2012年-IMF国际货币组织全球_Exogenous_Shocks_and_Growth_Crises_in_Low_41页_1mb

Summary of "Exogenous Shocks and Growth Crises in Low-Income Countries: A Vulnerability Index"

Core Content

This paper introduces a vulnerability index designed to predict the likelihood of growth crises in low-income countries (LICs) following large external shocks. The index is constructed using a combination of multivariate regression analysis and a univariate signaling approach, drawing on policy, structural, and institutional indicators to assess the susceptibility of LICs to growth downturns.

The paper emphasizes that exogenous shocks—such as terms-of-trade swings, export demand fluctuations, FDI changes, aid, remittances, and natural disasters—have a more pronounced and frequent impact on LICs compared to advanced and emerging market economies. While neoclassical growth models suggest that negative shocks would lead to a quick return to the steady state, LICs often lack the policy buffers and institutional capacity to absorb such shocks, leading to persistent output and welfare losses.

Main Viewpoints

• Vulnerability Index: A composite index that integrates various indicators to assess the likelihood of a growth crisis in the context of exogenous shocks.
• Two Approaches:
  • Multivariate Regression: Uses a correlated panel probit model to estimate the probability of a growth crisis based on a set of policy and structural variables.
  • Univariate Signaling Approach: Identifies critical thresholds for each indicator that signal the onset of a growth crisis, and combines them into a weighted summary index.
• Growth Crisis Definition: A crisis is defined as a sharp decline in real GDP per capita following a shock, with the post-shock two-year average falling below the pre-shock three-year trend and negative growth in the first year.
• Key Determinants:
  • Institutional quality (measured by CPIA)
  • Exchange rate regime flexibility
  • Reserve coverage and fiscal balance
  • Past GDP growth performance
  • Balance of payments pressures
• Empirical Findings:
  • The overall vulnerability index has declined since the early 1990s, but has risen again in recent years due to exhausted fiscal buffers.
  • Institutional quality and exchange rate flexibility are the strongest predictors of growth crises.
  • Reserve coverage and fiscal balance have a significant impact on reducing crisis probability.
  • Contemporaneous shock size is a key determinant, especially for non-commodity exporters and non-small island economies.

Key Information

• Time Period: 1990–2009 for 71 low-income countries.
• Shock Definition: A shock is identified if the annual percentage change of a relevant variable falls below the 10th percentile of the country-specific distribution.
• Threshold Probability: Derived using a loss function minimization approach, with a threshold of 19% to classify a growth crisis.
• Model Performance:
  • The median predicted probability of a growth crisis is 38%, while for normal episodes it is 10%.
  • Out-of-sample predictions show that the model correctly identifies 14 out of 15 crises and has lower misclassification errors than in-sample predictions.
  • Robustness Check: The model performs well even when including all growth crisis events, not just those from large external shocks.

Methodology Highlights

• Shock Identification: Based on country-specific distributions, with six types of shocks considered: external demand, terms-of-trade, FDI, aid, remittances, and climatic shocks.
• Dependent Variable: A growth crisis is defined as a sharp decline in real GDP per capita after a shock, with negative growth in the first year.
• Signaling Approach:
  • Each indicator has a threshold that differentiates between crisis and non-crisis events.
  • The overall vulnerability index is calculated by summing weighted indicators based on their signaling power.
  • Optimal Thresholds: Determined by minimizing total misclassification errors (TME) or maximizing the signal-to-noise ratio (SNR).

Conclusion

The paper concludes that the vulnerability index is a useful tool for early warning of growth crises in low-income countries. While the index has improved over time, recent years have seen increased vulnerability due to the exhaustion of fiscal buffers following the global financial crisis. The index highlights the importance of institutional strength, exchange rate flexibility, and fiscal policy in mitigating the effects of external shocks. It also underscores the long-term impact of shocks on output and welfare, especially in countries with weak policy fundamentals.

Key Tables and Figures

• Table 1: Median real GDP per capita growth for shock and non-shock samples.
• Table 2: Benchmark probit regression results for all countries and subgroups.
• Table 3: Marginal effects of explanatory variables on crisis probability.
• Table 4: Impact of changes in variables on crisis probability.
• Table 5: Predicted probabilities and error rates.
• Table 6: Composition of the vulnerability index and performance of individual indicators.
• Figure 1: Identification of external shock episodes.
• Figure 2: Distribution of real GDP per capita for crisis versus normal episodes.
• Figure 3: Relationship between covariates and threshold crisis probability.

References

• Collier and Goderis (2007, 2009)
• Berg et al. (2010, 2011)
• Rodrik (1999), Pritchett (2000), Hausmann et al. (2006)
• Dabla-Norris et al. (2011)
• Demirguc-Kunt and Detragiache (1999)
• IMF (2007, 2010, 2011)
• Easterly et al. (2000)

Appendix

• Appendix Table 1: Correlation matrix for exogenous shocks.
• Appendix Table 2: List of countries in the sample.
• Appendix Table 3: Variables used in probit regressions and signaling approach.
• Appendix Table 4: Distribution of covariates.
• Appendix Table 5: Robustness check results.