Dynamic Behaviour and Policy Characteristics of the Model

Abstract

Any dynamic econometric model which purports to show the impact of macroeconomic policies must be able to track the historical data reasonably well. The process which enables us to examine this property of a model is called simulation. This way of evaluating and validating simultaneous-equation models, like the one presented in this book, is of paramount importance because, although each individual equation may have a very good statistical fit (as conventionally measured), the whole model may fail to reproduce the historical time series satisfactorily. Similarly, the individual equations of a model may perform poorly on standard statistical criteria, and yet the model as a whole may be capable of reproducing the historical data rather closely. This could easily happen ‘since the behaviour of the model as a dynamic system may bear little relation to the way individual equations fit the historical data’ (Pindyck and Rubinfeld, 1976, p. 317).