An Experimental Study of Coordination in Games with Strategic Complementarities: Long Run vs. Short Run Relationships Kyle Hyndman Southern Methodist University, hyndman@smu.edu, http://faculty.smu.edu/hyndman Santiago Kraiselburd Zaragoza Logistics Center & INCAE Business School, skraiselburd@zlc.edu.es Noel Watson Zaragoza Logistics Center, nwatson@zlc.edu.es In this paper, we study behavior in a series of two-player supply chain game experiments. Each player simultaneously chooses a capacity before demand is realized, and sales are given by the minimum of realized demand and chosen capacities. We focus on the differences in behavior under fixed and random matching. Intuition suggests that long-run relations should lead to more profitable outcomes. However, our results go against this intuition. While subjects are better-aligned under fixed matching, average profits are more variable and learning is slower. Over the last 5 periods, average profits are actually higher under random matching. We also show that subjects in the fixed matching experiments have a first-impressions bias, in which the greater the misalignment in initial choices, the lower are average profits. Finally, under fixed matching, subjects react more strongly to lagged variables, which may lead to coordination failures. Key words : Long-run relationships, Coordination, Global Games, Supply Chains 1. Introduction Firms have a choice of the use of long term or short term relationships to structure their supply chains. These relationships serve both to align incentives and to improve and align actions between partners. In this paper we focus on the latter role of these relationships and collect experimental evidence for the benefit of long term versus short term relationships. The context of this study is coordination in games with strategic complementarities. Such games typically have multiple Pareto rankable equilibria, making the question of which equilibrium will arise in the actual play of the game important. Our game involves two players choosing costly individual production capacity, such that the minimum of capacity across the players dictates available capacity for meeting the uncertain demand. In such an environment it is natural to expect that long run relationships should allow players to reach more profitable equilibria than short run relationships. The goal of this paper is to put this conventional wisdom to the test through a series of experiments. 1