Communication and Bargaining Breakdown: an Empirical Analysis

Matthew Backus (Columbia, NBER, CEPR)
Thomas Blake (Amazon)
Jett Pettus (Columbia)
Steven Tadelis (Berkeley, NBER, CEPR)

Abstract : Bargaining breakdown—whether as delay, conflict, or missing trade—plagues bargaining in environments with incomplete information. Does communication alleviate these costs? Prior theoretical and experimental evidence is ambivalent. We examine this question using field data from an online bargaining marketplace: eBay Germany’s Best Offer platform. On May 23, 2016, the platform introduced unstructured communication allowing buyers and sellers on the desktop version of the site, but not the mobile app, to accompany price offers with a message. Using this natural experiment, our differences- in-differences approach documents a 14% decrease in the the rate of breakdown among compliers. Though adoption is immediate, the effect is not. We show, using text analysis, that the dynamics are consistent with repeat players learning how to use communication in bargaining.

Outside Options, Bargaining, and Wages: Evidence from Coworker Networks

Sydnee Caldwell (Microsoft Research/UC Berkeley)
Nikolaj Harmon (University of Copenhagen)

Abstract : This paper analyzes the link between wages and outside employment opportunities. To overcome the fact that factors that affect a worker’s outside options may also impact her productivity at her current job, we develop a strategy that isolates changes in a worker’s information about her outside options. This strategy relies on the fact that individuals often learn about jobs through social networks, including former coworkers. We implement this strategy using employer-employee data from Denmark that contain monthly information on wages and de- tailed measures of worker skills. We find that increases in labor demand at former coworkers’ current firms lead to job-to-job mobility and wage growth. Consistent with theory, larger changes are necessary to induce a job-to-job transition than to induce a wage gain. Specification tests leveraging alternative sources of variation suggest these responses are indeed due to information rather than unobserved demand shocks. Impacts on earnings are concentrated among workers in the top half of the skill distribution. Finally, we use our reduced-form estimates to identify a structural model that allows us to estimate bargaining parameters and investigate the relevance of wage posting and bargaining across different skill groups.

Optimal Bargaining on Ebay Using Deep Reinforcement Learning

Etan Green (University of Pennsylvania)
Barry Plunkett (University of Pennsylvania)

Abstract : Reinforcement learning algorithms now outperform the best humans in a wide variety of Markov Decision processes (MDPs), such as chess and Go. We 1) formulate bargaining in "Best Offer" listings on eBay as an MDP; 2) train neural networks to behave like human buyers and sellers using a large, publicly available dataset of Best Offer listings; 3) train a reinforcement learner to play optimally against these agents as either the seller or a buyer; and 4) characterize the learner's behavior. More generally, we provide a template for estimating optimal policies in economic settings where experimentation is infeasible but data are plentiful.