Estimating the returns to information: a gaming approach