IMPROVED ESTIMATES AND FORECASTS OF ERROR-CORRECTION MODELS IN ECONOMICS

This paper is concerned with improved estimation and better forecasting of major economic activity within the context of error correction (EC) models as pioneered by Sargan in 1964, and extended by numerous economists over the past three decades. A new and general econometric methodology with superior average mean squared errors (MSE) for EC models is introduced and, for illustration, applied to the estimation and forecasting of private consumption and demand for money in Australia for the period 1960-1992. The improvement in terms of historical and ex post forecasting MSE of our new methodology over the conventional statistical approaches currently in use for EC models is found to be up to 15.43%.