Short-term Interest Rates, Weekly Money Announcements and Rational Forecasts

Recently there have been a growing number of studies of the “money announcement” phenomenon. Towards the end of the week, the U.S. Federal Reserve Board announces its estimates of the narrowly defined monetary aggregate for the statement week ending some eight to ten days previously. The tendency for interest rates and other financial prices to respond to these “money announcements” has been well established. The importance of this phenomenon rests on the test it provides for the effect of monetary policy on real variables. However, existing tests incorporate at least two key assumptions. Firstly, that the Money Market Services, Inc. survey of market participants expectations is the best predictor (i.e., the rational expectation) of the weekly money announcement. Secondly, that the money announcement itself is the best predictor (i.e., the rational expectation) of the “true” change in the money supply (in practice, the true money supply is taken to be the final estimate after several data revisions). The analysis in this paper shows that both of these assumptions are rejected by the data. Readily available information can be used to improve upon both the Money Market Services forecast of the announcement and the Fed’s own preliminary estimate of the “true” money supply change. This information is used to estimate rational expectations of the change in the money stock, both with respect to information available just before the announcement and that available just after the announcement. Appropriate econometric techniques are used with these “generated regressors” to obtain consistent and efficient estimates of the announcement effect on short-term interest rates. This enables tests to be calculated to determine whether this effect, and/or the structure of the forecasts, change in response to changes in Fed policy or in its measurement of the announced monetary data.