Financial innovations, money demand, and disaggregation: some time series evidence

This paper explores the instability in estimated money demand functions. Using a new data series on credit card usage, we evaluate the role of financial innovations in stabilizing the M1 demand function over three troubling episodes. We find that our measure of financial innovations improves the short-term predictive ability of the M1 demand function, but does not generate stable long-run elasticities. Structural instability remains even after accounting for seasonal adjustment, the turbulence in the second and third quarters of 1980, and an alternative transactions measure. ; Financial innovations are likely to have differential effects on the components of M1, and we estimate separate models for currency, demand deposits, other checkable deposits, and total deposits. Our financial innovations series continues to improve short-run predictions, and the currency demand equation is much more stable than the M1 equation. ; Lastly, we analyze the sluggish adjustment of money holdings as a source of structural instability. We argue that theory fails to identify the adjustment parameter, and establish that minor variations in this parameter lead to minor variations in the likelihood function but major variations in long-run elasticities. ; We conclude that financial innovations are a useful element in forecasting short-run money demand, but are not the primary cause of money demand instability, which stems from deeper problems with the basic specification. Modeling and estimating the components of M1 (and perhaps M2) appear promising directions for future research.