Measuring markups with production data (with Amit Gandhi and James Traina).
Abstract: We show standard methods to estimate production functions do not identify markups. This nonidentification creates spurious skewness in estimated markup distributions. We also show that ex-ante structure on the returns to scale solves the identification problem. In US public firm data and in a Monte Carlo experiment, we find that applying constant returns to scale performs remarkably well and reduces the skewness in the markup distribution among public-firm by as much as half in comparison to nonidentified estimates. This results in half the efficiency losses in output and labor shares when calibrated to a recent macroeconomic model.
Check it out at: https://papers.ssrn.com/sol3/papers.cfm?abstract_id=3358472.