The OLS estimator of the effect of concentration on price is biased for two reasons. First, performance feeds back into structure, causing a simultaneous equations bias. Second, as a function of outputs or revenues, measured concentration is correlated with determinants of price that are, at best, measured with error, so measured concentration is correlated with the error term. With panel data, fixed-effects procedures can be combined with instrumental variables to eliminate bias from both sources. In concentration-price regressions for the airline industry, the bias is found to be substantial and negative.