COINTEGRATION ANALYSIS OF MONEY DEMAND-FUNCTIONS - CASE OF CHILE

This paper estimates a cointegration relation in a money demand equation for Chile using three alternatives methodologies: ordinary least squares (Engle, Granger, 1987), non linear least squares (Stock, 1987), and maximum likelihood estimation in a fully specified error correction model (Johansen, 1988). It has been shown that the one proposed by Johansen have better small sample properties; once the multivariate cointegration relation is established, the cointegrated vector is used in an error correction framework to estimate the money demand equation. The estimated model eliminates the problem of parameters instability found in earlier papers, whithout the need to introduce additional hypothesis related to the financial innovations occured in the eighties or to a structural change in the long-run money demand.