Qualitative survey responses and production over the business cycle

An examination of Swedish manufacturing data on real output and qualitative business tendency survey (BTS) responses from 1968 through 1998 reveals that survey-based attitude data typically improve the fit of simple autoprojective models of manufacturing output growth. It also turns out that traditional autoregressive distributed lag (ADL) models based on business survey data can provide more accurate one-quarter-ahead forecasts of output growth than naive alternatives. Another finding is that when BTS variables concerning ex post (ex ante) output growth are included in the empirical specifications, then no other ex post (ex ante) business survey variables seems to include any additional information about output growth.