Commitment Devices: Nice or Mean Means

Introduction

Do people always act in their best interest? Standard neoclassical economics claims so. But everyday experience and the growing literature on behavioral economics point otherwise. There may be (more or less common) situations in which behavior is hard to explain by utility maximization of a rational agent.

One of these situations is time-inconsistency, i.e. the inability of an agent to stick to a plan, he once realized as optimal. E.g. an alcoholic may well decide in the morning that he would be better off, if he remained sober for the rest of his life. Yet in the evening having a drink may be just too tempting to resist. Everybody gives in to temptations once in a while and almost everybody can be watched doing ridiculous things like saving in Christmas clubs or locking refrigerator doors at night just to commit to spending or eating less. And opportunities to commit in one way or another are mushrooming up everywhere in society –ranging from disulfiram (“Antabuse”), a drug which causes sickness when combined with alcohol and is used in treating chronic alcoholism, to websites like “stickK.com” or “Aherk.com”, which are basically “self-blackmailing” services. People can define their goals and if they do not achieve them, they will lose either money or dignity (by having a compromising picture published on their Facebook-account).¹

But do people benefit from commitment devices like these? Or are there other, superior means to cope with issues of self-control?

A lot of research has been done especially on self-control problems, in which the agent knows what his best interest would be, but somehow cannot manage to act accordingly (inter alia Ainslie, 1992 and 2001; Elster, 1979 and 2000; Ross, 2005; see also e.g. O’Donoghue & Rabin, 1999 for work on the interrelated problem of procrastination). One possible explanation, which sticks to most of the rationality axioms of standard expected utility theory, is hyperbolic or quasi-hyperbolic discounting (Phelps & Pollak, 1968; Laibson, 1997). If the discount function of an agent is more convex than with exponential discounting, present utility is overweighed compared to future utility resulting in time inconsistency. Left alone this kind of behavioral assumption can easily lead to completely inconsiderate agents, who cannot exert even the slightest bit of willpower – a result as implausible as the perfectly controlled, rational homo oeconomicus. One way to explain willpower and self-control attempts nevertheless is to view a decision as precedent for future decisions. In this way “utility from anticipation” (Bénabou & Tirole, 2004; see also Caplin & Leahy, 2001; Carrillo, 2005; Kőszegi, 2010 for treatments in more general context) is relevant. If my choice today informs me about my future choices, resisting temptations becomes more compelling, as I anticipate resisting in the future as well (This paper draws on Bénabou & Tirole, 2004 mainly; see also Ainslie, 1992 for a slightly different approach).²

Another way to keep oneself from acting impulsively is using commitment devices. Anticipating self-control problems a rational agent is likely to “tie himself to the mast” to resist future temptations (see Brocas, Carrillo, & Dewatripont, 2004 and Bryan, Karlan, & Nelson, 2010 for reviews). These commitment devices may be as loose as a personal rule or as binding as “Antabuse”. By the very definition of a commitment commitment devices ease, if not eliminate self-control problems. But is their availability really advantageous to the people suffering from weakness of will? At first glance the answer is obviously yes.³ And empirical evidence mainly confirms this view. Probably the most famous reported evidence of people using commitment devices stems from Ariely and Wertenbroch (2002). They allowed students to choose their deadlines for delivering papers themselves. Contrary to the expected rational behavior of always choosing the latest possible date, students chose earlier dates, although they knew that not meeting these deadlines would be costly. Ariely and Wertenbroch (2002) interpret this behavior as strategically committing to start working early and interestingly students who did so performed better than those who did not.