| We measure forecast power of both implied forward rates and the yield spreads between long- and short-term bonds from the term structure from 1959:2 to 1995:12 by various methods. The major hypothesis to test is whether or not forecast power varies across monetary policy regimes; the first interest rate targeting regime (59:2-79:9), the non-interest rate targeting regime (79:10-82:9), and the second interest rate targeting regime (82:10 to date). It turns out that the forecast power does vary significantly between the 79-82 and the 82-95 eras but not between the 79-82 and 59-79 eras. The second part of this dissertation is concerning a few possible causes for these inconsistent test results. We take a look at two macroeconomic variables--inflation and output growth rates--suggested by the Taylor Rule and find that the variance of inflation or unemployment rate is negatively related to the forecast power of the term structure. Then we consider a fact that the Fed funds rate targets were not available before 1985. If the Fed sets the targets to control future short rates more effectively and the targets are publicly available, the market will be able to predict the future path of short rates more accurately with the targets than without. It turns out that the market does predict future interest rates much better with the targets. Thus, all three factors--inflation, output, and the fed funds targets--prove to be closely related to interest rate forecast, and this is partly why the forecast power does not always vary the same way we expect when we consider only policy regimes as a variable to affect interest rate forecast power. |
