PUTTING YOUR MONEY WHERE YOUR MOUTH IS – THE VALUE OF PROFESSIONAL YIELD CURVE FORECASTS

Abstract – Professional yield curve forecasters do not necessarily share their end-users ’ objectives, i.e., forecast precision. Indeed, using ten years of yield curve forecasts from the Blue Chip Financial Survey, we find that forecasters clearly do not pursue a minimum mean squared error objective; rather they under-revise as revision opportunities arise (a form of rational bias). As a result, bond portfolios formed on their advice not only earn lower profits and have greater exposure to losses relative to the advice given by a standard statistical model but there is virtually no value added in professional forecast revisions. 1 Why is there a market for survey-based forecasts when we know them to be biased and inefficient relative to a mean squared error (MSE) metric? One response is that minimizing MSE may not be the appropriate objective function from which to judge forecasting performance. Shorting a bond, for instance, on the advice of a forecasted rise in its yield will earn a profit even if the forecast is biased as long as it is directionally correct. Profit does not depend on MSE