Too Many Cooks? The German Joint Diagnosis and Its Production

The "Gemeinschaftsdiagnose" [Joint Diagnosis (JD)] is the most influential semi-annual mac-roeconomic forecast in Germany. Jointly produced by up to six institutes, its accuracy as well as the large number of involved participants is often criticised. This study examines the JD's growth and inflation forecasts from 1970 to 2007, including most of the contributions of the forecasts submitted by the five institutes at the start of the JD. Four questions are addressed: (i) Are these forecasts unbiased and efficient? (ii) How do results change if we presume an asymmetric loss function? (iii) Are any of the institutes more accurate than the JD? Are five/six institutes necessary and at what cost? (iv) Do the institutes make strategic forecasts to influence the JD forecast? Results show that there is no strong evidence of bias or inefficiency of the institutes' forecasts and no evidence of asymmetric loss functions. Five institutes are not necessary, but it is very hard to predict the redundant institutes; however, the loss of accu-racy by employing only two is small.