Executive Summary

1         Short-term interest rates are typically expected to fall as an economy slows down and heads toward a recession. In response to comments that a negative-sloping yield curve for US Treasury securities is the harbinger of deflation in the US economy, this paper assesses the reliability of the yield curve in providing information about future inflation and economic activity.

2         Our analysis indicates that the positive relationship between yield spreads on Treasury securities of different maturities and future changes in inflation becomes more evident, as the difference in maturities increases. The strength of this relationship may be expressed as a function of the variations of the expected change in inflation and the ex ante real interest rate spread. We find that as we move along the Treasury yield curve, variations in expected inflation changes become large relative to variations in the real interest rate spread, so that inflation becomes the dominant factor driving the returns on Treasury securities. In other words, the relationship between the yield spread and future changes in inflation becomes stronger.

3         Next, we evaluate the effectiveness of the nominal term structure in predicting growth in real GDP. Our findings indicate that while yield spreads calculated using different segments of the yield curve are all statistically significant, the yield spread between three- and one-year Treasury bills provides the maximum predictive power of growth over the next one and two years. For a given segment of the yield curve, the forecasting power of the spread diminishes as the forecast horizon increases.

4         We also use the nominal term structure to forecast the probability of a recession. The results show that yield spreads from various segments of the yield curve are significant predictors of recessions four quarters ahead for the period 1960Q1 to 1998Q3. However, the forecasting power of our equation diminishes substantially when it is used to forecast the probability of a recession eight quarters ahead. Plots of the estimated probabilities show that they are relatively higher in recession quarters than in non-recession quarters, although there a false alarm was generated in the early 1980s.

5         Based on our model estimates and the yield spread between three- and one-year Treasury bills, the probability of a recession in the next four quarters fell from 29% on 31 Aug 98 (before the rate cuts instituted by the Federal Reserve) to 27% on 4 Jan 99 (after the rate cuts) and 17% on 15 Mar 99. Furthermore, the positive yield spread between three- and one-year Treasury bills on 15 Mar 99 suggests that real GDP will grow by 3.0% over the next four quarters. This compares favourably with 1999 growth of 3.3% projected by Consensus Forecasts on 8 Mar 99.