On the constancy of time-series econometric equations

Abstract

Parameter constancy is a fundamental requirement for empirical models to be useful for forecasting, analysing economic policy, or testing economic theories. However, there are surprises in defining a constant-parameter model, such that models with time-varying coefficients, and expansion of the parameterisation over time are both compatible with constancy, yet unbiased forecasts may not entail a sensible model choice. In-sample tests cannot determine likely post-sample predictive failure. A comparison of two models of UK money demand illustrates the analysis empirically, as one suffers considerable predictive failure yet the other does not, despite being identical in-sample.