An examination of vector autoregressive forecasts for the U.K. economy
The vector autoregressive (VAR) approach to forecasting is applied to economic variables for the U.K. Unrestricted VAR and Bayesian VAR models are contructed using the data available in November 1981. Forecasts for 1981-1984 are compared with forecasts from four economic models and also from simple AR and ARIMA models. The economic models give the best forecasts for the growth of output and inflation but the other methods perform well for the remaining variables. The accuracy of VAR forecasts varies with the data used for fitting the models and the prior assumptions made. Finally, forecasts for 1987-1992 are presented, using the data available in November 1987.