An experimental examination of subjective forecast combination

Little is known about how individuals subjectively combine forecasts. This study uses two experiments to investigate how subjective forecast combination is influenced by both features of the decision context and characteristics of expert forecasters (financial analysts) who provide the individual forecasts. Experimental results indicate that individuals often are conservative in their combined forecasts; this conservatism is related to both the intended purpose of the forecast and individuals’ combined forecasts on average reflect the relative historical accuracy of forecasters but not the relative historical dependence between forecasters. Although some subjects used dependence correctly in their combined forecasts, others either used it incorrectly or indicated that they did not know how it should be used. These results suggest that subjectively combined forecasts are likely to differ from those based on theoretical forecast combination models.