2004年-世界发展银行全球_The_Distribution_of_Income_Shocks_during_Crises___An_Application_of_Quantile_Analysis_to_Mexico_1992-95_20页_341kb

Summary of "The Distribution of Income Shocks during Crises: An Application of Quantile Analysis to Mexico, 1992-95"

Core Content

This article examines the distribution of income shocks among different types of households in Mexico during the 1994-95 economic crisis, using quantile analysis to move beyond traditional average-based measures. The authors argue that understanding the full distribution of income shocks is crucial for assessing household vulnerability and designing effective safety nets.

Main Points

• Economic Context: Mexico's 1994-95 crisis, known as the "Tequila crisis," led to a 30% average decline in household incomes. This crisis was not unique, as similar downturns occurred in other Latin American and Asian countries.

• Methodology: The study uses a quantile regression approach to analyze the distribution of income shocks across various household types. This allows for a more nuanced understanding of how different groups are affected at different parts of the distribution, not just the average.

• Quantile Regression: Unlike ordinary least squares (OLS), which focuses on the mean, quantile regression estimates the effect of variables at different percentiles (e.g., 20th, 50th, 80th). This method is more robust to outliers and provides a clearer picture of the distributional impacts of the crisis.

• Data: The analysis is based on the Mexican National Urban Employment Survey (ENEU), a rotating panel dataset covering 1992-96. It includes detailed information on household income, labor market participation, and demographic characteristics.

• Household Categories: The study considers various household types, including:

  • Education level of the head (incomplete primary, complete primary, incomplete secondary, complete secondary)
  • Age (young, old)
  • Number of children (more than 1.3)
  • Household structure (single mothers, single women, single men)
  • Sector of employment (informal self-employed, informal salaried, no remuneration)

• Key Findings:

  • The distribution of income shocks across household types remained largely unchanged during the crisis compared to normal times.
  • Households headed by less educated individuals, single mothers, or informal sector workers did not experience disproportionately large income drops.
  • Outliers and extreme values significantly affect OLS estimates but are less influential in quantile regression.
  • The median regression provides a more accurate and robust estimate of income changes, especially when dealing with skewed or non-normal distributions.

Critical Insights

• Vulnerability and Distribution: Vulnerability to income shocks is not just about the current income level but also about the variability of income over time. Quantile analysis helps uncover this variability and its implications for policy design.

• Policy Implications: Traditional measures of income change may obscure important differences in how various groups are affected. By analyzing the entire distribution, policymakers can better target support to the most vulnerable households.

• Methodological Advantages:

  • Quantile regression is more robust to outliers and non-normality in the data.
  • It allows for the detailed exploration of the distribution of shocks, which is essential for understanding inequality and risk.

Key Information

• The base group consists of households headed by married, middle-aged, college-educated males in the formal sector with fewer than the mean number of children.
• The log difference specification is used to estimate percentage changes in income, though it is only a close approximation for small changes.
• The corrected coefficients are calculated using the formula:
  $$
  \hat{\delta}_j^* = \exp(\hat{\delta}_j - [1/2] \hat{V}[\hat{\delta}_j]) - 1
  $$
• The number of bootstraps is determined using the algorithm by Davidson and Mackinnon (2000).
• The F-test is used to compare the significance of coefficients across different quantiles.
• Outliers are identified and handled using bootstrapping rather than arbitrary trimming.

Conclusion

The study concludes that while the average income decline was significant, the distribution of income shocks did not change substantially across household types during the crisis. This suggests that the impact of the crisis was relatively uniform, and that pre-existing characteristics of households may not be the primary determinant of vulnerability. The authors advocate for the use of quantile analysis in future studies to better capture the full range of income shocks and their distributional effects.