1996年-BIS国际清算银行_Does_the_term_structure_predict_recessions_The_international_evidence__30页_275kb

Does the Term Structure Predict Recessions? The International Evidence

Core Content

This working paper by Henri Bernard and Stefan Gerlach investigates the ability of the term structure of interest rates to predict recessions across eight countries: Belgium, Canada, France, Germany, Japan, the Netherlands, the United Kingdom, and the United States. The study builds on earlier research by Estrella and Hardouvelis (1991) and Estrella and Mishkin (1995a, b), which showed that term spreads (the difference between long-term and short-term interest rates) can be useful in forecasting economic downturns.

Main Findings

• Term Spreads as Predictors: The yield curve provides useful information about the likelihood of future recessions in all eight countries. The slope of the term structure is a significant indicator for monetary policy purposes.

• Forecast Horizon: Term spreads are effective in predicting recessions up to two years ahead. The predictive ability varies by country, with the highest being in Germany, followed by the United States and Canada, and the lowest in Japan.

• International Differences: While domestic term spreads are useful, foreign term spreads add limited information for most countries. Exceptions include Japan (where the German spread adds predictive value) and the United Kingdom (where the US spread provides additional insight).

• Recession Probabilities: The estimated recession probabilities based on the term spread shift considerably in advance of actual recessions. For example, in Germany, a term spread of -4% indicates a 99% probability of recession within four quarters.

• Leading Indicators: Although leading indicators contain some information about future recessions, this information is only useful for the immediate future (up to one to three quarters). Term spreads, on the other hand, remain informative for longer horizons.

Key Information

• Methodology: The study uses probit models to estimate the probability of a recession based on term spreads. The dependent variable is a dummy variable indicating whether a recession occurred, and the independent variable is the term spread.

• Recession Dating: To ensure consistency, the authors use recession dates from Artis et al. (1995), which are based on the same criteria across all countries, thus avoiding potential biases due to differing definitions.

• Data Period: The analysis covers quarterly data from 1972:1 to 1993:4, with Belgium and the Netherlands starting from 1977:1.

• Significance of Results: The significance of term spreads is measured using t-statistics, with *, **, and *** denoting significance at the 10%, 5%, and 1% levels, respectively.

• Explanatory Power: The explanatory power of the term structure, as measured by pseudo R², is "hump-shaped" — it increases with the forecast horizon up to a certain point and then declines. This suggests that while term spreads are useful for predicting recessions, their effectiveness diminishes over time.

Conclusion

The paper concludes that term spreads are a valuable indicator for predicting future recessions, particularly for the United States and Germany. While foreign term spreads add limited predictive value for most countries, they are more informative in Japan and the United Kingdom. Leading indicators provide short-term forecasts but do not offer significant information for longer horizons, where term spreads continue to be useful. The findings support the use of term spreads in monetary policy decision-making as they offer reliable, timely, and comprehensive insights into the likelihood of economic downturns.