A Large-Scale Evaluation of Merger Simulations

The PIs will review all decisions for mergers and will also devote effort to coding and testing the post-merger simulation. While merger simulation under Bertrand-Nash is available through pyBLP (Conlon and Gortmaker, 2019), the extensions to Bertrand-Nash require de novo code that can be robustly applied to many mergers. The PIs need to conduct all data analysis of intermediate and final results as well. The PIs will also participate in data cleaning as needed to support the hourly RAs. The most direct policy implication of this study will be how it informs the use of merger simulations in practice. The Horizontal Merger Guidelines (§6.1) note that agencies can use merger simulations to supplement decisions, especially if a wide range of simulations point to significant price increases. If this analysis suggests that simulations systematically underpredict post-merger prices, agencies that take this into account in their holistic evaluation of simulation evidence, perhaps weakening the price effects one must expect to challenge a merger