2009年-BIS国际清算银行_Consumption_and_real_exchange_rates_in_professional_forecasts_47页_512kb

Summary of "Consumption and Real Exchange Rates in Professional Forecasts"

Core Content

This working paper by Devereux, Smith, and Yetman investigates the relationship between consumption growth and real exchange rates in professional forecasts, with a focus on the implications of incomplete asset markets for international risk sharing. The paper explores the Backus-Smith puzzle, which refers to the lack of empirical support for the theoretical prediction that relative consumption growth and real exchange rate depreciations are positively correlated in a world with complete asset markets and full risk sharing.

Main Viewpoints

• Theoretical Prediction: In a world with complete asset markets and full risk sharing, the relative consumption growth and real exchange rate depreciations should be positively correlated.
• Empirical Evidence: Historical data shows little to no such positive correlation, sometimes even a negative one, suggesting that the assumptions of complete markets and full risk sharing may not hold.
• Incomplete Markets: Theoretical models with incomplete asset markets can produce negative or low correlations between consumption growth and real exchange rate changes, which aligns better with the observed data.
• Professional Forecasts: The authors use professional forecasts for 28 countries over the period 1990–2008 to test the implications of incomplete markets. They find no evidence of a positive correlation, deepening the Backus-Smith puzzle.
• Consumption Forecast Errors: The paper suggests that the presence of 'hand-to-mouth' consumers—those who do not adjust their asset holdings optimally—may explain the weak link between consumption and real exchange rate forecasts.
• Testing Risk Sharing: The authors explore different tests for risk sharing, including the state-by-state test, the monotonicity-in-means test, and the covariance test. These tests are generally weaker than the complete model prediction and may not be reliable in detecting the conditional relationship.

Key Information

• Empirical Findings:

  • There is no evidence of a positive correlation between consumption growth and real exchange rate changes in professional forecasts.
  • The correlation between consumption and real exchange rate changes in forecasts is close to zero, consistent with the Backus-Smith puzzle.
  • The paper provides graphical evidence showing that forecast data points do not lie along an upward-sloping line, which is expected under the complete markets assumption.

• Theoretical Models:

  • Models with incomplete asset markets, such as non-contingent bonds, predict a negative or weak correlation between consumption and real exchange rates.
  • These models suggest that real exchange rate depreciation can be associated with an increase in relative consumption due to wealth effects from shocks to traded goods.

• Forecasting Challenges:

  • Consumption and real exchange rates are difficult to forecast econometrically, especially under floating exchange rates.
  • Weak instruments and the difficulty of constructing real-time forecasts challenge the use of instrumental variables and GMM estimation methods.
  • The use of a vector of ones as an instrument leads to the monotonicity-in-means test, which may not be informative if unconditional means are constant or not different across countries.

• Implications:

  • The failure of the state-by-state test suggests that the complete markets assumption is not valid.
  • The paper highlights the importance of understanding consumer behavior and the limitations of using aggregate forecasts to test for risk sharing.
  • The presence of hand-to-mouth consumers may help explain the weak link between consumption and real exchange rate forecasts.

Structure of the Paper

1. Introduction: Introduces the Backus-Smith puzzle and the role of incomplete markets in explaining it.
2. Risk-Sharing Indicators and Incomplete Markets: Explains the theoretical framework for risk sharing under complete and incomplete markets.
3. Tests and Previous Evidence: Reviews previous empirical tests and their limitations, and introduces the new tests used in the paper.
4. Forecast Data Sources: Describes the data used for the analysis, including professional forecasts for 28 countries.
5. Empirical Evidence: Presents graphical evidence and statistical results from the analysis of forecast data.
6. Interpretation of Results: Discusses the implications of the findings, particularly in the context of consumer behavior and risk sharing.
7. Conclusion: Summarizes the key findings and their significance for international macroeconomic models.

Conclusion

The paper provides a critical analysis of the relationship between consumption growth and real exchange rates in the context of incomplete asset markets. It finds no evidence of a positive correlation in professional forecasts, which deepens the Backus-Smith puzzle. The authors argue that the presence of hand-to-mouth consumers and the challenges of forecasting consumption and exchange rates may help explain this result. The paper highlights the importance of using appropriate testing methods and the limitations of empirical tests in capturing the true nature of international risk sharing.