Identifying Market Power in Production Data
43 Pages Posted: 3 Apr 2019
Date Written: March 2019
Production-based estimates of markups require output elasticities for a flexible input, but these elasticities are not identified under the standard assumptions of proxy variable estimators. We show markups are identified given an additional economic restriction: constant returns to scale technology. We present Monte Carlo evidence that ignoring the identification problem, as in prior literature, introduces significant bias in estimated markups. Comparing estimators on US public firm data, we find that our approach is also more robust to testable forms of misspecification. Emerging macroeconomic models imply output and labor share wedges using our markup estimates of 1 and 11 percent for our baseline specification, half the size of non-identified estimators. A more demanding specification implies respective losses of 4 and 21 percent.
Keywords: Market Power, Markups, Production Functions, Productivity, Labor Share
JEL Classification: E2, L11, L13, D24
Suggested Citation: Suggested Citation