Biased Arbitration (joint with Gregor Matvos and Amit Seru)

Abstract:

We examine investor outcomes in securities arbitration. Using a novel data set containing roughly 7,000 arbitration cases, we find evidence suggesting that arbitrators exhibit persistent disparity in their decisions. Brokerage firms appear to utilize this information in the arbitrator selection process. Arbitrators that were more industry-friendly in the past are almost forty percent more likely to be selected relative to arbitrators who were more investor-friendly in the past. The FINRA Code of Arbitration Proceedings, which all brokerage firms are bound to, provides both claimants and respondents control over the arbitrator selection process. We develop and estimate a structural model of arbitrator selection to quantify the costs and benefits of the current arbitrator selection system. In the model arbitrators use their reputation to compete against each other for the attention of claimants and respondents. We find evidence suggesting that limiting respondent's and claimant's inputs over the arbitrator selection process could improve outcomes for investors.