DEC Course on Poverty and Inequality Analysis Module

DEC Course on Poverty and Inequality Analysis Module 7: Evaluating the Distributional and Poverty Impacts of Economy-wide Policies Session VII: Simulating the Distributional Impacts of the 1999 devaluation of the Brazilian Real. Francisco H. G. Ferreira

Can the distributional impacts of macroeconomic shocks be predicted? : A comparison of top-down macro-micro models with historical data for Brazil Francisco H. G. Ferreira, Philippe G. Leite, Luiz A. Pereira da Silva (World Bank) and Paulo Picchetti (Universidade de São Paulo) Chapter 5 in Bourguignon, Bussolo and Pereira da Silva (eds. ) 2008. The Impact of Macroeconomic Policies on Poverty and Income Distribution (Washington, DC: Palgrave Macmillan and the World Bank)

The Brazilian Financial Crisis of 1999. The “float” of the currency on January 15, 1999 average annual parity with the USD goes up from R$1. 161 (annual 1998 average) to one USD to R$1. 816 (annual 1999 average corresponding to a 56. 4% devaluation) A temporary rise the central bank policy rate (BACEN’s Selic) during the period corresponding to October 1998 till May 1999. Monthly rate raised from 1. 47% in August 1998 up to 3. 33% in March 1999 (corresponding to annualized rates of almost 50%). SBA arrangement with the IMF credibility of the policy framework tightening in the fiscal stance corresponding to a reduction of the Consolidated Public Sector Borrowing requirements (PSBR) from 7. 5% of GDP down to 5. 8% of GDP, i. e. a cut of 1. 7% of GDP).

Objective: impact evaluation of a program or policy Define impact for individual i or household as the difference in income with and without the program (policy): yi wi Li Ei Ri Ai Ci p : real income : wage rate : labor supply : self-employment, non-wage income : net transfer income : socio-economic characteristics : consumption characts. household-specific P price index : general price index

DC Dp prices Dw wage DL employt. DR transfers DA households character. - D filter Household Survey (HHS), i individual households Compare the distribution of y|P=1 with the distribution of y|P=0. Calculate changes in inequality or poverty across the two distributions. Partial equilibrium independent shocks

Evaluation of macro economic policies. Macro to micro linkages Macro framework, general/partial equilibrium Dp prices Dw DL DR wage employt. transfers DA households character. Instead of « exogenous and independent » shocks use « endogenous and dependent» shocks to 'microsimulate' the effect of policies on all individuals in the micro data sets, and the poor some consistency constraints will be « binding » (e. g. , budget envelope for social programs, real GDP growth, etc. ) LAVs Household Survey (HHS), i individual households

Top-down macro-micro-simulation approach “Top” Level : Macro General Equilibrium Macroeconomic Model With Sectoral Disaggregation to model Factor Markets Linkage Aggregate Variables ( LAVs) “Bottom” Level : Micro Individual /Household occupational choice model Household income determination model Household occupational choice model Individual/Household income determination model

Novel Aspects of the Linkage Exercise Test top-down linkage with macro-econometric model on top (not CGE, confidence intervals for parameters) and micro-simulation at bottom Changes in LAVs respond to “known” periodicity (e. g. , annual) at the top (not “convergence process” of CGE) Endogeneize in macro model key features of emerging markets: Structural features : e. g. , increase of “informality” in labor market; usage of substitutable semi-skilled labor Shocks: change in ERR portfolio choices by banks and holders of domestic debt financial crisis If historical simulation (H) is robust , counterfactual (C) is possible as alternative macro-responses with different outcomes for poverty and distribution (compare program/policy (P) with counterf. (C)).

Brazil – Top Layer = Macroeconometric Model (Castro, Pereira da Silva and Picchetti [2003]) • • Conventional IS-LM macroeconometric model with a disaggregated labor market and financial sector, estimated with 1981 -2001 annual data or more (NA and historical HHS) Real economy: 6 sectors: Urban/Rural, Tradable/Non-Tradable, formal/Informal Labor market: 3 skills, skilled, semi-skilled and unskilled, mobile across skills; supply and demand modeled by sector and skill endogenous unemployment, supply side – sector specific production functions with threelevel nested CES. Fernandes and Menezes-filho (2001) substitution between capital and labor and all kinds of labor, except between high-skill and low-skill labor. financial sector, see Bourguignon, Branson and de Melo [1989]: 8 assets: currency, deposits, bonds, dom. loans (debt) and equity shares; forex currency, forex loans to residents and forex bonds. 6 agents: households, private firms, commercial banks, Government, central bank and foreigners. Bottom Layer - Micro-simulation model (Ferreira and Leite)

An overview of the macro model

Historical Simulation 96 -2001 Macro Variables

Sectoral Production Functions Composite Labor: Qualified and Unqualified Jobs Qualified Jobs: High and Intermediate Skill Workers Unqualified Jobs: Intermediate and Low Skill Workers

Brazil, Financial Crisis Scenario – What we do: Simulate 1999 financial Crisis with Macroeconometric Model 48 LAVs Run the micro-simulation model ER shock and policy rate change (1999) Run historical simulation with macroeconometric model generate 48 LAVs to feed microsimulation model Depart from 1998 HHS (PNAD), use LAVs generated by macro model to simulate 1999, converge micro simulations to match macro generated LAVs Compare results of combined micro-macro simulation with actual 1999 data from HHS (PNAD)

Types of simulation experiments undertaken

The Household-Level Data and the Micro-econometric model Data Set: Pesquisa Nacional por Amostra de Domicílios (PNAD) 1998 & 1999 Main variables earnings occupation total household income per capita Insufficient detail on capital incomes, production for own consumption and incomes in kind

The Household-Level Data and the Micro-econometric model The model consists of three equations: Occupational Choice

The Household-Level Data and the Micro-econometric model Earnings equation Household income aggregation

Recall: Structure of the micro-macro model “Top” Level : Macro General Equilibrium Macroeconomic Model With Sectoral Disaggregation to model Factor Markets Linkage Aggregate Variables ( LAVs) “Bottom” Level : Micro Individual /Household occupational choice model Household income determination model Household occupational choice model Individual/Household income determination model

Recall: The LAV structure (One for Urban; one for Rural) Sectors Formal Tradable Formal Informal Non-Tradable Unemployed HH Low Skill f w f Groups Int. Skill f w f High Skill f w f

Employment: Actual and Simulated

Wages: Actual and Simulated

Micro-simulations Solution of system of 42 equations ( ) ˆ å å Exp a gs + x ih b g + e ih = p g w g Wg. . "g iÎg sÎg

Simulation Solve the system of 42 equations changing all constant ( 0 and ) terms. Calibrated so that micro-simulation reproduces changes in aggregate structure of employment obtained in macro-economic framework. Newton-Rapshon algorithm. Minimize the sum of squared differences between the left- and the right-hand side of equations.

Results: Earnings (I)

Results: Earnings (II)

Results: Earnings (III)

Results: Aggregate Poverty and Inequality Indices

Winners and Losers

Conclusions: Occupations The macro-micro model captures a great deal of the occupational effect of the 1999 crisis on the occupational structure in Brazil. a significant increase (+12. 8% / +12. 7%) in unemployment in both rural and urban areas a rise in unemployment particularly large for workers with intermediate and high skill levels in urban areas (+14. 6% / +14. 4 and +18. 1% / +17. 3% respectively) a decline in the employment of urban workers with low skills (-1. 8% / -0. 3%) an increase in the level of informality in both rural and urban areas (+1. 1% / +0. 1% and +3. 5% / 4. 0% respectively) a growth of informality in particular in urban areas for workers with intermediate and high levels of skills (+6. 5% / +3. 1% and +5. 3% / +10. 4% respectively)

Conclusions: Earnings The model underestimates slightly changes in earnings for all but one category of workers (i. e. the workers with intermediate level of skills in the formal tradable sector) Overall, the macro-micro model captures also a great deal of the actually observed changes in (nominal) earnings in Brazil from 1998 to 1999. Mean earnings fell for all three urban categories of workers; by – 1. 12% (+0. 32%) for workers with low skill level; by – 2. 84% (-4. 87%) for workers with intermediate skill level; by – 4. 31% (-6. 06%) for workers with high skill level; The picture is more mixed in rural areas. There, the only winners among low-skilled workers were those employed in the formal non-tradable and the informal sectors (and this is well predicted by the model). The main losers (-4. 04%) among intermediate and high skilled workers were those in the formal tradable sector (and this is predicted by the model, 7. 78%). And the main winners (+12. 07%) among intermediate and high skilled workers were those in the formal tradable sector (and this is overpredicted by the model, 29. 33%).

Conclusions Occupations: predictive performance of the macromicro model is relatively good Earnings: less satisfactory. Under-prediction of declines in wages. May be due to an insufficient disaggregation of the wage LAVs across occupations, or to the functional form of factor remuneration –negatively affected by lower economic activity and rise in unemployment-- in the macroeconomic modeling. The end result in terms of the counterfactual income distribution for Brazil in 1999: compensating errors, leading to a relatively good prediction of the poverty and inequality levels RHGs worse than macro-micro approach