Econometric modelling of the relationship between money, income and interest rates in the U.K., 1963-1978

This thesis investigates empirically the relationship \nbetween money, income and interest rates in the U.K. over the \nperiod 1963 to 1978. After developing univariate models of \nthe time series' proxying these theoretical variables, the \nparadox existing between the conventional theoretical model, the \nIS/LM framework, and the usual empirical practice of directly \nestimating the demand for money function is investigated. It \nis shown that the crucial issues are the exogeneity assumptions \nplaced on the IS/LM framework. As such assumptions cannot be \ntested in a static framework, a dynamic analogue of the IS/LM \nmodel is developed, along with appropriate methods for testing \nexogeneity in dynamic multivariate systems. Empirical tests \nshow that the assumptions of the exogeneity of money and government \nexpenditure are invalid, but that the direct estimation of \ndemand for money functions is appropriate. This leads to an \ninvestigation of the dynamic structure and functional form of \nthis function using recently developed techniques based on specification \nsearch procedures. \nA major conclusion of this study is that the IS/LM model is \nan invalid framework for empirical research, and in particular \nmoney cannot be regarded as being exogenously determined. Indeed, \nthere is no evidence of feedback from money to either real income \nor prices, although both statistical and economic reasons are \nadvanced for the possibility that such feedback cannot be detected \nby the techniques employed. Important short run dynamic effects \nare found on the demand for money with respect to real income, \nprices and interest rates. Furthermore, both the wage rate and \nan own rate of interest variable are also important determinants \nof money demand. The demand for narrow money function also \nexhibits sensible long run behaviour and has an adequate \npredictive performance but, unfortunately, the broad money function \nhas no long run properties and predicts unsatisfactorily.