Econometric
Models

Abstract of "Econometric forecasting," Geoff Allen, Department of Resource Economics, University of Massachusetts and Robert Fildes, The Management School, Lancaster University


Several principles with a considerable history still appear useful: keep the model simple, use all the data you can get, use theory, not the data, as a guide to selecting causal variables. Early econometric models failed in comparison with extrapolative methods because they paid too little attention to dynamics. In a fairly simple way, the vector autoregression (VAR) approach that first appeared in the 1960s resolved the problem by shifting emphasis towards dynamics and away from collecting many causal variables. The VAR approach also resolves the question of how to make longer term forecasts where the causal variables themselves need to be forecast. When causal variables do not need to be forecast, or the analyst can use other sources, a single equation with the same dynamic structure can be used. Ordinary least squares is a perfectly adequate estimation method. Evidence supports estimating the initial equation in levels, whether the variables are stationary or not. While useful to permit a large number of lags in the initial estimation, simplification by reducing the number of lags pays off. The particular test used does not seem to matter. Evidence on the value of further simplification is mixed. Error correction models will do worse than equations in levels when tests fail to find cointegration, but are only sometimes an improvement when there is a cointegration relationship.

Evidence is even less clear on whether or not to difference variables that are non-stationary on the basis of unit root tests. While some authors recommend applying a battery of misspecification tests, few econometricians use (or at least report) more than the familiar Durbin-Watson test. Consequently, there is practically no evidence on whether model selection based on these tests will improve forecast performance. Limited evidence on the superiority of varying parameter models hints that tests for parameter constancy are likely to be the most important. Finally, econometric models do appear to be gaining over extrapolative or judgmental methods, even for short-term forecasts, though much more slowly than their proponents had hoped.

Keywords: econometric forecasting, vector autoregression, error correction model, specification testing, forecast comparisons