رفتار توده وار پیش بینی چرخه کسب و کار   Herding behavior of business cycle forecasters

1. Introduction Business cycle and growth expectations play a major role in understanding macroeconomic relationships. They also determine the extent to which economic policy agents, including central banks, can influence macroeconomic outcomes. One way to deal with forecast uncertainty is to pool the expectations of professional forecasters (Zarnowitz, 1984) in order to hedge against the errors of individual forecasters, thus improving the forecast quality. Such surveys of professional forecasters are provided by either central banks or private companies. The idea of these consensus forecasts is that, although individual forecasters may outperform the average of a group of forecasters in certain cases, an individual forecaster rarely outperforms others systematically. Zarnowitz and Lambros (1987) find that the forecast errors of consensus forecasts are smaller than those of most individual forecasters. Batchelor (2001) shows that consensus forecasts are more accurate than the projections published by the OECD or the IMF. The reliability and superiority of consensus forecasts depends crucially on whether the forecasters actually reveal their own best forecast or behave strategically, i.e., show herding or anti-herding tendencies. Forecaster herding arises if the forecasters ignore their private information and instead follow the forecasts of others (Scharfstein & Stein, 1990). For example, Bewley and Fiebig (2002) show that interest rate forecasters tend to indicate values in the safe consensus range, in order to avoid sticking their neck out with ‘‘extreme’’ forecasts. This is because a poor forecast may not damage a forecaster’s reputation if other forecasters also delivered poor forecasts. Thus, herding behavior biases the distribution towards the mean. Forecaster herding should not be confused with forecast clustering, where similar forecasts may be observed because all forecasters have access to the same set of economic data and similar forecast techniques. Herding behavior, on the other hand, refers to forecasters deliberately deviating from their best private forecasts for strategic reasons. Forecaster anti-herding may arise if forecasters, for strategic or other reasons, deliberately scatter their forecasts away from the forecasts of others. This may arise when a forecaster’s income (or reputation) depends not only on the accuracy of their own forecasts, but also on their relative performances. If some of the customers of professional forecasters buy forecasts only occasionally,