Jan Walliser

1
Poverty Analysis Macroeconomic Simulator (PAMS)
and PSIA with an application to Burkina Faso
Jan Walliser Senior Economist The World Bank
2
Outline of the Presentation

• Introduction
• PAMS Inputs and Outputs
• A brief tour of PAMS
• A Set of Policy Experiments

3
Introduction why macroPSIA?

• Changes in the macro framework such as the
  fiscal, inflation and exchange rate targets? How
  do they affect the poor?
• Exogenous shocks such as trade shocks, capital
  flows volatility, changes in foreign aid and
  foreign payment crises? How can policy mitigate
  these effects on the poor?

4
Introduction why macro PSIA?

• Improving public expenditure targeting? How can
  public expenditure be better targeted?
• Structural reforms such as trade policy,
  privatization, agricultural liberalization? How
  are the poor affected?

5
Modeling Implications and Challenges

• Maintain simplicity of macroeconomic consistency
  frameworks (e.g., RMSM-Xs or other country-based
  models)
• Link macro-consistency frameworks directly with
  household survey data

6
The Logic of PAMS

• Three Recursive Layers Consistent with Incidence
  Approach
• Macro-framework GDP, national accounts, taxes
  government spending, BOP, prices
• Labor model breaking down population by skill
  level and economic sectors using categories from
  HHS
• Model to simulate income changes by group,
  allowing calculation of poverty incidence and
  inter-group inequality

7
Top-down HHL "micro-simulation" approach
General Structure 3 Layers
Macroeconomic Model Macro Accounting (RMSM-X),
CGE (123), Econometric
Layer 1 Macro
Sectoral Disaggregation, Factor Markets ? Linkage
Aggregate Var For k representative groups of
households
Layer 2 Meso
Household Survey (HHS), i individual households,
Macro "consistent" changes in real household
incomes and change in the distribution of welfare
Layer 3 Micro
(yi) with poverty line, z, ? indicator of poverty
Pi for each household i and indicators of
within-group inequality (e.g., Gini, etc.)
8
Limitations

• Not all policy challenges covered
• PAMS best suited to simulate poverty and
  distributional implications of
• PRSP-PRGF macro baseline scenarios
• Sensitivity analysis along the base case
• Sectoral growth scenarios
• Average tax burden (standard incidence analysis)
• Average social transfer

9
PAMS Inputs and Outputs

• Micro input
• Macro input
• Micro-Macro Linkage

10
PAMS Micro Input

• Household Survey Data
• Expenditure or income
• Size of household
• Household weight in population
• Data arranged by socioeconomic groups of
  representative households

11
PAMS Macro Input

• Macro framework from any macro consistent model
  (IMF macro projections, World Banks RMSM-X
  model, other domestic macro models)
• ?Aggregate variables (GDP, BOP, fiscal accounts,
  monetary accounts, inflation)

12
PAMS Micro-Macro Linkages

• Labor market module breaks down the economy into
  sectors rural/urban, formal/informal,
  tradable/non-tradable
• Labor supply is driven by exogenous factors
• Labor demand is demand is broken down by sector,
  skill level and location and depends on sector
  demand and real wages
• Labor model produces wage income by
  representative households of SEG and location
  based on income aggregates, group-specific tax
  and transfer variables

13
PAMS Micro-Macro Dynamics

• Base year as starting point
• Simulation of macro variables/population
• Simulation labor demand and supply, wages and
  incomes by groups
• Simulation of changes in HH-level income data to
  calculate poverty indicators assuming unchanged
  intra-group distribution of incomes

14
PAMS Outputs

• 1. Standard macroeconomic Indicators
• 2. Standard poverty and inequality indicators
  (P0, P1, P2, Gini, etc.)
• 3. Poverty decompositions Growth, inequality and
  population effects with respect to P1 and P2

15
PAMS Outputs

• 4. Pro-poor growth indicators
• Pro-poor growth index (Kakwani and Pernia, 2000)
• Growth Incidence Curve (Ravallion and Chen, 2003)
• Poverty Equivalent Growth Rate (Kakwani and Son,
  2003)

16
PAMS
17
Simulation with PAMS
Update Macro
Update Earning Trans. Module
Pov. Ineq Simul. Scen.
Household survey
Pov. Ineq Baseline Scen.
Iteration Process
18
Country Applications
19
PAMS Burkina Faso

• 1994, 1998, 2003 HHS
• Longstanding macroeconomic Program with IMF
• HIPC CP in 2000 (original) and enhanced (2002,
  with topping up)
• Growth rates averaging 5 percent
• Largely rural population

20
PAMS Burkina Faso

• Poverty rates (1998) of 45 percent based on
  national poverty line (which is below 1/day)
• Cotton as major cash crop 50-60 percent of
  exports, and significant growth of cotton
  production
• Cereal production stabilized due to promotion of
  small-scale irrigation

21
PAMS development

• Work started before 2003 HHS in context of PRSP
• Interest in having better handle on poverty
  projections using macro-growth projections
• Home-grown excel-based macro-model (IAP) with
  technical assistance of GTZ
• Collaboration on PAMS based on 2003 HHS
• PAMS model linked to IAP output tables

22
PAMS development

• PAMS model linked to IAP output tables with
  support from local GTZ adviser and team
• Close collaboration with macro forecasting
  division in Ministry of Economy and Development
• (Political) challenge integration of 2003 HHS
  because of weaknesses in data analysis

23
SEGs and Poverty, 1998-2003
24
Macro baseline scenario
25
Poverty baseline scenario
26
Inequality-growth tradeoff
27
20-percent decline in cotton prices
28
20 percent decline in cotton volume and cotton
price
29
Increased primary sector contribution to growth
30
Lessons learned

• Strong payoffs of building a close early
  collaboration with the government forecasting
  team
• Close collaboration with the local GTZ technical
  assistance crucial
• Close involvement of World Bank country office
  staff essential
• Need to make greater allowance for the collection
  and analysis of poverty data when embarking on
  PAMS modeling