Short-term forecasting of industrial production with business survey data; Experience from Finland’s great depression 1990-1993

The severe Finnish recession in the early 90s provides an interesting testing ground for forecasting models specified and estimated before the recession. The authors use recent data to evaluate some short-term forecasting models for industrial production. The main explanatory variables are from business surveys and the models themselves are based on the use of the Kalman filter. The recession years present difficulties for forecasting especially in the textile industry and metal industry. In the food industry and to some extent in the forest industry the forecasting performance during the recession is actually better than in earlier periods. Mechanical re-estimation of the models yields better forecasting results in four out of six branches studied. The importance of business survey information however, seems to have increased during the recession. The improvement in production accuracy after taking account of relevant business survey information is statistically significant in the forest industry and in manufacturing of non-metallic products when the precision of autoprojective forecasts is used as a baseline.