Analysis of Three Variable VECM Models Including Demand Functions for Real Balances

Abstract

The review of the existing literature on empirical money demand functions presented in Chapter 2 suggests that very little confidence is derived from existing single equation estimates, either with respect to the existence of a stable aggregate demand function for real balances, or with respect to the magnitude of the parameter values of such a function, should it exist. In Chapters 3 and 4 newly developed techniques for multiequation analysis are discussed. The purpose of this and subsequent chapters is to apply these techniques to investigations into the existence of long-run aggregate demand functions for real balances. The analyses involve testing for cointegrating vectors among macroeconomic data series and identification of a long-run demand function for real balances among those cointegrating vectors. The identified long-run money demand functions are examined for stability across various sample periods, particularly the periods of “missing money” and those of velocity trend shifts. The robustness of the long-run money demand function is examined with respect to the dimension of the multivariate process and the number of cointegrating vectors. Finally, various techniques of identifying permanent and transitory “economic shocks” are applied to the vector error correction (VECM) models containing the money demand equation, and the resulting dynamic patterns are analyzed.