ON COMPARING MACROECONOMIC MODELS USING FORECAST ENCOMPASSING TESTS

It is clearly of interest to macroeconomists to be able to evaluate wh ether one large-scale macroeconometric model 'is better' than another. Although comparisons between models are sometimes invidious, because the purposes for which the models were built differ, it is the case th at formal comparisons of two models' statistical properties are rare. This is in spite of considerable theoretical advances in the econometr ic methodology, namely the development and use of non-nested and encom passing tests. Chong and Hendry (1986) advocate the use of the forecas t encompassing regressions, where the outturns are regressed on compet ing (one-step-ahead) forecasts. This paper reports the findings of app lying this rather easy-to-use method of comparing large scale macroeco nometric models. The forecast data we use are those published by three macroeconometric modelling groups, namely: Liverpool; the National In stitute; and The London Business School. Forecasts for up to three yea rs ahead are published for unemployment, growth, and inflation, throug hout the 1980's. Forecast encompassing tests fail to separate one mode l from another, based on one-year-ahead forecasts. Each model 'wins' o nce. However, the conclusions are not the same as using root-mean-squa re-forecast-error criteria, illustrating Clements and Hendry's (1994) observation that minimum root-mean-square-forecast-error is neither ne cessary nor sufficient for a model to have constant parameters, to pro vide accurate forecasts, or to encompass its rivals.