The Effect of Payment Methods on Risk Aversion

Article excerpt


Management of the risk-return Wadeoff is a--or perhaps, the--central issue in the study of finance. Von Nenmann and Morgenstern (1944) initiated formalization of this area with expected utility theory. Within this framework the degree of curvature of the utility function expresses the individual's degree of risk aversion. Over the succeeding 60 odd years, expected utility theory has been extended in many directions in an effort to refine and generalize its outcomes. In recent years, risk aversion has been the subject of numerous field and laboratory experiments that, among other things, serve as tests of the oft-used risk-neutrality assumption (e.g. Binswanger 1980; Kachelmeier and Shehata 1992; Beetsma and Schotman 2001; Harrison et al. 2007). These studies predominantly have found participants to be moderately risk averse rather than risk neutral as assumed in some theoretical models. The validity of early risk aversion studies was questioned, as they involved hypothetical or quite small payoffs. To address some of these concerns, studies by Holt and Laury (2002, 2005) [hereafter HL] compared hypothetical and real, increasingly higher, payoff levels within a single study. They found that real, higher payoffs led to greater degrees of risk aversion and that most subjects exhibited decision consistency. As is common, the resolution of this question led, in turn, to other issues being raised. One of those was whether the method of payment used influences observed risk aversion levels (Laury 2006; Lee 2007). The results of our study, reported below, provide new evidence with regard to this question.

Within most studies of the impact of financial incentives on risk aversion, one of two payoff mechanisms has been used. Random round (RR) payoffs occur when a participant repeats several variations of an exercise after being told that one of the trials will be selected at random to determine the size of their payment for participating in the experiment. In the accumulated value payoff method (AV), the sum or the average of the decisions from the several trials determines the subject's payment. Lee (2007) surveys this literature and offers a meta-analysis of 43 studies suggesting that "[s]ince the number of studies using RR is small and hence there are no other studies in this review that use RR and AV for the same task, the effects of different repetition payoff mechanisms (especially RR vs. AV) are not exactly known." Thus, results from studies of risk attitudes conducted to date may or may not have been influenced by the payment mechanism that was used.

Our study was conducted using a novel methodology. We first trained participants and then had them play a non-violent, adventure-style video game. We obtained permission to modify a game that was originally designed to study NATO peacekeeping teams' behavior by BBN Technologies under contract to the U. S. military (BBN Technologies 2006). The system includes a data logging function that records all game behavior in a Sequel database and also provides a utility for statistical analysis of the data. We modified the NATO game as a first illustration (in a civilian rather than military context) of the use of this system for academic research. (1) In this study, we use the game to replicate the HL experiment for measuring risk aversion, but employ both RR and AV payment methods. This allowed us to shed light on one of the fundamental questions that has been asked about this body of research and to illustrate the potential of this promising new methodological tool. In the pages that follow, the HL experiments are explained in more detail, followed by a description of the experiments we conducted to replicate the HL results and validate our methodology. Some of the most notable outcomes of these experiments are then reviewed, with an emphasis on the impact of alternative payment methods. Finally, implications of the research and directions for future research are summarized in the concluding section of the paper. …