Contract theory typically assumes contracts reflect deliberate attempts to maximize profit. At the same time there is a large and fiercely competitive industry – full-service restaurants – where different firms all use the same basic wage contract, and have done so for decades. How can the same basic contract maximize profit at all these firms, in all these eras? This paper uses a combination of experimental, interview, and observational data from large-scale restaurants to investigate whether and to what extent the customary contract – tips and hourly wages – is consistent with profit maximization. The paper explains that the contract creates an incentive problem relating to multitasking, even if the worker is risk neutral and has no effort costs. It uses a field experiment, lasting the better part of a year, to show there is an alternative that makes both better off. Under the alternative workers earn 10 percent more. The firm earns at least 49 percent more profit in the short run. There is no reduction in long-run profit. Data from worker
interviews show the gains are generated by workers who know the opportunity costs of their tasks. Workers who know adjust when opportunity costs change. Workers who do not know do not adjust. The firm’s other behaviors, including their use of informal incentives, all reffect attempts to rectify the incentive problem. Only the use of the customary contract does not. The firm reverted to the customary contract after the experiment, even after acknowledging the alternative’s superiority. The paper explains how reversion implies the firm weighs material costs more heavily than benefits. It then uses data from interviews with owners of other restaurants to show that owners are in fact loss averse around a reference point of 0, and explains how loss aversion justifies reversion to the customary contract. It also shows owners with more experience
are more loss averse. Taken together the results suggest experience, via loss aversion, compels contractual choices which are materially inefficient