Financial innovation and money demand : theory and empirical implementation

Empirically, traditional money demand equations are frequently characterized by periods of"missing money", unstable parameters, and autocorrelated errors. The common practice to solve these problems consists of changing the specification of the regressions once the shifts (which are usually associated to financial innovation) are identified. This paper provides an alternative approach to dealing with the unobservable process of financial innovation. It consists of modelling financial innovation as shocks that have permanent effects on the money demand, analogous to productivity shocks in production functions. This paper describes the theoretical model used and shows the failure of traditional money demand equations using cointegration techiques. It describes a simple GLS-iterative econometric model which allows the authors to recover the path of financial innnovation and obtain sensible estimates of the relevant elasticities. It also shows Monte Carlo simulations to evaluate the behavior of the estimation procedure for particular samples and data generating processes, and to study how robust the procedure to some deviations from the basic assumptions is.