RDP 8602: Short-Term Interest Rates, Weekly Money Announcements and Rational Forecasts 6. Concluding Comments

Two important assumptions implicit in existing studies of the money stock announcement effect have been shown to be rejected by the data. The measure of the expected change in the money supply provided by the median of Money Market Services' survey, is not found to be a rational expectation of the change that is announced on the weekly H6 statistical release. Furthermore, neither these data nor the announced data are found to be rational expectations of the actual change in the money supply that occurred in the relevant statement week.

When these results are taken into account, the estimates of the announcement effect on short-term interest rates yield different inferences. The main difference for the Federal Funds rate is finding a much larger increase in the announcement effect following the Fed's change to a nonborrowed reserves operating policy in October 1979.

The differences for the T-bill rate are more substantial. The modified model provides evidence of a significant fall in the response of this interest rate following the October 1982 change in Fed operating policy (to a borrowed reserves target), as well as a significant increase following the earlier change. The 1982 response is not detected by hypothesis tests on the standard model. Moreover, the modified model removes many of the problems raised by the rejection of the market efficiency hypothesis in the standard model.

At least two puzzles remain for the period between these two changes in Fed policy. Firstly, the modified model rejects the market efficiency hypothesis for the period immediately following the first policy change (as does the standard model). Secondly, breaks in both the formation of expectations and in the announcement effect on the T-Bill interest rate, are detected following the renaming of the MIB aggregate to “new” MI.

Given that the modified model produces substantially different conclusions for a number of the hypotheses examined, it is likely to produce different inferences with respect to the announcement effect on longer-term interest rates and in other asset markets. The extent to which that will help distinguish between the different explanations of the announcement effect, and thus shed light on the efficacy of monetary policy, remains a topic for further research.