RDP 9204: The Term Structure of Interest Rates, Real Activity and Inflation 5. Summary and Conclusions

The evidence presented in this paper suggests that the slope of the nominal yield curve is useful in predicting the future path of economic activity and inflation. The results suggest that the steeper the upward slope of the yield curve, the faster will be the rate of growth of output over the next one and a half years. However, slower growth after about six quarters eventually dissipates the additional increase in output. The results suggest that the yield curve should not be used to forecast economic activity in the near term (that is, less than three quarters). Such forecasting is much better done by the index of leading indicators. The yield curve does, however, provide a good guide to the cumulative growth in activity over the following six quarters. On the inflation front, the slope of the yield curve provides little information on changes in inflation over periods less than one year but is useful in predicting the difference between the inflation rate over the next two years and the inflation rate over the next year.

If expected inflation affects only nominal interest rates and has no effect on real rates then the slope of the nominal yield curve should provide no information about the path of real activity but should help predict the change in inflation. The fact that the slope of the nominal yield curve does help predict activity suggests that the slope of the real yield curve is not constant. Indeed, the fact that, in the inflation change equations, the long-short spread provides more information than the slope of the yield curve over shorter maturity differences also suggest that the real yield spread is not constant.

Taken together, the results on activity and inflation are consistent with monetary policy having effects on real interest rates through price stickiness. Long term interest rates more accurately reflect expected inflation than do short term rates which are heavily influenced by the liquidity effect. For example, higher expected inflation as the result of higher expected future monetary growth initially has little effect on short term rates as real money balances remain unchanged. In contrast, the higher inflationary expectations are reflected in long rates. Over time, as prices gradually adjust, real short rates can rise to reflect the higher inflation.

The above empirical results are not without their qualifications. Foremost amongst these is the relatively short period over which market determined interest rates are available. Such data exists for only nine years. This leaves relatively few truly independent observations, especially when looking at cumulative output and inflation changes over periods longer than a year. At a more abstract level, it is difficult to assess the degree to which both the monetary authorities and the agents in the economy take into account the above correlations in the making and interpreting of policy. The above tests are predicated on the assumption that both the authorities and other agents in the economy did not change their behaviour with respect to the yield curve over the sample period. Any such changes could alter the relationship between the slope of the yield curve and the future path of economic activity. Notwithstanding these qualifications this paper provides both theoretical and empirical support for using the slope of the yield curve as an indicator of the future paths of both real economic activity and inflation.