Disappointment at doing worse than expected can be a powerful emotion. This emotion may be particularly intense when the disappointed agent exerted effort in competing for a prize, thus raising her expectation of winning. Furthermore, a rational agent who anticipates possible disappointment will optimize taking into account the expected disappointment arising from her choice.
In this paper we use a laboratory experiment to test whether agents are disappointment averse when they compete in a real effort tournament. In particular, we test whether our subjects are loss averse around reference points given by endogenous expectations which adjust to both an agent’s own effort choice and that of her rival. Pairs of subjects complete a novel computerized real effort task, called the “slider task”, which involves moving sliders across a screen. The First Mover completes the task, followed by the Second Mover, who observes the First Mover’s effort before choosing how hard to work. A money prize is awarded to one of the pair members based on the pair’s relative work efforts and some element of chance which we control. After each repetition, the subjects are re-paired. We impose probabilities of winning the prize which are linear in the difference in the agents’ efforts, so the marginal impact of a Second Mover’s effort on her probability of winning does not depend on the effort of the First Mover she is paired with. Therefore, if agents care only about money and their cost of effort, the Second Mover’s work effort should not depend on the effort of the First Mover. However, as predicted by our model of disappointment aversion, the experimental data show a discouragement effect: the Second Mover shies away from working hard when she observes that the First Mover has worked hard, and tends to work relatively hard when she observes that her competitor has put in low effort. Thus First and Second Movers’ efforts are strategic substitutes.
Our primary contribution is empirical. First, we offer evidence consistent with disappointment aversion from a reduced form linear random effects panel regression. More substantively, we exploit the richness of our experimental data set to estimate the parameters of a structural model of disappointment aversion using the Method of Simulated Moments. This allows us to estimate the strength of disappointment aversion on average and the extent of heterogeneity in disappointment aversion across the population. Goodness of fit analysis shows that the estimated model fits our data well.
Together with random variation in the monetary prize across pairs of subjects, the design of our slider task generates sufficient variation in behavior to enable us to estimate the structural parameters of our model of disappointment aversion. In particular, the slider task gives a finely gradated measure of performance over a short time scale. As the task takes only two minutes to complete, we can collect repeated observations of the same Second Movers facing different prizes and First Mover efforts, while the fineness of the performance measure allows us to observe accurately how Second Movers respond to different prizes and First Mover efforts. The resulting panel data permit precise quantification of the distribution of the cost of effort and the strength of disappointment aversion across agents in the population.
The formal model that we test is a natural extension of disappointment aversion to situations in which agents compete. Models of disappointment aversion (e.g., Bell, 1985; Loomes and Sugden, 1986; Delquié and Cillo, 2006; K?oszegi and Rabin, 2006, 2007) build on the idea that agents are sensitive to deviations from what they expected to receive; in particular, agents are loss averse around their expected payoff so losses relative to this expectation are more painful than equal-sized gains are pleasurable. We model expectations-based reference points as adjusting to an agent’s own effort choice and that of her rival: in the terminology of K?oszegi and Rabin (2007) they are choice-acclimating. The endogeneity of an agent’s reference point is crucial with loss aversion around a fixed reference point, even if given by a prior expectation, a Second Mover will continue to disregard First Mover effort.
Download
A Structural Analysis of Disappointment Aversion in a Real Effort Competition
