The intertemporal stability of the US money demand function: new evidence from switching regressions

The demand for money remains one of the topics most extensively studied in macroeconomics. This article contributes to the debate on the money demand stability and presents further evidence of a structural shift in the US money demand function. The switching regression technique developed by Goldfeld and Quandt (1972) shows that the US money demand function displays a gradual structural break during the 1994--1995 period. The traditional Goldfeld money demand model was estimated by the nonlinear optimization methods. Consumer and corporate interest rates were included in the model specifications. In all specifications, the results show a two-regime money demand model with a significant structural shift common to the 1994--1995 period. The study period from 1966:I to 2009:IV suggests that any identified shift is the most significant break in the series. Thus, this study demonstrates that the most significant transition from the first to the second regime is gradual rather than abrupt, as suggested by the previous studies. We believe that the cause of the gradual break may be associated with the US recession in the 1992--1993 period. This finding suggests that a two-regime demand model can be used in US money demand analysis and forecasting in future.