EVALUATING CONFLICTING STABILITY RESULTS IN GERMAN MONEY DEMAND REGRESSIONS

In this paper we have tried to find the central assumptions necessary to obtain stable money demand functions for the Federal Republic of Germany. We found the hypothesis about the seasonal process to be such a crucial assumption. Once changes in the seasonal pattern are accounted for, the conventional Goldfeld specification yields relatively good results concerning stability. More complex formulations of money demand using complicated lag structures, the theoretically dubious rate of inflation and foreign interest rates are not necessary to achieve stability. The instability of money demand as a consequence of a break in the seasonality of the income process is attributed to measurement errors, as quarterly income reflecting changes in seasonal production patterns does not properly take into account the seasonal structure of the volume of transactions. This conjecture also helps to explain the differing regression results based on alternative data transformations. Taken together, the evidence suggests to accept the old-fashioned Goldfeld specification in levels as not such a bad description of the long-run characteristics of Germany's money demand. Inferences on the speed of adjustment derived from this equation should, however, not be trusted. © 1990 Institut fur Weltwirtschaft an der Universitat Kiel.