Multi-period Survey Forecasts Of Business Investment: Is There Room For Improvement?

Abstract

The ASA-NBER multiperiod survey forecasts of business investment are compared with univariate forecasts to assess predictive information content. In general, the survey forecasts fail to be unbiased, and, none fully exploit the information in the past history of business investment. Interestingly, however, they contain predictive information on other relevant (quantitative or qualitative) variables. Combined forecasts of survey and univariate models score significant improvements over either, suggesting their potential usefulness in policy-making.