The purpose of this work is to examine certain hypotheses about the predictability of the narrowly defined demand for money function in Japan and, in addition, to see what light these tests might shed on the United States. Similarities in the performance of empirical money-demand equations for both the United States and Japan raise serious questions about the function's overall predictability.1 Moreover, they support the general belief among policymakers that the assumed stability of the narrowly defined money demand relationship has collapsed.
Judd and Scadding2 have summarized most of the relevant literature for the post--1974 instability and unpredictability of the U.S. demand-for-money function. Prior to the mid-1970s, most money-demand experts believed that the demand for money exhibited a stable relationship with a small set of macroeconomic variables.3 Empirical money-demand equations for the pre-1974 period, though, proved unable to describe accurately the post-1974 period for either country. As a matter of fact, the "missing money" phenomenon common to both countries4 served to show that conventional money-demand