uno.statby(sum,nomean,nostd) Ipseg 'To test for empty groups ENDSUB The above subroutine is embedded in the following program that does two things. It recodes the global categorical variable if there are some empty groups. After recoding, it processes a group of variables e.g. household size, household weight, per capita expenditure, total household expenditure, or total income. This process constructs distributions of these variables by socioeconomic groups. The final result is then exported to Excel creating a separate file for each socioeconomic group. 'BFALPGROUPS.PRG does two things: (1) It creates socioeconomic groups according to LP classification (this is done by calling the subroutine BFALPCLASS);(2) it processes a group of variables according to that classification and exports the results into Excel files. The program requires two arguments: the workfile and the categorical variable used in the creation of subsamples. These arguments must be specified within the call as in the following examples: 'Run(v) bfalpgroups bfa981p Ipseg or run(q) bfalpgroups bfa98lp Ipseg 'B. Essama-Nssah, PRMPR (World Bank) November 17, 2001 'Revised March 7, 2002 Include BFALPCLASS Load %0 Call BFALPCLASS 'Recode categories since preliminary tests revealed that LPSEG=4 yields an empty set; If not just set lp=lpseg Series LP 'To use in the call Smpl if lpseg<=3 lp=lpseg Smpl if lpseg>4 lp=lpseg-1 %v="lp" smpl Wall Freeze(unotab) Uno.statby %1 !cat=@max({%v}) Total number of sub-samples 'Create group of variables to be processed Group grp rural pdum hhsize hhwgt pcx tothhx totinc !gsz=grp.@count 'Initialize first column to receive results !t=1 For !j=1 to !gsz %st= grp.@seriesname(!j) for !i=1 to !cat smpl if %v=!i !k=@obs({%v}) if !k then 'To protect against vectors of null dimension vector(!k) v{%stH!i}={%st} Page -36 PAMS - Annex 2 matrix(!k, !gsz) mat{!i} colplace(mat{!i}, v{%stX!i), !t) endif next !t=!t+1 Next SmpI @all For 1i=1 to !cat mat{!i}.write(t=xls, a2) bfa{%v}mat{!i} Next 'Note: if any of the above matrices has more than 8,193 rows EViews will issue an error message stating that the maximum size for a spreadsheet has been exceeded. When this happens, one may resort to the copy/paste facility or to an altemative plafform such as SPSS. If any subgroup has more than 65,536 observations then there is no way to export the results to the current version of Excel (Excel 2000). 'Tum matrices into tables For !i=1 to !cat freeze(bfa{%v}tab{!i}) mat{! i} setline(bfa{%v)tab{!i}, 3) Next 'Label table columns For !col=1 to !gsz %st= grp.@seriesname(!col) for !i=1 to !cat bfa{%v}tab{!i}(1, !col+1)=%st next Next 'End of Program One needs to keep the above limitations in mind. EViews will issue an error message if any of the socioeconomic group has more than 8,193 observations. The error will say that the maximum size for a spreadsheet has been exceeded. We are told by the makers of EViews that this is to maintain backward compatibility with an older version of Excel! However, one can go around this constraint by resorting to the copy/paste facility, provided that the number of observations to be exported is les than or equal to 65,536 (Excel 2000). Otherwise, one should think about moving to another plafform. Ill. Data Availability Information on available household data sets may be found on the following Websites The Poverty Monitoring Database HTTP://WWW.WORLDBANK.ORG/POVERTY/DATAIPOVMON .HTM This site has six main components: o Household Survey for 124 countries (currently) classified by country, year or region. o News on upcoming surveys, studies and poverty assessments. o Social Indicators. o Basic information on participatory poverty assessment by the World Bank Group. o Links to other Relevant sites. Africa Household Survey Databank Page -37 PAMS - Annex 2 THTTP:/NVWW4.WORLDBANK.ORG/AFR/POVERTY * The site provides information on access policy for both Bank and non-Bank users. * As of August 1, 2001 the status of the Databank showed the following: o 110 surveys covering 37 countries o 28 Priority Surveys o 15 Household Budget Survey/income and Expenditure Surveys o 21 Integrated Surveys o 1 Core Welfare Indicators Questionnaire (CWIQ) Survey o 40 Demographic and Health Surveys (DHS) o 1 Demographic and Health survey (non-DHS) o 4 Others Page -38 PAMS - Annex 3 Annex 3'7: Labor Demand Elasticities for the PAMS I. Objective The Labor and Poverty module features a simple segmented labor market. The user of the PAMS needs to estimate the coefficients (elasticities) of the labor demand equations that are described in the PAMS main document for the specific country he/she is concemed about. In the absence of adequate time-series, this requirement could be a problem. This Annex aims at providing the user with a review of the literature on labor demand elasticities of wage and output for two heterogeneous labor groups (skilled and unskilled). The present literature survey is not exhaustive. It aims simply to (1) provide an illustration of acceptable ranges of wage and output elasticities for labor demand; (2) identify labor databases that are accessible for economists; and (3) provide actual estimates of wage and output elasticities of labor demand using aggregated data of manufacturing and agricultural sector for two regions (Africa and Latin America). II. Data Sources on Employment, Wage Rates and Output 11.1. Sources Useful labor data can be found in the following sources. (1). Martin Rama and Raquel Artecona [1999] n A Database of Labor Market Indicators across Countries". This database covers most of the countries and report five years average figures to increase the reliability of the data from 1945-49. While this procedure tends to strengthen the quality of the data, only a limited number of observation is available. (2). World Bank Database: SIMA. The World Bank database covers most of the developing countries for aggregate labor data (although there is some disaggregated data). However, this database comprises missing observations and is difficult to use for country specific research. (3). IMF's IFS. This database is helpful for country specific analysis since it provides time series annual data for employment, output and aggregate average wage rate (disaggregated data is not available). 11.2. Extraction of Data (2). The World Bank SIMA The data can be easily downloaded from Bank Intranet in Excel format. Here are some tips for the first users. First: go to Bank Intranet and select Data shortcuts from full-down menu Second: select SIMA query from the menu Third: Double click QUERY in the box to activate the extraction process Fourth: Select database (e.g., Regional Africa or others depending on your target) Fifth: Select target countries Six: Click series and select series (e.g. employment of agriculture as a % of total employment, Monthly wages of agricultural sector, value added as a % of total GDP) This Annex was written by Hong-Ghi Min (PRMPR) Page -39 PAMS - Annex 3 Seven: Click periods and select periods (e.g., 1980-1995) Eight: Click show data, this will show you requested data in Excel format Finally, Click File and double click export data from full-down menu Ten, Select Excel and assign a file name for data file you extracted. Ill. Literature Review In this section we report the results of a rapid survey of the literature on elasticities of labor demand grouped by industry, educational level, occupation, region, and countries. 111.1. Wage Elasticity of Labor Demand Table A3-1 shows the results for some published studies on wage elasticity of labor demand. The 'results are that, first, estimates varies from -0.12 (Nadiri, 1968) to -1.54 ( Symon and Layard, 1984) and most of elasticities are usually lower than 1 (in absolute value). Second, Heckman and Sedlacek (1985) report that in most cases, the manufacturing sector's age elasticity is smaller than that of non-manufacturing sectors. 111.2. Output Elasticities of Labor Demand Table A3-2 provides the output elasticities of labor demand for different countries. First, the minimum value is 0.2 for the United States (Estevao, 1996) and the maximum value is 1.0 for the United States (Shapiro, 1986). In particular, Hamermersh (1993) finds that output elasticity is relatively insensitive to the functional form of the production function and that most estimates vary between 0.15 to 0.75. IV. Data Collection and Estimation We also collected data and conducted our own estimates for labor demand elasticities. Our findings are below. IV.1. Data Collection We used "A Database of Labor Market Indicators across Countries" by Martin Rama and Raquel Artecona., 1999 as well as (2) World Bank Database: SIMA. We have a set including Latin American and African countries. There is a relatively small number of observations available. IV.2. Panel Estimatlon The tested functional form is as follows: Log(Labor demand) = aO + al*Log( Wage) + a2*Log (Output), where, a, is wage elasticity and a2 is output elasticity of labor demand. (1). Labor Demand Elasticity In Latin America Fixed effect estimation, random effect estimation, and White (1980)'s heteroscedasticity consistent estimation method were employed and labor demand elasticities for Latin America as a whole is reported in Table A3-3. The estimation shows that the wage elasticity lies between - 0.20 and -0.78. Output elasticities lie between 0.3 to 0.87. All estimates are significant as is the case with wage elasticity. Actual estimation results are consistent with the literature surveyed in the previous section. Page -40 PAMS - Annex 3 (2). Labor Demand Elasticity in Africa First, we estimated the labor demand elasticities of manufacturing sector and results are reported in Table A3-4. For the manufacturing sector, wage elasticities hover from -0.2 to -0.71 and that of output hover between 0.32 to 0.92. All estimates are significant except the wage elasticity of fixed effect model. Next, we estimated the elasticities of labor demand for the agricultural sector and results are reported in Table A3-5. For the agricultural sector, wage elasticities hover from -0.44 to -0.88 and that of output hover between 0.64 to 0.72. All estimates are significant at 1 percent critical level. When we compare the absolute value of agricultural sector with that of manufacturing sector, estimates of elasticities for the manufacturing sector are higher than that of the agricultural sector. IV. Conclusion In the absence of country specific time-series, the recommended range of labor demand elasticities for the calibration of Labor and Poverty module, based on literature survey and actual estimation, can be summarized as follows. (1). Estimates of manufacturing sector (proxy for the demand for urban labor) We would recommend values between -0.20 to -0.9 for wage elasticity and 0.3 to 0.9 for the output elasticities. (2). Estimates of agricultural sector (proxy for the demand for rural labor) We would recommend value between -0.10 to -0.9 for wage elasticity and 0.2 to 0.9 for the output elasticities. Page -41 PAMS - Annex 3 Table A3 - 1. Wage Elasticities of Labor Demand: Literature Authors Industry, Period, Country Kollreuter (1980) Manufacturing, 71-77, West Germany -0.20 Hsing (1989) Manufacturing, 53-78, Unite States -0.70 Heckman and Sedlacek (1985) Manufacturing, 68-81, United States -0.49 Non-manufacturing -0.93 Franz and Konig (1986) Manufacturing, Translog, 64-83, U.S. -0.96 Pencave and Holmlund (1988) Manufacturing, 50-83, Sweden -0.75 Harris (1985) Manufacturing, 68-81,U.K. -0.21 Nadiri.(1968) Manufacturing, Translog, 68-81, U.S. -0.12 Layard and Nickell (1986) Aggregate, 54-83, U.K. -0.93 Andrews (1987) Aggregate, 50-79, U.K. -0.51 Hairris (1990) Aggregate, 65-87, New Zealand -0.24 Symons and Layard (1984) Manufacturing, 56-80, 5 OECD countries -1.54 Wadhwani (1987) Manufacturing, 62-81, U.K. -0.38 Begg et al. (1989) Aggregate, U.K., 53-85 -0.40 _________________________ Japan, 53-86 -0.45 Maximun - 1.54 Minimum -0.12 Table A3 - 2. Output Elasticity of Labor Demand: Literature Study Country OEL* Nadiri & Mamuneas(1996) US 0.36-0.37 Munnell (1990) US 0.59 Estevao (1996) US 0.2-0.22 Shapiro (1986) US 1.00 Feldstein (1967) UK 0.75-0.90 Lesli & White (1980) UK 0.64 Hart & McGregor (1987) Germany 0.31 Roberts & Skoufias (1997) Colombia Skilled 0.733, (Manufacturing) Unskilled 0.661 Pessino (1997) Argentina 0.25 Lim (1976) Malaysia 0.45-0.67 Gujarati (1999) (Manufacturing) 0.67 '*'he World Bank (2000) Taiwan(Agricult.) 0.38 (=share of labor) Maximum 1.00 Minimum 0.20 Page -42 PAMS - Annex 3 Table A3-3. Labor Demand Elasticity in Latin America: Manufacturing Sector Fixed Effect Random Effect White's Estimation Constant ___ 4.97 (0.46)*' 4.80 (0.79)** Wage -0.20 (0.08)* -0.45 (0.07)** -0.78 (0.12)** Output 0.30 (0.07)** 0.57 (0.04)** 0.87 (0.04)** Adjusted R-squared 0.45 0.88 0.95 No. of Observations 33 33 33 Note: 1. Panel of 11 countries with 11 observations (missing values) 2. Double asterisks denote that estimates are significant at 1 percent critical level And single asterisk at 5 percent critical level. Table A3-4. Labor Demand Elasticity in Africa: Manufacturing Sector Fixed Effect Random Effect White Estimation Constant ----- 1.54 (0.78) -1.76 (0.48)** Wage -0.20 (0.21) -0.63 (0.15)** -0.72 (0.09)** Output 0.32 (0.07) * 0.62 (0.05)** 0.92 (0.04)* Adjusted R-squared 0.25 0.58 0.95 No. of Observations 96 96 33 Note: 1. Panel of 12 countries with each of 16 observations are used (many missing values). 2. Double asterisks denote that estimates are significant at 1 percent critical level And single asterisk at 5 percent critical level. Table A3-5. Labor Demand Elasticity in Africa: Agricultural Sector | Fixed Effect Random Effect White Estimation Constant l-_-_-| 4.97 (0.46)** 1.06 (0.77) Wage -0.45 (0.21)* -0.45 (0.07)** -0.87 (0.07)** Output 0.64 (0.08)** 0.57 (0.04)-- 0.72 (0.04)* Adjusted R-squared 0.45 0.88 0.89 No. of Observations 76 33 33 Note: 1. Panel of 11 countries with 16 observations are used (many missing values). 2. Double asterisks denote that estimates are significant at 1 percent critical level And single asterisk at 5 percent critical level. Page -43 PAMS - Annex 3 V. References Carlos Pessino, 1997, "Argentina: The labor market during the economic transition" in Labor Markets in Latin America: Combining Social Protection with Market Flexibilitv Craine, 1973, "On the service flow from labor", Review of Economic Studies, vol. 40, 39-46. Estevao, M.,1996 ,"Measurement error and time aggregation: a closer look at estimates of output- labor elasticities", Board of Govemors of the Federal Reserve System, Washington DC. Feldstein, 1967, " Specifications of the Labor Input in the Aggregate Production Function", Review of Economics and Statistics, vol. 34, 375-86 Gujairati, D., 1999, Essentials of Econometrics, 2nd edition, McGrw-Hill. Hamermesh 1993, Labor Demand, Princeton University Press, 1993 Hart and McGregor, 1987, "The retums to labor services in west German Manufacturing industry, European Economic Review, vol. 32, 4, 947-64. Lawrence J. Lau, 2000, The Macroeconomics of Development, unpublished manuscript Lesli and White, 1980, "The productivity of hours in UK manufacturing and production industries, Economic Joumal, vol. 90, 74-84. Lim,1976,"On estimating the employment-output elasticity for Malaysian manufacturing", The Journal of Developing Areas, vol. 10, April, 305-316. Munnell, 1990, "How Does Public Infrastructure Affect Regional Economic Performance" New Englend Economic Review, Sep/Oct. Nadiri and Mamuneas, 1996, Contribution of Highway Caoital to Industrv and National Productivity Growth, unpublished manuscript Oldrich and L. Kyn,1976, 'Macroeconomic Production Functions for Eastern Europe" in On the Measurement of Factor vroductivities, ed. By Altman, Kyn, and Wagner, Vandenhoeck & Ruprecht, Goettingen. Roberts, M. and E. Skouflas, 1997,"The Long-run Demand for Skilled and Unskilled Labor in Colombian Manufacturing Plants", Review of Economics and Statistics, vol. 79, 2, 330-34 Shapiro,1986, "The dynamic demand for capital and labor," Quarterly Journal of Economics, vol. 101, 5'13-42. The World Bank, 2000, India: Policies to reduce Dovertv and accelerate sustainable development.. Washington DC. The World Bank, Economic Trends in MENA Region, Washington DC. 2000. Page -44 PAMS - Annex 4 Annex 4: Linking PAMS with macro-consistency frameworks Procedure for the RMSM-X18 We use the built-in linkages between cells in Excel worksheets to link the Labor and Wage- Income module and the RMSM-X. The procedure to link cells in Excel is straightforward. You select the cell that you want to link by clicking on it. You type a equal sign (=) to prompt the cell for receiving an input. You then move your cursor to the RMSM-X sheet that you need, by clicking on the bottom bar of names for the relevant sheet. Then you go to the relevant cell and click on it. Type retum and the link will be automatically selected. Usually, the mode under which the link is established in absolute. It means that the cell will register the link with the name of the worksheet, the name of the sheet inside the worksheet and the absolute references of the cell using dollar signs for the cell row and column (e.g., ='[RMSM-X_BRZOO.xIs]PIT'!$E$45). When you need to copy the same reference to the right (in order to project the line) you will need to remove the first $ sign (the column sign) to avoid copying the cell for this particular year only. 1. Loading files, databases and re-saving them Load the Light Data Base (LDB) and any other Excel database that is necessary for your work (e.g., IFS, UNIDO, ILO, WBDI, WBGDF, etc.).and that might be available in the World Bank SIMA system of information. Load the RMSM-X and the shell of the Labor and Wage-income module, both in Excel. Save under the shell of the Labor and Wage-Income module under the same directory as your RMSM-X, with a different name, related to the country you are working with. Re-link the shell with the RMSM-X worksheet you are working with. Go to Edit in the bar menu of Excel, go down to Links... and select it. When the Links window opens, click once on the RX file to highlight it. Then chose the Change Source button and click on it. For each database that is needed, use the same procedure. Changing the data sources will produce some 'errors", #NA' in the Labor and Wage-income module, because Excel select exactly the cell referenced in the shell, and the relevant data might not in the exact same cell in another database. Updating databases will therefore imply now a thorough examination of each spreadsheet in the Labor and Wage-income module. 18 In this Annex, we depart from the rule explained in the PAMS of a fixed number of 6 groups of RH plus a 7th group of 'Rentiers". We add from the Burkina Faso case, used as an example, 'Self-Employed', 'Unemployed" and a small category of skilled workers in the Urban Non-Tradable Urban" sector. Page -45 PAMS - Annex 4 II. Updating the Original Data Go to the Original Data spreadsheet (ORIGINAL DATA) in the Labor and Wage-income module. The top of the spreadsheet should look like this. You need to update all the PrRoDIUT'IoN ' ' information in the blue 'Gross Dornestic Product at 'm column, including changing IRRAL the base year for the iGIDP nip Agriculture (Primnry Sector) - . projections (which should be jGDP rnp Agriculture, Tradable (E-rts)- I consistent with the base year GDPnpTAgriculture, Non-Tradable1 o the v ~~~~~~111. .- ---1 -. 1--- - - I-.- ---- ----1 I Of the RMSMlv-X). InvestO (PbE ector in Agric. Infrastruct.) j Some of the links with the GDP np Urban LDB data base and/or the CEnp Urban, Fonmal Tradable (HtrrS)I RMSM-X that you are using GDP rnp Urban, FornTL Non-Tradabie Q_ might have worked well, some CUDP ip, Urban, Infornml not because of discrepancies between the exact location of the cell referenced by the new module. You will have to check each cell in detail and make sure that it is the adequate information that is uased by the Labor and Wage-income module. Most of the data will come from the macroeconomic information available in the LDB data set. However, in other instances LABOR FORCE TOTAL (unit: Nillions) (particularly regarding the .APLOYTh , ABO, ,,O,,,,,,,,,,,, T,T, ,,,,,,, , t 7810 composition and breakdown UMrA-LOO IABOR l (i Iors ! of the labor force and its 1 URBAN IABOR,_ Silled. associated remuneration) the !Unkilled information most likely will . . . j ! come from the local Public Sector household survey. T-- -Ood- ---- For example, you have to fill Skill_d i ~ all the relevant cells that will Non-Tradable [iJ inform the model about the Skilled composition of the labor force Unsidlled i jInformtl(unskilled) Page -46 PAMS - Annex 4 111. Linking/Updating the Production Data Go to the Production spreadsheet (PRODUCTION) in the Labor and Wage-income module. PRODLUIlON - __,, -7 2011 2M - Check how the automatic N B pinknumbers =newduction ICOR change of Links has worked 'N-.H ronumbrsfmSMXNAh out. Whether the projected PROU,1XION C9NSTANT PRICES N.B. bue nunbers =onRMSM.X TRApD production data seems -__s Don-nstic Produc at nip F i 1 108,75, 1,1 7.1 Rr!AL consistent. For example, G?Prip Agriculture 1,004,097 1,029,199 there should not be any ?5PnpAgricutureTradabIe(Er4 r1s). jj1j 37,436 46,948 sudden jump in time series (G?P "?mpgr.culture, Non4niclable m.v4- W'66,i6 9V,251 liked the one you can see in Invest (Public sector: Agric. Infrastnct.) . ' 1j,712 150,614 the GDP for Agriculture. URBAN i r?P?npU?b? iS4fil,, 24778 88,17 In the first line, GDP at market ' 2Pni VPUrba ,Foennl,T,dablT ',rt) .- 72?i61 7 prices (mp), the first red cell GD?P nip Urban, Fo?nrl, Non-Tradable O . I 674,228 676,348 (h (iGS?P nip, Urban, 1n nu1j., J . 611)2l, (65,sis34i (the first projected year) . ...--..--.. . .~ should be linked to the same cell (for the same year) of GDP in constant prices in the SNA sheet in the RMSM-X. In the second line (and the first red cell) GDP mp in Agriculture, should be linked to the Agricultural GDP in constant prices in the SNA sheet in the RMSM-X. The third line (blue), should be linked to the GDP in constant prices in the trade sheet in the RMSM-X (TRA). The relevant cell there is the one (or the sum of those) accounting for the exports of Agricultural products. Similar procedures should apply for the GDP mp in Urban sector. The pink line represents the only line where the module projects its own view about a production function. It represents the private sector urban formal sector. You can chose your own view, or mimic the RMSM-X fixed coefficient ICOR procedure. IV. Linking/Updating the Prices and Population Sheets Similar procedures are needed for linking the price and population projections in the Labor and Wage-Income module with the relevant projections in the relevant sheets in the RMSM-X. _________________________ Go to the sheet PRICES in the I-bae year .. ..... 2 Labor and Wage-Income E8CIERNAL -N B red anaten bra, RomiSM-X PiTshed . module. The first step is to Eultauge Rate, $Nararat(LCU per 1USD) L1-0I1- ! 6 ensure that the exchange rate E4at A8ncutunnprrar(olto) , (based 1 in the base year for the ;EvportMaroBotunnain Lr I O.70io. 10o12 simulations) is properly set up. PRODL~1oN . . Go to the first line where the Qua. DarroeatnecPedurlatnp | .ISt.j t-044l; !5 nominal exchange rate in Local RURIPAL 1pamesa 1rjes. r-- i 1.0225 19 Currency Units (LCU) per one W?eecA8nn.iare,Tea4ableR5) a), 10196 1.3i2 US dollar is displayed. The (DP epAgneultun eNon- Traabbk , 0250 n0506d value of the line should follow PaipmPU bar np- -b Forral,Tnx r 04-59 1506 the nominal depreciation of the qDimpoTb-,Fonr* Non-!nd Ei:O 1.0500 i lo05 exchange rate. If not, you WPa~Urba.1afam1~j., _ ''~ should link the exchange rate cells with the relevant cells in the RMSMS-X, in the Price and Interest Rates sheet (PIT). A similar procedure has to be performed for the sheet POPULATION. Ensure that the serie of projection is adequately related to the population projections of the RMSM-X. Page -47 PAMS - Annex 4 V. Linking/Updating the Income sheet. Go to the Income sheet (INCOME) of the Labor and Wage-income module. Two lines need to be linked properly and/or updated. The Income sheet ensures that the projected sum of incomes of all agents in the economy adds up to the total disposable nominal income that is calculated in the macro-consistency framework (here the RMSM-X). That is done in the first line of the sheet, where the income of 'Capitalists & Rentiers" is calculated as a residual between the total disposable nominal income that appears in the lPrivate Sector sheet of the RMSM-X (PRS) and sum of incomes of all agents in the economy. The other line that ensures consistency is the total direct tax payments by all agents that need to be consistent with the total income (direct) tax reported in the Govemment sheet in the RMXM-X. This line has to be linked to the GOV spreadsheet in the RMSM-X (direct taxes). VI. Linking/Updating the Budget sheet. Go to the Budget sheet (BUDGET) of , BUDl)GEr [-,'2COOI the Labor and Wage-Income module. -- - - ~ . - ~---j Several items need to be reconciled in BUDXEr(Millio. odt Ll) order to make the budget allocations - ............ ..i in the Labor and Wage-Income l REcNuE(MIIIOn otLCLh) -. . module consistent with the overall TIotal lne nnueilud ng grants fro"Rms.. . 44s072 envelope of public resources in the TRY Revenue 291836 p iO/wIncon!Ta1eS5TctTaum S)gcnerated 5194 RMSM-X. The following lines need to 1.b . (o. be linked to the proper lines in the Percent ofarrilurd CDPllc! (from M (98,866) RMSMS-X. ____f -9.3% EXP N'DiTURE(Mllion oC.s. E All the necessary links concem the GOV sheet in the RMSM-X. A. [PubIic Sector Wages - 0 45,171 N B Wsges f..RM,S. Mbudget 2, Other Non-Wages Curent E,enditures (Residual Iten . 3,469 | Miailry Eapend:ures 73,388 b)Socia Expengitures l10,081 i, l)Education ofw! ich 36,694 Prinary Ed i 11,008 Second. Ed. 3 669| Ten 'yad - 22,016 b2) Health l 36,694 b3) IncgmieTransfers 40,363 -~~~~~~~~e ~ ~ IB. Cajdtal&p8enitures j.. |Inv.stnentElpenditures frornRMSM-X F2!3 lInes t C ItrsPyuns l ,_3 | ]ntercSt Payrnens frgDRMSM-X I 223,136 I C.l IneetPynnsI- Tota!l Ependutv Cnstreined byp RMSM-? C) . 446,732 in Percent of tDP 420%_ Page -48 PAMS - Annex 5 Annex 5 Household Survey Simulator (HHSS) I - Introduction The HHSS is a set of two Microsoft Excel-Based worksheets that stores detailed relevant variables for analyzing and forecasting the impact of policy shocks (e.g. exogenous change in public wages, institutional change in the national minimum wage, change in public transfers, tax system, pension system, etc.) or macroeconomic performance (e.g. Economic growth, Inflation, etc.) on poverty reduction and income distribution. In fact HHSS (i) stores original data from a country 's household survey (ii) simulates poverty incidence and income distribution effect by category and by geographic region over ten (with possible extension to fifteen ) years head from the base year for any given geographic poverty line profile and/or any income profile by category. 11 - HHSS package The HHSS package comprises two Excel worksheets. The original Household Database (called HHbase-country_acronym) and the Simulator module (called Hhproj-country_acronym). Both are described in the present guide that serves as a preliminary reference manual for running and extending the method to other countries. This guide contains explanations of files, worksheets, buttons, simulation rules, and output sheets. III - The Household data (HHbase-country_acronym) The HHbase stores the original data from a given country 's household survey (HHS) into a specific format (the data is retrieved and arranged according to the procedure described in Annex 2). HHbase is the input file for the simulator module. HHbase is just a database file and does not contain any Excel macro. It contains fourteen worksheets. o The first sheet: Content worksheet describes the relevant information contained in the file o From the second to the eleventh sheet HHbase contains data for ten labor/income categories (See below for detailed information about these categories). o The twelfth worksheet contains the weights of each observation, according to the HHS information, by category and by individual observation. o The thirteenth worksheet contains the income by category and by individual observation o And the fourteenth worksheet is the expenditure by category and by individual observation. The last four worksheets constitute input data for the simulator Module (HHProj). Page -49 PAMS - Annex 5 The Categories of the Household data (HHbase)19 PS-1 Public Sector assumed Skilled RSE -2 Self-Employed assumed from Rural Zone RNTU-3 Rural Non-Tradable assumed Unskilled RTU 4 Rural Tradable assumed Unskilled UIU-5 Urban Informal assumed Unskilled UNTU-6 Urban Non-Tradable assumed Unskilled UNTS-7 Urban Non-Tradable assumed skilled UTS-8 Urban Tradable assumed Skilled KAP-9 Capitalist assumed Skilled UNEMP-10 Unemployed and other non active UNWEIGHT Unweighted series or Household weight from each category's sheet Weighted series or Individual weight with respect to the population from WEIGPHT each category's sheet Income Individual or Household Income series from each category's sheet Expend Individual or Household expenditure series from each category's sheet IV - The simulator module (HHProj) The iHHProj is a separate worksheet and simulates poverty incidence and income distribution effect by category and by location (rural/urban) over fifteen years starting from a given base year, for a given poverty line and for income growth assumptions by category. It can operate as a stand-alone simulator as described in the main text of the PAMS. It suffices to input "exogenous' guess-estimates of the growth of nominal income for each of the ten categories of the Hhbase and run the Excel macro to obtain a simulation of the projected poverty incidence and Gini. Naturally, such an exercise will not be 'macro-consistent" because the inputed growth rates would not necessarily be consistent with macroeconomic equilibria. IV-1- Description: HHproj is linked to HHbase by a macro The sheet "McrSht" is the starting worksheet containing the macro instructions to undertake a simulation. The first step is to select a Baseline scenario. The worksheet is updated (and a Baseline is selected) by clicking on the button 'Update the Baseline". This selected the last output of any simulation as a base. The second step is to run a simulation by clicking in one of the four buttons accordina to the type of simulation that the user wants. Four types of simulations are possible, each corresponding to one of the four buttons available. The user can choose to simulate the poverty reduction and income distribution indicators in terms of household "Expenditure" or "Income' and in terms of weighted (W) representative household or un-weighted representative households (UW). Weights correspond to the number of people in the household, i.e. a "correction" for large households that will increase the likelihood of being in poverty. 19 In this Annex, we depart from the rule explained in the PAMS above of 6 groups of RH plus a 7t group of "Rentiers". We add from the Burkina Faso case, used as an example, 'Self-Employed", "Unemployed" and a small category of skilled workers in the Urban Non-Tradable Urban" sector. Page -50 PAMS - Annex 5 The sheet "McrSht" '_ UE f ' Click Below to Update At Individual Level (Weighted) At Houselold Level (Unwuighted) LExpenihftu (th3, . EpeJdInhi,u;I lncom . ,Jncone rnee Upcte Coovr rhe aBe Yea and 1@ Yean ahead Pjecdon. alick here to update - Update theBmeIini< - 'l The baseline scenario sceiirlonr - HU: Check that Excel i set In Automallc Calculaion mode (Menu 'Tools', then 'Optlons, then 'Calculatlon) Chek thdt Household Survey Dete Bess i Open The sheet 'Assumpt" is where assumptions are inputed. It contains (in columns): a. Two poverty lines: In the Rural poverty line, the user has to inDut in the areen highlighted cells, the starting level ($/day or local currency/day) of the Poverty Line and the rule for indexing it to a given price index (or not). The Rural Poverty Line will be the threshold against which the incomes of households in the following categories will be assessed: Rural Non tradable Unskilled (RNTU), Rural Tradable Unskilled (RTU), Self-Employed (RSE), Unemployed and other non-active (UNEMP). A similar requirement applies for the Urban poverty line. Again, the Urban Poverty Line will be the threshold against which the incomes of households in the following categories will be assessed: Public Sector (PS), Urban Non tradable Skilled (UNTS), Urban Non tradable Unskilled (UNTU), Urban Informal (UIU), Capitalists (KAP), and Urban Tradable Skilled (UTS). b. Assumptions of Nominal Income Growth by category. Page -51 PAMS - Annex 5 The sheet 'AssumDt" Assumptions for Projecton Scenarios, Growth Rates (Nominal, Annual ) I Obs~ Year Poverty lines Ocupatisnel Celeganes, Rapresentative Groups E i , '. . | . ! i i . 1 ' 1 , . .. ''. 1 _ ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~11. .ea'Po-nlyLine 1VSda 2dv last updtstlch nW!heEPrditure. . . Growtl Rdes q! Gravrh reen at "mrg htD e (nu ; Po,rty Lbe of Ocupatonal CeSorIs, Rep-messetwb Groups riflation conedl only- Base Y 'I9gtl 0.0% 30% 30% 9 0.0%. O.03% 0.0%! 00%, 00%', 00% d%| o 0% I 1999 0 a 0. % O 30% 30% 0.0%, 0.0%! 00%1 00%- 30Y%i 0.0% 0.0%j 30% 200~0 03%! OO0% 0.0%! oO%[ 070%! D.O%| dml dn%! ! % 0% 0%' 00%j 00%! 0.0%; 3 2001i 3.iY| 36% 1%! 7.2% 7.2% -2.2% 12 7%34 1% 2.5%W 2.5% 11.0%1 0.0%, 2002 1.F%; 1% 0.9%! E.1% E.1% 03t 193% 59% 12 F%' 2 %' 4.7%' 30% 5 203 0.9%' 03% 1.0%! 3.%! 3,8% 58% i63% E 1% 01%, 25%: 5 0.5 (% o :5 2004i 1.6%! 1.6% 1.0% 0.2%i 32% 10 % 14 9%$ 59% 123% 25%1 6.5%1 0.0,% 7 210d 15.; E Ia I1C% 22% 22% 49% 14 5% 73%k 12%0 25% 57% 0 .% 2006 E.S0%: 605 1%0% 20% 20t3% 80% 129%' 8'% 119% 25h 5 2%: 0C% r Up d4e 9 2C07 .1% 30%- 10 22% 2 2% 7d% 12Er%, 87% lI5% 25%1b 58%, 0% ! rowt 0 2C03 3 0v 3 D 10% 2 1% 21% 7 1% 12 2% BB% 11 4%j 2 5% 5 4% 0 0% . 2009 SJ~ 311 3r 10%h 1 9% 1 9% 7 2% 11 9% 59% 11 9% I 25% S 1%, 00% .. By 12 2010 30 % 30% 10% 17%. 7% 7 3% 1165%, 90%*109%1 26% 5S0%, 0°% Categor 13 2311 3Ir%' 30% 10% 4$% 1%4 741% 114%, 90%,1(?%. 26% S% 4 ' F -- Id 2012 3'hIo 3Dit 10% 12% 1 2% 75ib 11 I% 91% 105%, 2bit S 0%; dV Fr0o% IS 2013 31J% 30IC 10% 09% 0 9% 77% 10 9%' 92% 102 % 25% 4del% 0 0% LP- There are two modes that can be used to input the assumptions on Nominal Income Growth by category. The first is -as mentioned earlier-to input 'exogenous" numbers for each year and each category. The second is to use the Excel Macro button 'Update Growth By Categoty from the LP". This instruction will seek the relevant growth rates in the Labor and Poverty LP Module of the! PAMS, provided that there is one for the country being analyzed and that it has been stored in the same directory of the HHProj. It is also possible to relate directly each cell of this sheet to the relevant cell of the Labor and Poverty LP Module using an Excel link or an hyperlink. Page -52 PAMS - Annex 5 The sheet "ORIGIN-CFA" The "ORIGIN-CFA" sheet contains the income or expenditure series by category, using the household data from HHbase Household Survey Data for Burkina Faso Per Capita Expenditures Demominated In CFA Francs 1 1 31,~2180 31.560! 1,430! 7~35 LO 520 45,9751 10771048; 0 18,900! 3,471 39.072 __ 185 954 ~ 20___ 58.31 13457 0 26,735 10JJ078; 3i 4.1 .8.0 .~d b . . ..... . ..... 44414 30178 4,960, ~I0)6 312. 7,8 i5) 0 26,65 140 4 43171 34,341: 5,881) 1 1889: 40,650: 71.5931 128,30)0 0 23,1501) 1 2,9 And the individual weight by category Individual Weight 1i 4940.95 1359 121 1096:271 2603.761 1352.7 418.92 116961 0 1292.9 6450.08i 21i065.9 443.68: 1479 4534.541 276.43' 699.8 1011.91 0 - 58139- 8 31 59j2 298 83976 4961.6 2-3-61 . 51 -77-3571 1250.1 690.611 0 780.56 32 4I 6603 1 10683 119085 21631 8976 102081 7 -62105i 0 1556.4 522 * 1 916091 1 006 234915 1457375' 79266 762451 8752 0 1506.72 3601.73 44368: 877.68 497 7110261 81 838.6 5 76. 1075.68 0 2536.71 3579.91 Any zero "0d in the column "UTS-US" means that that category does not exist in the simple. Other sheets in HHProj The 'ProjCalc' sheet contains all the calculations needed to produce the poverty and income reduction indicators. The Projections sheet contains the primary output calculated in "ProjCalc'. 'CHARTS' plots poverty incidences and income distributions indicators over ten years. "BASE' stores the baseline case (see above). "Chart 1" and 'Chart2' provides the pictures of PO and Ginis for the Baseline Case. 'Chart3" provides the picture of the current simulation plotted against the Base as a first difference. 'OUT-TABLEl" contains the first set of summary tables "OUT-TABLE2" contains a detailed set of indicators by category over ten years. IV-2- Five steps for simulation of a Scenario Caution: Before starting running the HHProj, make sure that the following options are active under Excel's 'Tools" bar menu, and 'Options' command: The 'Calculation" (instruction) has to be set into an "Automatic" mode. Page -53 PAMS - Annex 5 Step 1: Fill out the assumption sheet ("Assumpt). Chose one of two modes. Stand-alone simulation or Macro-consistent simulation. In the laKter case, make sure that the information contained is consistent with or linked (using the Excel Macro) to the Labor module of the PAMS. Step :2: Make your decision about the type of simulation (income or expenditure, weighted or not) - you would like to use - by clicking on one of the four buttons of the sheet "McrSht". Step :3: Wait while the macro processes the calculations. Recall that each cell in the HHS of the base year is now "multiplied" by the growth rate of the category to which the household head belongs to. Step 4: Get your results in 'CHARTS", 'Chartl", Chart2" , "Chart3", "OUT-TABLEI", and 'OUT- TABLE2". Step 15 (option): The simulator can also update the based line scenario. You have to click on the baseline scenario button. Page -54 PAMS Annex 6 Annex 6: 20 Application of PAMS to Burkina Faso: Poverty and Distribution associated with the current macro framework After the completion of its full PRSP, Burkina Faso is receiving a Poverty Reduction Support Credit from the World Bank (IDA). This annex summarizes the findings obtained using PAMS to project ex-ante the changes in the social and poverty indicators that one can expect from this credit over the medium-term (up to 2010). We first set the context. Burkina Faso is a poor landlocked country with a limited resource base, high vulnerability to external shocks and very low human development. Since 1991, the country has embarked upon a comprehensive reform program and has made to date significant headway in its transition towards a market-oriented economy and a more selective role of the State. Despite the good progress achieved, the country remains one of the poorest in the world. Real GNP per capita was estimated at US$230 in 2001. In the same year, on the basis of its human development index, Burkina Faso was ranked 159th out of 162 countries. According to the most recent poverty survey (1998), some 45.3 percent of the population lives below the poverty line. I. Macroeconomic and Structural Reforms Stabilization and Growth. Over the last decade Burkina Faso has established a relatively strong track record on macroeconomic performance. Since 1991 the country has implemented a wide range of economic reforms under a series of stabilization and structural adjustment programs supported by the Bank, the IMF and other donors. Real GDP growth picked up in 1994, after declining during the first half of the 1990s, and averaged 5.1 percent per year between 1997 and 2001. The relatively strong growth performance is attributable to the competitive gains following the 1994 CFA Franc (CFAF) devaluation, the large public investment program, and to financial and structural policies aimed at consolidating the market orientation of the economy and maintaining macroeconomic stability. Inflation, as measured by the consumer price index, steadily declined from 6 percent in 1996 to -0.2 in 2000. Real GDP growth decelerated to 2.2 percent in 2000 because of adverse weather conditions which resulted in a smaller cotton crop and a significant cereal deficit, and exogenous shocks.24 In 2001, real GDP growth reached 5.7 percent. The primary sector performed well and grew by 12.7 percent. Cotton production increased by 45 percent from the 2000 campaign, reaching 400,000 tons. This was due to good rainfall and an increase in farm-gate prices, which, in turn, led to an expansion of the cultivated area by 35 percent.22 The extemal current account deficit, excluding current official transfers, declined from 17.6 percent of GDP in 2000 to 15.9 percent of GDP in 2001. 20 The descriptive sections of this Annex draw on the Intemational Development Association (IDA) proposed Poverty Reduction Support Credit to Burkina Faso, April 17, 2002, Washington DC 21 Shocks included the increase in oil prices; the appreciation of the US dollar; and the deterioration of the political situation in Cdte d'lvoire, where 3 million Burkinabe reside, a factor that significantly reduced workers' remittances. 22 Producer prices were raised in 2001 by 10 percent (from CFAF 159 to CFAF 175/Kg of seed cotton). In addition, a bonus of CFA 25/kg was paid, representing the distribution of half of SOFITEX pretax profit in the previous campaign. Page- 55 PAMS Annex 6 The fiscal performance in 2001 was mixed: the overall budget deficit (including grants) reached 4.8 percent of GDP (0.3 percent lower than programmed). The mobilization of fiscal revenue, however, was low, reaching 12.5 percent of GDP (1.5 percent of GDP lower than programmed). External assistance was also lower than expected, due to late delivery of certain creditors' contributions under the enhanced HIPC Initiative. Nevertheless, savings on current and capital expenditures contributed to offset this shortfall in revenues while budget expenditure in the social sectors continued to increase. The 12-month inflation rate in 2001 reached 4.9 percent, due to an upward pressure on food price resulting from the previous year's deficit in cereal production. Higher inflation and slight decline of the US Dollar against the Euro resulted in an appreciation of the real exchange rate, although the gains in competitiveness achieved through the 1994 devaluation were largely preserved. Table A6- 1(a). Burkina Faso: Selected Economic Indicators (1 999-2004) Actual Projections Estimate 1999 2000 2001 2002 2003 2004 Real GDP Growth (%) 6.3 2.2 5.7 5.7 5.4 5.4 GDP per capita (%) 3.4 -0.6 2.9 2.9 2.6 2.6 Population growth (%) 2.8 2.8 2.8 2.8 2.8 2.8 Inflation (CPI, %) -1.1 -0.3 4.9 2.0 2.0 2.0 Tax Revenue/GDP 11.9 12.2 .12.8 13.8 14.1 14.7 Overall Fiscal Balance/GDP Incl. Grants -4.0 4.2 -4.8 -5.9 4.4 -2.7 /excl. Grants -13.3 -12.7 -13.0 -13.5 -11.6 9.5 Current Account Balance/GDP, incl. Transfers -12.8 -14.6 -12.4 -10.3 -9.0 -7.9 /excl. Transfers -16.0 -17.6 -15.9 -14.1 -13.0 -12.1 Structural Reforms. Burkina Faso has been implementing a wide range of structural reforms in recent years. The country remains the most compliant member regarding the regional norms of the West African Economic and Monetary Union (WAEMU). The Govemment has greatly improved its budgeting and expenditure management practices and established a multi-year program to enhance staff capabilities on public financial management matters. Under Phase IlIl of the prvatization program (2001-2004), the Govemment aims at privatizing or liquidating 21 public enterprises, including the main public utilities (ONATEL, SONABHY and SONABEL). The gradual liberalization of the cotton sector has continued. The State has become a minority shareholder of the cotton company SOFITEX (36 percent of capital) and has already indicated its intention to reduce further its share in favor of producer's associations (currently with 30 percent of capital). The monopoly of SOFITEX was ended in December 2001 and two new cotton zones have been opened to private investors. A number of important measures have been implemented in the area of civil service reform, including the adoption of a more compressed salary grid applicable to all civil servants, the interconnection of the payroll and civil service databases, and a new system of merit-based promotion. Furthermore, it is expected that the Supreme Audit Court (Cour des Comptes), with senior jurisdiction over the control of public finances, will be fully operational by end-December 2002. II. Medium-Term Prospects and the Baseline scenario from the RMSM-X Real GDP growth could reach 5.7 percent in 2002. SOFITEX is expected to break even in the next cotton campaign, and a further increase in cotton production is expected. Seed cotton producer price will remain unchanged at CFAF 175 per Kg. Assuming a normal cereal crop, the primary sector is expected to grow by 3.2 percent. Growth will pick up in the secondary and tertiary sectors (6.6 and 7 percent respectively), as a result of stronger demand linked to the 23 In reality, producer price will decrease by 12 percent, in that the bonus paid during the current campaign, amounting to CFAF 25 per Kg of seed cotton, will not be paid to the producer during the next campaign. Page- 56 PAMS Annex 6 economic recovery. Inflation should decline to some 2 percent. The current account deficit, excluding current official transfers, is projected to improve to 14.1 percent of GDP, as a result of substantial increase in the volume of cotton harvested in late 2001 but exported in 2002. The overriding fiscal objective for 2002 and the medium-term is to consolidate Burkina Faso's budgetary position, which entails significant efforts to increase revenues, to improve budgetary management, and to introduce greater efficiency in public spending to support the govemment's poverty reduction program. For 2002, fiscal revenue is projected to rebound to 13.8 percent of GDP. The disbursement of HIPC resources will be accelerated to make up for the slow start in 2000-2001. Expenditure will continue to be contained to levels compatible with revenue performance, while social expenditure will continue to increase in 2002 (See Annex 6). For the medium term, assuming no exogenous shocks and the maintenance of sound economic policies, growth could remain at around 5.5 percent. Burkina Faso's PRSP set an ambitious target for the decline in the proportion of people below the poverty line, from some 45 to 30 percent between 2000 and 2015. In the light of recent exogenous shocks that lowered growth in 2000, lower than expected growth in the medium-term, and the continued vulnerability of the Burkinabe economy to export shocks, this ambitious target may not be attained unless significant efforts are taken to ensure that growth will be sustained and pro-poor. In order to do so, it is necessary to find an appropriate mix of fiscal, monetary and public investment policies that would bring the economy to a higher growth path while maintaining an adequate level of consumption. In the light of the shortcomings of current macroeconomic models utilized by the authorities a new growth/poverty scenario for the medium term is being elaborated in collaboration with the World Bank. This work will benefit from the utilization of new macroeconomic modeling tools that can simulate the impact of altemative policy packages, and shocks of various kinds, on growth, poverty and distribution for the population as a whole and selected socio-economic groups. The first result, including the presentation of growth/poverty scenario for the period 2002-2015, will be presented by end-April 2002. Page- 57 PAMS Annex 6 Table A6- 1(b). Burkina Faso: Projected Baseline Economic Indicators (2001-2010) Der. Indicator Table for the Public Closure 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 Growth Rate of GDP 6.2% 5.8% 5.7% 5.5% 5.5% 5.00/0 5.0% 5.0% 5.0%/o 5.0% Absorption Growth 5.0% 5.4% 5.7% 5.3% 5.4% 4.8% 4.8% 4.8% 4.8% 4.8% Consumption Growth 3.4% 5.5% 6.9% 4.9% 5.6% 4.8% 4.8% 4.8% 4.8% 4.8% Private Consumption 2.9% 3.3% 7.2% 4.9% 5.6% 4.3% 4.4% 4.5% 4.6% 4.3% Investment Growth (GDI) 10.2% 5.0% 2.0%h 6.6% 4.8% 5.0% 5.0%!, 5.0% 5.0% 5.0% Investment Growth ( GDFI) 10.2% 5.0% 2.0% 6.6% 4.8% 5.0'!, 5.0% 5.0%1. 5.0% 5.0%/6 Import (GNFS) growth 4.8% 4.4% 3.9% 3.9% 3.7% 3.3% 3.2% 3.2% 3.2% 3.2% Export (GNFS) growth 15.7% 6.9% 1.0% 3.5% 2.1% 2.3% 2.3% 2.3% 2.4% 2.4% Real Per Capita Growth Rates: Gross Domestic 'roduct (GDP) 3.3% 2.9% 2.8% 2.6% 2.6% 2.1% 2.1% 2.1% 2.1% 2.1% Total Consumption 0.6% 2.6% 4.0% 2.0% 2.7% 1.9% 1.9% 1.9% 1.9'!. 1.9% Private Consumption 0.0% 0.4% 4.2% 2.0% 2.7% 1.4% 1.5% 1.7P% 1.8% 1.5% Per Capita USS Le.vels: Gross Domestic Plroduct (GDP) 204 219 235 247 259 269 279 289 300 311 Total Consumption 182 193 208 216 227 235 243 251 260 269 Private Consumption 164 171 184 192 202 207 214 221 228 235 memo item: GDP in USS 2369 2608 2877 3111 3365 3586 3822 4073 4341 4627 Debt and Debt Ser vice (LT+ST+IMF): Total DOD(USSM) 1396 1488 1568 1677 1820 2020 2234 2500 2792 3121 Total Debt/GDP 58.9% 57.1% 54.5% 53.9% 54.1% 56.3% 58.5% 61.4% 64.3% 67.5% Debt Service (USSM) 71 60 60 70 26 44 70 132 134 180 Debt Service/Total Exports 22.2% 18.2% 17.0% 17.9% 6.4% 10.2% 15.6% 27.9% 27.00% 34.7% Debt Service/GIDP 3.00/6 2.3% 2.1% 2.2% 0.8% 1.2% 1.8% 3.2% 3.1% 3.9% Interest Burden (LT+ST+IMF): Interest Paid (USSM) 19 -3 -2 0 7 30 44 60 80 102 Interest Due (US$M) 19 -3 -2 0 7 30 44 60 80 102 InterestlTotalExports 5.9% -0.9% -0.6% 0.1% 1.6% 6.9% 9.8% 12.7% 16.1% 19.7% Interest/GDP 0.8% -0.1% -0.1% 0.0% 0.2% 0.8% 1.2% 1.5% 1.8% 2.2% Goods Market As a Share of GDP in LCU: Resource Balance -17.6% -16.7% -15.6% -15.0%/ -14.7% -14.5% -14.4% -14.2% -14.1% -13.9% Exports 11.3% 10.7% 10.3% 10.3% 9.9% 9.7% 9.5% 9.3% 9.1% 8.9% Imports 28.9% 27.4% 26.0% 25.3% 24.6% 24.2% 23.9% 23.5% 23.1% 22.8% Consumption 88.90% 88.4% 88.6% 87.7% 87.6% 87.4% 87.2% 87.00% 86.9% 86.7% Private 80.4% 78.3% 78.6% 77.8% 77.7% 77.1% 76.7% 76.3% 76.1% 75.6% Public 8.5% 10.2% 10.0°/ 9.9/o. 9.9% 10.2% 10.5% 10.7% 10.7%! 11.1% Investment 28.7% 28.3% 27.1% 27.4% 27.1% 27.1% 27.2% 27.2% 27.2% 27.2% Private 12.6% 12.9% 13.1% 12.7% 13.5% 13.5% 13.6% 13.9% 14.4% 14.6% Public 16.1% 15.4% 14.0% 14.7% 13.6% 13.7!/o 13.5% 13.3% 12.9%h 12.7% GDFI 28.7% 28.3% 27.1% 27.4% 27.1% 27.1% 27.2% 27.2% 27.2% 27.2% Changes in stocks 0.0% 0.0% 0.00/. 0.0% 0.0% 0.00% 0.0% 0.0%0 0.0% 0.0% Gross Domestic Sivings 11.1% 11.6% 11.4% 12.3% 12.4% 12.6% 12.8% 13.0% 13.1% 13.3% Total Savings 28.7% 28.3% 27.1% 27.4% 27.1% 27.1% 27.2% 27.2% 27.2% 27.2% ForeignSavings 13.3% 11.5% 10.7% 9.9% 9.9% 10.1% 10.2% 10.4% 10.7% 11.1% Gross National Savings 15.3% 16.8% 16.4% 17.4% 17.3% 17.0% 17.00/o 16.8% 16.5% 16.1% Monetary Savings. 0.0% 0.0% 0.0% 0.0%h 0.0%0 0.0%0 0.0% 0.00% 0.0% 0.00% Government Savilgs 3.1% 3.3% 3.4% 3.6% 3.6% 2.3% 1.6% 1.0%0 0.4% -0.4% Private Savings 12.2% 13.5% 13.0% 13.8% 13.7% 14.8% 15.5% 15.8% 16.1% 16.5% Page- 58 PAMS Annex 6 Ill. Poverty Profile and Social Indicators Poverty Profile. In 1996 and 2000, the Government issued poverty profiles based on the results of priority surveys conducted in 1994 and 1998. Over this period, poverty'incidence remained broadly stable (from 44.5 percent in 1994 to 45.3 percent in 1998). Poverty in Burkina Faso is predominantly rural, accounting for 94.5 percent of national poverty, and the incidence of poverty is markedly higher in rural (51 percent) than urban areas (16 percent). But the incidence of urban poverty has increased from 10 to 16 percent between 1994 and 1998, and this has been accompanied by an increase in urban inequality. The southem region of the country has the lowest poverty incidence, at 37 percent, while the center-north part of the country displays the highest level, 61 percent. Income disparities amongst socio-economic groups vary dramatically. The richest 20 percent of the population account for over 61 percent of aggregate national income, whereas the poorest 20 percent account for less than 5 percent. The analysis of poverty among socioeconomic groups (based on source of income) shows that between 1994 and 1998 the incidence of poverty increased for all groups except cash crop farmers. It is highest among food crop farmers, who account for most of the population living in poverty. Table A6- 2. Burkina Faso: Distribution of Welfare Across Economic Regions (1994 and 1998) Headcount Index (%) Poverty Gap Index Relative Contribution (%) (%) 1994 1998 1994 1998 1994 1998 Wage pubilc 2.2 5.9 .40 1.6 .20 .50 Wage private 6.7 11.1 2.2 2.5 .40 .70 Artisans/Commerce 9.8 12.7 2.8 2.7 1.4 1.6 Other activities 19.5 29.3 6.4 7.0 .30 .40 Exportfarmers 50.1 42.4 13.8 12.5 11.8 15.7 Subsistence 51.5 53.4 16.3 16.5 78.9 77.1 Farmers Inactive* 41.5 38.7 14.5 12.1 7.1 4.0 National 44.5 45.3 13.9 13.9 100.0 100.0 *Category including all observations that did not fit the remaining socio-economic groups. Social Development. Expenditure on social sector (including the utilization of HIPC resources) increased from 4.6 percent of GDP in 1996 to 6.2 of GDP in 2001 .24 This reflects the commitment of the authorities to execute the poverty reduction program embedded in the PRSP in its entirety. Overall, social expenditure in 2001 (including HIPC resources) is estimated at 23.3 percent of total expenditure, up from 18.3 percent in 1999 and 20.6 percent in 2000 (see Annex 6) Despite the substantial increase in govemment resources allocated to social services, commensurate progress in social and human development has not been forthcoming. Although key and education and health indicators have improved in recent years, most social indicators of Burkina Faso are poor, even by Sub-Saharan Africa standards (see also Part IV. B, Section 2). Table A6- 3. Burkina Faso: Selected Health Indicators (1998/99) Infant Mortality Under five Life Rate (per mortality HIV Malnutritio Country expectancy 1000 lIve (per 1000 Total Prevalence n Rate at birth births) live births) Fertility Rate (%) (weight for (years) Rate age, %) Burkina Faso 46 105 219 6.8 7 32.7 (34.3) SSA 52 91 151 5.6 8 32 24 Expenditure refers to the sum of current and capital expenditure. Social sectors indude health, educaton, women's welfare and other social expenditure and rural roads. Page- 59 PAMS Annex 6 The PRSP Process PRSP Main Objectives. The PRSP outlined and prioritized the Government's poverty reduction strategy along four main pillars: (i) accelerating broad-based growth; (ii) ensuring that the poor have access to basic social services, (iii) expanding opportunities for employment and income- generating activities for the poor and (iv) promoting good govemance. The PRSP focused mainly on three priority sectors - education, heath and rural development - where public interventions traditionally have had the highest payoffs in terms of fostering economic growth, increasing employment opportunities, and raising the Burkinabe standard of living. PRSP First Year Progress Report. Burkina Faso had unforeseen constraints in executing its priority poverty reduction program, chiefly exogenous shocks and lower than expected receipt of funds under the enhanced HIPC Initiative because of delays in finalizing agreements and non- participation by certain creditors in 2000 and 2001. Despite these constraints, the Govemment has continued to promote an enabling environment to foster growth and carry out its poverty reduction program, ensuring the effective implementation of priority actions, as listed in the PRSP. In October 2001, the Govemment presented its first PRSP-PR, prepared with extensive consultation with civil society and the donor community. The PRSP-PR and related JSA were presented to, and endorsed by, the Boards of the World Bank and IMF Boards on November 30, 2001 and December 6, 2001, respectively.25 Overall, the balance of the first year of the implementation of the PRSP was deemed satisfactory. In particular, notable progress was achieved in intensifying the debate on poverty reduction in the country, fostering the participatory process associated with the implementation of the PRSP and its revision, and in the health sector where most of the PRSP targets for 2000 were met or exceeded. Areas lor further improvement were also identified, including: (i) updating of the medium-term macroeconomic framework and its links to poverty reduction, in the light of lower than expected domestic growth for 2000 and beyond and global economic slowdown, (ii) taking concrete action with a view to setting up a monitoring and evaluation system for the PRSP and strengthening the quality of social statistics, and (iii) developing a global vision on rural development, by relying on the program outlined in the PRSP. Since the elaboration of the PRSP-PR, efforts have continued to press ahead with the implementation of the poverty reduction strategy (See Annex 3). Moreover, the proposed PRSC 11 has taken concrete measure to address some of the recommendations that emerged from the PRSP-PR exercise, as reflected in the Matrix of Policy Action (MPA), presented in detail in Annex 9 and discussed in the Section IV below. Support to the Implementation of the PRSP: The PRSC Program. Rationale. The overarching objective of the PRSCs series is to support the implementation of Burkina Faso's PRSP. PRSCs provide highly concessional financial assistance to help Burkina Faso implement the PRSP on the understanding that the resource allocation mechanism is supporting efficient, poverty-oriented expenditures in all sectors. These credits do not provide support to all the policy measures but selectively focus on key policy and institutional reforms in areas where IDA has a comparative advantage. Given the emphasis In the Burkina Faso PRSP on public expenditures as the main tool for PRSP implementation, the focus of PRSCs is predominantly on the reform program to improve the effectiveness of public spending, as well as the transparency and accountability of public resource management. PRSCs, nevertheless, also 25 See IDA/ (...). In the JSA, finalized in October 2001, the staffs of the World Bank and IMF stated that the country's efforts to implement the strategy provide sufficient evidence of its continuing commitment to poverty reduction and, therefore, that the strategy continues to provide a credible poverty reduction framework and a sound basis for World Bank and Fund concessional assistance. Page- 60 PAMSAnnex6 support key sectoral reforms, in the sectors highlighted in the PRSP as priority for poverty reduction. Sequencing. The time horizon of the PRSC series corresponds to the PRSP and CAS periods. The CAS, finalized in November 2000, supports the Government efforts to achieve sustained growth rates, reduce the incidence of poverty and improve the health and education of the rural population, as described in the PRSP (See Annex 4). The Bank's Board approved the first annual single-tranche PRSC for Burkina Faso on August 23, 2001. The credit established a rolling medium-term policy framework setting out a three-year reform program aiming at implementing the PRSP, with specific progress benchmarks and outcomes indicators defined and agreed with the Govemment and all donors. The rolling nature of the policy framework should enhance predictability of funding for the Govemment and allow the Bank to better monitor progress towards the PRSP target indicators. To the extent feasible and depending on country performance, future PRSC series will be aligned with the country's budget cycle to make sure that resources are available during the first quarter of the calendar year. 2 IV. The Results of the PRSC as a Baseline for PAMS We took the macro scenario outlined above for the PRSC to project a macroeconomic framework with the RMSM-X. The results of this initial run are in Table A6-1(b). The simulated growth story looks quite promising. In particular, there is a steady 2.6% p.a. average growth of per capita income in Burkina Faso during the years of the program, followed by a continuous 2.1% growth in per capita income from 2006 onwards. This high level of growth -and its steadiness-requires a continuous investment effort (of about 27-28% of GDP) and is accompanied by a required inflow of resources to finance imports of about 14-15% of GDP. Therefor3, after the HIPC operation, the baseline scenario still projects an increase in debt-to-export ratios that will have to be addressed by other structural policies relevant to the extemal balance. Then, we incorporate this macro framework into the PAMS. We construct a baseline case with PAMS. The picture projected by PAMS looks as follows. For the labor market PAMS projects a stable participation rate. The labor force grows from 6.56 million in 2002 to about 8 million in 2009-2010. We assume a small migration of about 30,000 workers per annum between the rural and urban areas and a labor supply in the urban areas (respectively rural areas) that grows from about 800,000 in 2002 to 1.2 million in 2010 (respectively 5.7 million in 2002 to 7 million in 2010). Labor demand, in turn, in the urban areas (respectively rural areas) grows from about 440,000 in 2002 to 620,000 in 2010 (respectively 5.1 million in 2002 to 5.2 million in 2010). Therefore, unemployment rates tend to grow more in rural areas (and thus cause migration), rising from about 10.5% in 2002 to about 25% in 2010. In urban areas, there is no significant growth in unemployment rates but the overall level remains high throughout the simulation period (from about 45% in 2002 to 46% in 2010). Given the relative sectoral growth rates, labor demand grows faster for both the tradable goods rural and urban sectors, while by construction, there is no 'skilled" unemployment in Burkina Faso. 26 Bank lending to Burkina Faso will increasingly take the form of results-driven programmatic credits (PRSCs) and self-standing projects for capacity building, support for Community Driven Development activities (concentrating on rural, social or HIV/AIDS interventions) and infrastructure investments directly targeted to poverty alleviation and private sector development. There will be a continuation of analytical and advisory activities as knowledge is the critical input for progress in the substantive areas presented in the PRSP. Page- 61 PAMS Annex 6 The baseline scenario maintains throughout the 2002-2010 period the structure of public expenditures of the most recently audited budget (that of 1998). The taxes and transfers are therefore projected accordingly. In that context, the projected net (after income taxes) nominal income after transfers can be seen in Table A6-5 below. The beneficiaries of the relatively high growth performance of Burkina Faso are first the skilled workers in the tradable sectors of the economy. Their relative scarcity and the growth in demand explain that their income grow by about 10-12% in urban areas and 5-8% in rural areas on average during the 2002-2010 period. In contrast, unskilled labor in informal sectors in urban areas benefit from budgetary per capita lump sum transfers and their nominal income grow by 12-14% in nominal terms while unskilled workers in rural areas benefit less from growth and their incomes grow by 1 to 2% only. As a reisult of these assumptions and results, PAMS can project the poverty indicators for Burkina Faso (see Table A6-6 below). Note that there is a difference between the official Govemment poverty headcount (P0) initial number (45%) and PAMS' initial P0 (66%) due to the different poverty line that is used ($1/day in rural areas and $2/day in urban areas on a PPP basis) and that is higher than the official poverty line. There are also other minor differences between the way the Government projects its own poverty line over the simulation period, and PAMS' indexation of its 2 poverty lines to the RMSM-X inflation rate measured by the consumer price index. Figure A6-1: What is interesting is BURKINA-FASO: Poverty and Distribution Indicators PROJECTED the path obtained by (o.* Avo ln- 1. L100% PAMS for the projection 160,000i - 0% period (see Figure A6- 140,000 90% 1). The results show 120,000 80% that the PRSC baseline 10,000 -70% scenario does manage tO0,000 - 60% to reduce P0 by about 4 80,000 .50% percentage points 60,000 .. 40% (compared to a target of 30% reducing it from 45% to 40,000 20% 30% --but in 2015--). A 20.000 -Average noDno-PO -Gini1 linear interpolation of . .% -Gini-uraj Gi-U,ban our result would be that 1996 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 P0 in 2015 would be around 60% (measured by PAMS poverty lines). While this is encouraging, PAMS' results suggest that growth alone - and even relatively high growth-will not be enough to significantly reduce poverty in Burkina Faso, even when different measures of poverty lines are used. More worrisome, the baseline scenario shows that there is an increase in inter-group inequality in Burkina, measured by the inter-group Gini. The overall inter-group inequality increases from about 0.36 to 0.46. This result is consistent with the sectoral growth that benefits essentially the incomes of urban areas vis-a- vis that of the rural areas where there is a higher concentration of poor people. There is much more work that is required to refine the simulations of the baseline case. Furthernore, after refining the baseline, more work is needed to simulate 'corrective' policies that would reduce poverty further and at least stabilize the level of inequality in the country. One can think for instance in using additional fiscal instruments (e.g., social transfers, differentiated tax rates, etc.) to produce a more likable altemative scenario within the macroeconomic framework defined by the PRSC. Nevertheless, the objective here is simply to note that PAMS is capable of identifying issues that are relevant for the policy discussion using a consistent three-layer framework that shows poverty indicators projected consistently with a given macroeconomic framework. Page- 62 PAMS Annex 6 Table A6-4: Burkina Faso: Projected Baseline Labor Market (2001-2010) by PAMS LABOR 20 "205 20 20 LABOR FORCK (-I45Sd4oinpoyWd) 6.31 6.37 6,56 6.74 6.93 7(13 7.33 7.34 7.75 7.97 8 19 LABOR DEMAND1 URBAN LABOR 039 0.41 (044 0.46 0.48 030o 0.52 0.59 0.37 0359 0 62 S8818.4 0 24 0.26 0.24 0 24 024 0.24 0.25 0.25 0 23 027 0 25 U-.Id0d 0 Is 0.17 0(9 0 22 0 24 0.26 0.20 0 30 0 32 0 34 0.36 P.bIk S.-8o 0 23 0.23 0.23 0.23 0 23 0 23 0 23 0.23 0 23 0 23 0 23 T,.d.h5 Goo6 Skill.46(p(4y ~94o1046 0.00 046 04 046OD 0 46 0.46 O) 46 04 046OO 046 N.-T,.d.bk. SkflI6d (.lyo0 .ds ) 0.01 001l 0 01 0.01 0.01 001 0 01 001l 0.02 0 02 002 UJn66ilkd 0 05 046 0406 0.46 0.07 0.07 0.07 000 0 09 000 00(9 I46o,5(..kidId) OI1 0.A2 0 13 0 15 0.17 0 19 0.21 0 22 0 24 0 26 0 20 RURALLABOR 5.16 4.99 5.13 3 23 322 9 24 S.23 3 22 3 22 3 21 3(19 F-A~. (.,p,o8) 4-0d (.03 1.01 ( 03 1.07 I IS 11 is (.25 1.31 I 30 1.45 1.52 Ur.keWId-nI& SOf-E.opIyo 3 88 3 98 4 10 4.18 4 00 4.06 3 98 3.91 3094 3.76 3.67 LABOR DEMAND TOTAL 0356 3 41 3.37 3.70 5 71 3 74 5 76 3 771 979 34 908a LABOR SUPPLY tyBNLBR(Wt iyw U0 71 0 73 0.79 0,03 0.07 0.92 0.46 I101 1.06 1.11 1 16 06k8.d 0.24 0 24 0 24 0 24 0024 0 24 0 23 0 25 0.23 023 0 23 Ua.kHdO. 0.47 0351 0.33 0.39 0063 0367 0.72 0.76 0981 0896 0.90 P8b8k0.do19 opn46.9,I"d- .) 0.23 0 23 0 23 0.23 0.23 0 23 0.23 0 23 0.23 0.23 0 22 00k1.6 0.46 046 0.46 046 0.46 0.46 046u 046 0.46 0 46 046O E-.-W .i C-.My (O46na C.,il. F1s880 0.46 0 46 046 0.46 046 0 46 0460 0.46 0.46 0460 S(61.4 0.01 0 01 0.01 0 01 0.01 0 01 0.01 0.01 002 0.02 0 02 I.Jaw 0.pplydf46d,.4 nq.) 014 0. 13 0.13 0.13 0(35 0,16 0 16 0(16 0.17 0 17 0.17 UpSp.d-s 0 46 (.46 0 00 0.46 0.46 0.46 046 00 046 046 046 hh.f-i (.idO. 0 33 0.37 0.40 044 0.48 0952 0356 0.60 0.64 0 69 0.73 MIv.d- (- Rm w Ub- ~~~~~003 0 03 0.03 Q0.3 0.03 0.03 0 03 0 03 0.03 0.03 RURAL LABOR ((iW n d ppy. d~.W .by p4p..dd. . 0359 5.59 5.74 3468 6 03 6.18 6 34 6350 6.66 6.83 7460 F-" (.xpo .9)b04 1460 1.62 1 63 1 63. 1.67 (68 L.70 ( 72 1.73 1.75 1 77 U.4.1.dlnf-nW6A S.(f.aOwWo 3.99 346 4 10 4 23 4.36 4.50 4 64 4.78 4.93 5.00 3 23 LABORSOIPPLYTOTAL 6 31 6.33 6 53 6 71 6 90 7.10 7 30 7.51 7.72 7.94 5(16 UNEMPLOYNENr t8(RBAN LABOR 0 32 0.34 0 36 0 37 0 39 0.41 0 44 0 46 0.49 0351 0954 Sk088.4 046 046 046 046 046 0.46 0 46 0.46 0 00 0 00 0460 V.H.IdO4 0.32 0 34 0.36 0 37 0 39 0 41 0 44 0 46 0 49 0.31 0354 POtk S.I.8 0460 0.46 046D 04 046Ob 0.46 046D 046 0.46 046 0460 Tr.dAbI r-.& 31o(kd 0460 0.46 046 000O 0.46 0460 046 046 046 046 0.46 Sl1kd 10 00.00 046O46 .46 04 0.46 046 04 0460 046 0.46 O046 Ujn,kil8.d 0(10 0 09 0.09 009 0469 046 0.46 04 0.46~ 0.46 0468 W.-d (..ddW ~~~~~~0 22 0.23 0 27 0 29 0.31 0 33 0 35 0.38 0 40 0 43 0 46 RUR4AL LABOR 0 43 0.60 0.460 0 64 0891 0.94 1.11 ( 27 (.45 ( 62 1.81 F-.,d (.q.F.)o00iw4 0057 060 0460 0.38 0052 0.30 0 40 0.40 0.33 0460 0.23 1J,d.AW dl,08.ld& Sf.64,WoY.d 0 11 046 0 00 000 0.29 0644 0466 0587 (.49 ( 32 (.56 UNEMPLOYMENT RATE URBAN LABOR 4409% 4.9%A 44.9% 640%A 45 0% 45 1% 4505% 45.996 46296 46 394 46.7% 05k(1.4 0.0%A 0 0% 00% 0.0 00 O% 0.06A 0.0% 0.0% 0904 00% 0.0% v.w658. 67.7% 660% 64 6% 63.4% 624% 61.5% 61.1% 60.0% 604%A 4609 0%9.7% PI~.kS60.0 00%- 00% 00%- 00% 0,0% 0.0% 00% 0.0% 0.0% 0 0% 00% 50(8.4 0.0% 0 64 00% 00% 0.0% 00%A 0~0% 00%r 0.0% 0.0% 00% N.-Tr.46b8 SIoI8d 0 0% 0.0%A 00% 00% 00% 00% 0.0% 00% 0.0% 00% 0.0%A Uo.&Hkd 67.7% 595% 209% 57 9% 57 0% 5 90% 54.8% 53 6% 32,4% 0( 0% 49.5% bI.I.-I (-.d.0484 67.7% 6860% 660% 6302% 64 2% 632%l 6209% 62.7% 620S% 62.3% 62 1% RUBALLABOR 77% (008% (0.3% (00% (3 4% 15.3% 17.5% 196% 21.7% 2308% 2530% F-.W. (.0.pt) m*I0.4 35.9% 37.3% 36.9% 3532A 3(2% 29.6% 26 4% 23.50A 20 4% 17 2% 13.9% 12.66k8. I.f...( A S.L0f46Epkyd 2.7% 0.0% 00%A 1.3% 63% 9.9A 14 2% 18.2% 222% 26 0% 2909% Page- 63 PAMS Annex 6 Table A6-5: Burkina Faso: Projected Baseline Net Income after Transfers (2001-2010) by PAMS GROW7H RATES DANCOMI ArMtR TAXES AND WrIm TUAMsPZB UXEAN C.PIto4h&Ro,9m 110% 4.7% 5.5% 6.5% 5.7% 62% 5.4% 54% 51% 50% P.9bl s9 r 10% 0.9% 10% o 0% 10% 1.0% 1.0% 10% 1.0% 10% Tn.9.b G.d. Skil.4d 25% 2.5% 25% 25% 25% 2.5% 2.5% 25% 25% 26% 14o-T.dbI Sklld 2.5% 126% 124% 123% 120% 11.9% 11.6% 11.4% 111% 10 U1hilol 344% 69% 61% 609% 73% a.5% a7% as% 39% 90% 1.6,r.1 (.o.6a1.D 12.7% 193% 163% 14.% 14.5% 12.9% 12.6% 122% 11.9% 11H% RURAL Fo.o.(1,pl) eikl.d -2.2% 0.3% 5S% 10.1% 4.9% 9.0% 70% 71% 72% 7.3% Uskll IWd. R Sd1-E pyed 72% 6.1% 3.0% 0.2% 2.2% 2.0% 2.2% 21% 19% 1.7% TOTrAL 9.4% 4.9% 5.3% 5.5% 5.1% 5.6% 5.4% 51% 4.% 4 9% Table A6-6: Burkina Faso: Projected Baseline Poverty Indicators (2001-2010) by PAMS Country BURKINA FASO Table 1(a). Poverty Une and Income DISlulbuton, Year, TOTAL Stltistics by: e_ Ye_, PFn,ed Weighted Expendle 1999 1999 2000 2001 2002 2003 2004 2005 2008 2007 Povery line (i LCUeq.1-2SPPP/dfy 335 335 335 347 353 355 390 388 388 399 Pov rty lb. (in LCUyar) 80,509 80,509 80,508 83,350 54.822 85.275 85,418 87.803 93,039 95,778 Povyty One on mnt USWDyer) 0.57 0.54 0.47 0.48 0.50 0.51 0.52 0.53 0.55 0.58 Meals Income 97,792 97,792 97,792 103,531 108,684 113,400 116.734 120,870 125,928 130,691 Income GAP 0.40 0.40 0.40 0.39 0.38 0.38 0.37 0.37 0.38 0.38 TottlPopdaltion 10.730,330 10.995.700 11,274.000 11,594,119 11.923,142 12.292.302 12,910,978 12,988,271 13.338,002 13.712,077 Saliptte 5 (h-ehold) 9,933,181 9,933,191 9,933,161 9,933119 1 9,93 9,933,161 9.933.161 9,933.191 9,933.191 9,933,191 Poo, (HadcountIn semple) 9,543,950 9,543.950 8,543.950 6,487,42 9.300.491 8,116,018 9,081,232 6,002,025 9,134,898 8,140,020 PO (Heed Count index) 0.96 0.88 0.99 0.95 0.93 0.92 0.91 0.80 0.62 0.92 Pi (Poventy Gap) 0.28 0.28 0.29 0.25 0.24 0.23 0.23 0.23 0.24 0.24 P2 0.13 0.13 0.13 0.13 0.12 0.11 0.11 0.11 0.12 0.12 Ghin 0.38 0.38 0.39 0.39 0.39 0.40 0.40 0.41 0.42 0.43 Thedl 1.91 1.91 1.91 2.30 2.41 2.52 2.71 2.90 3.11 3.34 V. Alternative policies simulated by PAMS for Burkina-Faso Given that the baseline growth scenario is relatively optimistic, the objective of this section is to find within the base case, adjustments and new policies that will allow to reduce poverty further while at the same time controlling the rise of inequality. PAMS has three instruments to do so (we assume that these instruments can be implemented within the existing political framework of Burkina, or at least assuming that it will not cause excessive social and political unrest and/or that it could be done with involvement, participation from all parties in civil society through some sort of a 'social compact"): * PAMS can shift the composition of output growth toward the agricultural sector; * PAMS can increase the (average) income tax rates of the wealthier groups in the economy (e.g., the capitalists-rentiers and the urban civil servants); * PAMS can increase the room for maneuver for social spending by reducing military expenditures and allocating the resulting savings to transfers to the poorest groups; Page- 64 PAMS Annex 6 Table A6-7: .......____________________________ Hence, we construct !A~ a-=4Wi Z -° 3 ai1 ~B. fl~S- m°a ZI Zn37 XIF 9 :X, ~ an altemative LjIU,NIAaR j , , 1 4,a,,ii7 4,aDi4 ,,, 4Q-ml 4aEs 4QX7 ,4ai42 ,4,aZ, scenario along f 41 . ' -am7 4a1i4 ,4,,a= -om 4aiB, ,4am7,_ __ -d these lines. Within R&SMkr ! ; - .the same macro envelope of output growth, the Government ___-P4* X - promotes output f ,, Qa uam Q-g 4growth in the rural -a fM0l__ -9' ,QE areas (e.g., for ~~~LLAB~~~~~~~~~~1R ___ ~~~~~~~~~~~example through ziuLL^snR , ' . aa2.,G, aash QCh*Qlii; Qf4SQrXDX a2X4' ,,a,2s0i special programs of LEo { , ' Qa11 ,Qa ( Qfa,2,aiQaz a"f arm 2 I am incentives for __nRL GSD6 i0~ aaBi aosz~ aOu an r7~ _O a farmers, etc.). There is -as a consequence -more employment in rural areas but given PAMS' characteristics, i.e. by construction is maintains overall macro-consistency, there is simultaneously less employment in urban areas. Overall, unemployment fall in rural areas by about 1 to 4% and (new) job creation in rural areas rises (see Table A6-7) from 25,000 in 2004 up to 250,000 in 2011. Table A6-8: TAXES ON INCONE (Nlion of lCt_) In parallel, the Government decides to make its average income tax rates URBAN more progressive. Urban rentiers, civil servants and some workers in the non- 10% tradable urban sector get heavier tax ;Public Sector P0% rates. They get an additional surcharge of 10% or 0% over their TrlMde Goods 0 existing average tax rates that can be ..i . Bseen in Table A6-7). In parallel, rural N e0% workers get tax rebates (-5%) bringing Skilled " L ~255i2 their tax rates to zero. Overall, these u!nsi*4 , s|(9iL transfers in terms of income taxes are lpfornal(unsku e . so (P.1.0 done within the same tax envelope i provided by the baseline scenario. The 'RliRAi -L% workers' tax brackets end up being 'Fonmi~(e,qort)unakiLled ' 51)00 nt,.t respectively 35% for civil servants; Unskilled 519, 25% for workers in the urban tradable goods sector and skilled workers in the non-tradable goods sector; 20% for unskilled workers in the non-tradable goods sector; 15% for the unskilled workers in the urban informal sector, and 0% for workers in the rural areas. Table A6-9: . coMPosrnoN oF iuBfLcS PunDiNiG - ' Finally, there are some policy changes in the budget that affect linvenerentOfToaPixend) i social transfers. The Govemment -SiM 0.00% manages to reduce military Military(PercentofOtherEvenditures) 35.00%f expenditures to 35% (down from ,Soc,al(lestdua !,PercentofOther!lpnditures) tiS%65-0 50%) of non-wage, non-capital, a) Education (Percent of Social) '3.133 - E Pnmaryrd ~ IocoJ00 other expenditures. The available !Scond. Ed. 40 0ir4 savings, are allocated to a 'social Tertiary Ed, l0h)0 fund" that provides income transfers b) Healtb (PercentofSocil) . of c toithep c) Incgne TaTnsfers (Residual, Percent of Social) 3p 3or Page- 65 PAMS Annex 6 PAMS simulates the new scenario, which keeps the same basic macro framework that was simulated with the baseline. What can we expect? We have not changed significantly the macro story of the base case: we have the same "baseline" aggregate growth outcome, the same aggregate fiscal deficit and the same aggregate external balance (BoP). However, we have now made "policy changes" that brought a different composition of social spending, a more progressive income tax burden and more incentives to agricultural-led growth. Given the characteristics of Burkina (i.e. the poor are in the rural areas), we should expect a reduction in poverty and -hopefully-also a reduction in inequality due to the transfers that are made from the rich to the poor. Figure A6-2 below shows the deviations of the results of the new simulation vis- a-vis the baseline case. Figure A6-2: The new policies that we BURKINA FASO: Differences ('/I with Base Scenario listed above do indeed contribute to an increase in 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 15.00%. --...-. . the aggregate wage income (recall that we are 0.0o% i transferring resources from an overall unchanged 5.00% aggregate disposable income, and hence that the 0.00% "losers" are the "capitalists" -5.00% and "rentiers"). More 40.00% \importantly, there is also an .10.00% _-additional reduction in the Moan Income overall poverty headcount. -15.00% -In(HeaCoe t Gap ex) That comes from the set of -at-Gini transfers that targeted the -20.00% __- -- poorest groups of the rural areas (through lower income taxes and social benefits). Finally, in addition, there is also a reduction in the level of inequality, vis-A-vis the baseline case. Therefore, the overall conclusion is that the macro framework associated with Burkina Faso's PRSC does allow -with the proper set of additional fiscal, tax and transfer policies-to make a stronger attack on poverty levels. Unfortunately, the story is not that simple. Let us look now at the decomposition of poverty and inequality between rural and urban areas, as depicted in Figure A6-3. Figure A6-3: Although the new simulation BURKINA FASO: Differences (Y/4 with Base Scenario has managed to reduce (Urban plain lines, Rural - dotted lines) inequality "on aggregate", there is a significant 25v00 - -_ difference between the 25.00% - __.... _ ___ .______..... _ __ situations in the rural (plain 20.00% lines in Figure A6-3) and 15.00% ^9 -Meanincome urban areas (dotted lines in 10.00% 40- . . . Meanincome Figure A6-3). Recall that 5,00% --P(HaCO the policies envisaged 0.00% *, l- , . , . . . _PO(Hex) "favored" agricultural growth -00-Gtni and redistribution to the 00 0- P0 (Head Count rural poor. The counterpart .10.00% Index) of that is to reduce the size .15.00% of the urban economy vis-a- -20.00% I vis the base case and -25.00% --. . i hence to "increase" the level Page- 66 PAMS Annex 6 of urban poverty vis-a-vis the base case. This is due to the sensitivity of the urban informal sector to any slowdown and the consequent increase in urban unemployment. There is also a decline (always vis-a-vis the base) of urban wage income that increases urban poverty. The disaggregated result is thus mixed. Yes, on average, poverty went down further and so did inequality. But in fact rural poverty went down dramatically while urban poverty increased. And while urban inequality went down (an effect of the transfers that affected relatively more the rich urban households) there is a relative increase of inequality in rural areas, due to the combination of tax breaks and transfers affecting the two types of poor rural households. As a conclusion, PAMS shows that the intuitive results that one can get from cross-section regressions, where aggregate growth 'by definition reduces poverty', -by the sign of the estimated elasticity of growth to PO--) can sometimes hide different situations in terms of where (rural/urban) is poverty reduced, and whether it is done with an increase or a decrease of inequality. Although the simulations shown here are crude and tentative, they illustrate how the model is capable of initiating a richer policy dialogue and contribute to the design of better policies. Page- 67 Policy Research Working Paper Series Contact Title Author Date for paper W'PS2868 Universal(ly Bad) Service: George R. G. Clarke July 2002 P. Sintim-Aboagye Providing Infrastructure Services Scott J. Wallsten 38526 to Rural and Poor Urban Consumers W'PS2869 Stabilizing Intergovernmental Christian Y. Gonzalez July 2002 B. Mekuria Transfers in Latin America: David Rosenblatt 82756 A Complement to National/ Steven B. Webb Subnational Fiscal Rules? W'PS2870 Electronic Security: Risk Mitigation Thomas Glaessner July 2002 E. Mekhova In Financial Transactions-Public Tom Kellermann 85984 Policy Issues Valerie McNevin WPS2871 Pricing of Deposit Insurance Luc Laeven July 2002 R. Vo 33722 W'PS2872 Regional Cooperation, and the Role Maurice Schiff July 2002 P. Flewitt of International Organizations and L. Alan Winters 32724 Regional Integration W'PS2873 A Little Engine that Could ... Liesbet Steer August 2002 H. Sutrisna Domestic Private Companies and Markus Taussig 88032 Vietnam's Pressing Need for Wage Employment VWPS2874 The Risks and Macroeconomic David A. Robalino August 2002 C. Fall Impact of HIV/AIDS in the Middle Carol Jenkins 30632 East and North Africa: Why Karim El Maroufi Waiting to Intervene Can Be Costly V\'PS2875 Does Libert6=Egalit6? A Survey Mark Gradstein August 2002 P. Sader of the Empirical Links between Branko Milanovic 33902 Democracy and Inequality with Some Evidence on the Transition Economies W'PS2876 Can We Discern the Effect of Branko Milanovic August 2002 P. Sader Globalization on Income Distribution? 33902 Evidience from Household Budget Surveys W'PS2877 Patterns of Industrial Development Raymond Fisman August 2002 K. Labrie Revisited: The Role of Finance Inessa Love 31001 WPS2878 On the Governance of Public Gregorio Impavido August 2002 P. Braxton Pension Fund Management 32720 WPS2879 Externalities in Rural Development: Martin Ravallion August 2002 C. Cunanan Evidence for China 32301 W'PS2880 The Hidden Costs of Ethnic Soumya Alva August 2002 T. Bebli Conflict: Decomposing Trends in Edmundo Murrugarra 39690 Educational Outcomes of Young Pierella Paci Kosovars WPS2881 Returns to Investment in Education: George Psacharopoulos September 2002 N. Vergara A Further Update Harry Anthony Patrinos 30432 Policy Research Working Paper Series Contact Title Author Date for paper WPS2882 Politically Optimal Tariffs: Dorsati Madani September 2002 P. Flewitt An Application to Egypt Marcelo Olarreaga 32724 WPS2883 Assessing the Distributional Impact B. Essama-Nssah September 2002 0. Kootzemew of Public Policy 35075 WPS2884 Privatization and Labor Force Alberto Chong September 2002 H. Sladovich Restructuring around the World Florencio Lopez-de-Silanes 37698 WPS2885 Poverty, AIDS, and Children's Martha Ainsworth September 2002 H. Sladovich Schooling: A Targeting Dilemma Deon Filmer 37698 WPS2886 Examining the Feasibility of Jerry R. Skees September 2002 E. Laguidao Livestock Insurance in Mongolia Ayurzana Enkh-Amgalan 82450 WPS2886 Examining the Feasibility of Jerry R. Skees September 2002 E. Laguidao Livestock Insurance in Mongolia Ayurzana Enkh-Amgalan 82450 WPS2887 The Demand for Commodity Alexander Sarris September 2002 M. Fernandez Insurance by Developing Country 33766 Agricultural Producers: Theory and an Application to Cocoa in Ghana