For years, Zappos—an Internet shoe and clothing retailer—has been offering its employees money to quit their job. From the initial $100, this one-time offer has now reached $2000. The stated purpose of the policy is to select enthusiastic employees, in the hope the other types will leave with the money. Shakespeare’s Henry V and Napoleon used similar screening mechanisms, offering their armies a passport and ‘crowns for convoy’ or a free shot if they wished to leave or even kill their leader.

In a recent paper, ‘Reactive Incentives: Harnessing the Impact of Sunk Opportunity Costs’, we explore a different effect of such temptations. Rather than simply screening ‘good’ types, an offer to quit may boost resolve and increase subsequent performance among those who stay. To test this proposition, we ran a field experiment in an Amsterdam gym from September 5th, 2015 to July 15th, 2017. In the experiment, new subscribers were randomly assigned to either one of several ‘temptation’ treatments—and made a varying one-time monetary offer to quit—or to a control group, who received no offer.  

None of our subjects accepted the offer to quit—hence we had no selection effect to deal with—but subjects who were offered a moderate amount of money (a full refund plus a cash payment equal to 10% of their initial subscription price) to quit were statistically more likely to visit the gym, to resubscribe, and to earn higher post-treatment grades—most of them were students.

We call these and similar ‘temptations’ reactive carrots: ordinary, active carrots are payments made in the hope that the recipient will act in order to receive them. A reactive carrot, instead, is a payment that is designed to be refused and to induce the opposite behavior. So, an offer to quit is made in order to induce more, not less, gym attendance.

Even though this foregone opportunity is sunk, it nonetheless affects our subjects’ future behavior.  It is well-known that sunk costs matter, but we show that sunk opportunity costs also can change subsequent decisions.  We investigate several competing rationales for the incentive effect of reactive carrots. Two theories are consistent with our findings. Resisting temptations might (1) ‘self-signal’ information about an individual’s own resolve or (2) create a prospective risk of cognitive dissonance.

Under the self-signaling theory, individuals who resist a temptation learn something about themselves; in particular, they learn that they were able to resist the temptation. In the future, individuals have imperfect recall about their resolve (why they resisted the temptation) but recall their behavior (that they resisted the temptation), so that resisting a temptation builds a lasting positive self-image and hence boosts one’s self-confidence and resolve to act accordingly in the future.

Under the cognitive dissonance theory, an individual who resisted a temptation creates a mental yardstick of behavior. Future actions that are inconsistent with that benchmark become psychologically difficult to rationalize and, in turn, somewhat costly. Since not going to the gym on a regular day is inconsistent with having rejected an offer to do so for money on the treatment day, the individual is more likely to attend the gym in the future.

We also explore the possibility of reactive sticks, that is, the positive effect of having overcome a hurdle to go to the gym which in turn boosts future attendance. This time the experiment was done by Nature, not by us. We checked whether new subscribers who were exposed to bad weather during their first week were more likely to go to the gym than their luckier peers who had ‘good’—according to local standards—weather.

Reactive incentives can be found in many regulatory and contractual settings. Money back guarantees may boost customer satisfaction with products by offering customers a ‘temptation’ to return the good within a certain time-window. As our gym members, those who resist the temptation and keep the good may be less likely to report problems and more likely to purchase more of the same product in the future. Similarly, early termination clauses in relational contracts may boost performance among those who resist the temptation to exit early.

Many addiction-treatment programs and other programs designed to help people achieve certain goals, such as to quit smoking, succeed in the short term but are prone to high chances of relapse. Offering people who have enrolled in a smoking-cessation program a sum of money to smoke a cigarette in front of their peers may result in a boost of resilience to future temptations and consequent long-term success.

After the financial crisis, the Consumer Financial Protection Bureau, prohibited prepayment penalties on most residential mortgages. That might partially backfire. If lenders were to offer two contracts—one with a prepayment penalty and lower rates and a second without the penalty but higher rates—borrowers who resist the temptation to borrow at lower rates (and accept the penalty term) may actually end up repaying their mortgage faster than a control group who was only offered one contract without the prepayment penalty. Our analysis suggests that if lenders were required to offer both contracts rather than prohibited from offering the one with pre-payment penalties, repayment rates may increase.

Some voters have to go through often cumbersome registration procedures, which can be seen as a reactive stick. Those who endure the inconvenience may be more likely to vote later. Making voter-registration easier may increase registration but reduce voting.

Shareholders often receive offers to purchase their shares, for instance, in a share repurchase campaign or a hostile takeover. Our experiment suggests that those who resist the temptation to sell may be more likely to hold on to their shares in the future as compared to those who did not resist the temptation.

If a pop-up had given you a choice between reading this post and accepting a sum of money not to do so—an offer we hope you would have refused—would you now be more likely to read the next post in this blog?

Ian Ayres is William K. Townsend Professor of Law at Yale Law School.

Giuseppe Dari-Mattiacci is the Alfred W. Bressler Professor of Law at Columbia Law School.