A Comparison of U.S. Housing Starts Forecasts

This study examines the Federal Reserve and private forecast accuracy of housing starts. We show that the Federal Reserve (private) forecasts are (generally) unbiased and superior to the random walk benchmark. At the shorter horizon, the Federal Reserve and private forecasts embody distinct predictive information, indicating that one can gain a significant improvement in forecast accuracy by combining the two sets of forecasts. At the longer horizon, our findings support the asymmetric information hypothesis that the Federal Reserve forecasts embody useful predictive information beyond that contained in the private forecasts.