Friday, 21 October 2011: 4:55 PM
There is an enormous market for sports information, partly derived from the demand for betting on sports. We examine a large sample of forecasts of NFL game outcomes made by local and national analysts. We find evidence that local experts’ forecasts exhibit home team bias and that this bias is not the result of their superior knowledge. Our findings are consistent with the theoretical model of media bias developed by Mullainathan and Shleifer (2005), but not a model proposed by Gentzkow and Shapiro (2006). Neither set of forecasters provides predictions that would enable bettors to beat the spread.