2009年-世界发展银行全球_Pakistans_Investment_Climate___Laying_the_Foundation_for_Renewed_Growth_Volume_3_Background_Paper_on_Econometric_Methods_102页_1mb

Summary of Document: Econometric Methods for Investment Climate Assessment on Economic Performance in Pakistan

Core Content

This document presents an econometric analysis of the investment climate (IC) effects on economic performance in Pakistan, using firm-level data from the 2002 and 2007 Investment Climate Surveys (ICSs) in both the manufacturing and services sectors. The goal is to identify the impact of IC variables on productivity, employment, wages, export propensity, and FDI receipt, while addressing issues such as endogeneity, missing data, and measurement errors.

The analysis is structured into two main parts: identifying IC effects on economic performance and evaluating the contributions of these effects to productivity and other performance indicators. The methodology is based on robust econometric techniques developed by Escribano and colleagues, including the use of Olley and Pakes (1996) decomposition of productivity and instrumental variable (IV) estimation to address endogeneity concerns.

Main Points

• Investment Climate Importance: The investment climate significantly affects economic performance, particularly in emerging economies like Pakistan. Poor IC conditions have contributed to the decline in GDP per capita and labor productivity compared to East Asia.

• Data Sources: The study uses three IC datasets:

  • 2007 manufacturing firms (784 establishments)
  • 2007 services firms (151 establishments)
  • FY02-FY07 manufacturing panel (402 firms)

• Econometric Methodology:

  • A structural system of equations is used, with IC variables on the right-hand side and economic performance measures (productivity, employment, wages, export propensity, FDI receipt) on the left-hand side.
  • Robust productivity elasticities are estimated using single-step and two-step methods, and the analysis accounts for endogeneity and simultaneous effects.
  • Olley and Pakes decomposition is employed to break down aggregate productivity into efficiency, technical change, and allocative efficiency components.
  • IV estimation is used to address endogeneity in IC variables and production function variables.
  • Missing data mechanisms are carefully considered, with the assumption that data is missing at random (MAR), allowing for imputation and consistent estimation.

• Key Economic Performance Measures:

  • Productivity (TFP for manufacturing, labor productivity for services)
  • Employment and wages
  • Probability of exporting
  • Probability of receiving FDI

• Data Cleaning and Handling:

  • Missing data is a major issue in the ICSs, with response rates ranging from 5% to 100% across different IC variables.
  • Missing values are handled through imputation methods and replacement procedures.
  • The representativeness of the sample is affected by missing data, especially in the Baluchistan and NWFP regions.

• Results:

  • Infrastructure, access to finance, and informality are significantly associated with productivity differences.
  • Larger firms, in terms of market share, are better able to cope with IC bottlenecks and benefit more from positive IC aspects.
  • The FY02-FY07 panel is useful for analyzing the evolution of IC constraints and their effects on firms over time, despite limitations in representativeness.
  • IC contributions are evaluated in terms of both the Olley and Pakes decomposition and sample means of economic performance indicators.

• Methodological Considerations:

  • The study accounts for endogeneity and simultaneity in the variables.
  • Cluster standard errors are used to address correlation within industries and regions.
  • The Heckman selection model is suggested for checking sensitivity to non-ignorable missing data assumptions.

Key Information

• IC Variables Classified into 7 Blocks:

  1. Infrastructure
  2. Economic governance
  3. Finance
  4. Innovation and competition
  5. Labor markets and skills
  6. Corporate governance
  7. Other control variables

• Data Challenges:

  • Missing data is frequent and affects the representativeness and efficiency of the sample.
  • Missing values are more common in capital-related variables, particularly in the services sector.
  • Outliers are also identified and excluded from analysis to ensure robustness.

• Econometric Models:

  • OLS and random effects models are used for cross-sectional and panel data.
  • IV estimation is applied to address endogeneity in both production function and IC variables.

• Impact of IC on Economic Performance:

  • IC has a measurable effect on productivity, employment, and wages.
  • Firms with better IC conditions are more likely to export and receive FDI.
  • The Olley and Pakes decomposition highlights the role of IC in efficiency and allocative components of productivity.

• Simulation and Comparison:

  • Simulations based on IC improvements are conducted to estimate potential productivity gains.
  • International comparisons are made using demeaned productivity measures to assess IC performance across countries.

Structure of the Analysis

• Volume III of the report focuses on the econometric methodology.
• Section 2 describes the data sources and cleaning procedures.
• Section 3 details the estimation of IC elasticities and semi-elasticities on productivity.
• Section 4 explores the impact of IC on employment, wages, export propensity, and FDI receipt.
• Section 5 discusses the evaluation of IC contributions to productivity decomposition and other economic performance measures.
• Section 6 outlines the econometric approach for the services sector.
• Section 7 describes the methodology for the FY02-FY07 panel.
• Section 8 and 9 present the results and evaluations of IC effects.
• Section 10 concludes the report.

Conclusion

The econometric analysis confirms the significant role of the investment climate in shaping economic performance in Pakistan. By addressing endogeneity and missing data issues, the study provides robust estimates of IC effects on productivity, employment, and other performance indicators. These findings are valuable for understanding the critical IC bottlenecks affecting Pakistan’s economic growth and convergence with other regions.