2012年-世界发展银行全球_When_Should_We_Worry_about_Inflation__28页_508kb

Summary of "When Should We Worry about Inflation?"

Core Content

This article investigates the relationship between inflation and economic growth, focusing on identifying the threshold level of inflation beyond which growth is negatively impacted. The study uses a logistic smooth transition regression (LSTR) model to analyze data from 165 countries between 1960 and 2007, building on previous research that suggested a nonlinear relationship between inflation and growth.

Main Views

• Inflation and Growth Relationship: There is a nonlinear relationship between inflation and growth. At very low levels, inflation may have a positive effect on growth, but above a certain threshold, it becomes harmful.
• Threshold Levels: For most developing countries, the estimated inflation threshold is around 10%, while advanced economies do not show a clear threshold, suggesting that even low levels of inflation can be detrimental to growth in the long run.
• Transition Speed: The speed of transition from the low-inflation regime to the high-inflation regime is relatively high, indicating that inflation becomes costly to growth quickly once it exceeds the threshold.
• Oil Exporters: Inflation is more costly for oil-exporting countries than for other groups, especially when using non-oil GDP growth as the dependent variable.
• Policy Implications: The findings suggest that central banks should be vigilant about inflation, as it can have immediate negative effects on growth. However, there is also a case for maintaining some inflation as a buffer to allow for monetary stimulus during financial crises.

Key Information

Threshold Model

• The LSTR model is used to estimate the smooth transition between regimes, allowing for a more nuanced understanding of how inflation affects growth.
• The model includes control variables such as investment/GDP, population growth, initial GDP, terms of trade growth, and variability in terms of trade.
• The threshold level of inflation is estimated at 10% for most countries, with advanced economies not having a clear threshold.
• The transition speed is captured by the parameter $\gamma^*$, which is divided by the standard deviation of $f(\pi_{it})$ to allow for cross-model comparisons.

Empirical Results

• The baseline LSTR model shows that inflation has a significant negative impact on growth when it exceeds the threshold.
• The R-squared value for the LSTR model is 0.426 for all countries and 0.755 for advanced economies, indicating a strong explanatory power.
• The LM test for nonlinearity is highly significant (p-value = 0.00 for all countries), confirming the presence of a nonlinear relationship.
• The estimated threshold for advanced economies is 1%, and the transition speed is 13, suggesting a slower transition in this group compared to developing countries.

Robustness Checks

• Bootstrap techniques are used to validate the precision of the estimates, showing that the threshold is robust to outliers.
• The range of bootstrap estimates is wide, indicating variability in the data and the estimation method.
• Monte Carlo experiments show that the distinction between LSTR and TAR models is not crucial due to the rapid speed of transition.

Conclusion

The study concludes that while the relationship between inflation and growth is nonlinear, the threshold of 10% for developing countries is a critical point beyond which inflation significantly harms growth. For advanced economies, the relationship is more consistent, with even low inflation levels potentially slowing growth in the long term. The LSTR model provides a more accurate and flexible framework for analyzing this relationship compared to traditional linear or TAR models, especially when considering the speed of transition and the nonlinearity of the impact.

The results have important policy implications, suggesting that central banks should be cautious in allowing inflation to rise beyond a certain level, as it can have immediate and significant negative consequences for growth. However, maintaining a small buffer of inflation may provide more flexibility for monetary policy during crises.