TESTING AND ESTIMATION IN UNSTABLE DYNAMIC-MODELS - A CASE-STUDY

This paper discusses testing for parameter instability and estimation of time-varying parameters in the context of the Engle-Granger (1987) procedure. It reviews several developments in testing, in particular the new test by Bai, Lumsdaine and Stock (1991) for use in vector autoregression and error-correction models; it gives an account of the Kalman filter estimation technique; and it examines a variety of methodological matters. To illustrate the methods and issues raised, an example concerning the estimation of regional exployment multipliers for Northern Ireland is presented. The paper concludes with some remarks and recommendations for applied work in economics.