Using Incentives and Commitments to Overcome Self-Control Problems with Exercise: Evidence from a Workplace Field Experiment

Can financial-incentive programs help people improve behavior in health settings where self-control problems are important? We report the results of a large-scale field experiment with employees of a Fortune 500 company on the effect of financial incentives at encouraging exercise. The treatment group for the experiment received a onemonth incentive program, during which time they could earn $10 per visit to the on-site company gym for up to 3 visits per week. A secondary treatment offered half of the incentivized group at the end of the incentive period the opportunity to create a self-funded commitment contract for continued exercise at the company gym. We find that the temporary incentive led to lasting increases in exercise even after the incentive was removed. The availability of a commitment contract substantially improved the long-run effects of the incentive program. We discuss the implications of these findings for health-incentive programs and our understanding of time inconsistency and commitment. Preliminary and Incomplete: Please do not circulate 1 Can financial-incentive programs help people improve behavior in health settings where selfcontrol problems are important? This paper addresses this question in the context of exercise. There is strong empirical evidence that people exercise less often than they intend (DellaVigna and Malmendier, 2006) and these struggles are a common motivating example for models of time inconsistency (Laibson 1997, O’Donoghue and Rabin, 1999 and 2001). Low levels of exercise are also an important policy issue. The Centers for Disease Control estimates that as of 2009 approximately one third of adults engage in no regular exercise, and this lack of physical activity likely contributes to poor population health and rising health-care costs. In the face of spiraling health costs, policy makers, employers and insurance companies are increasingly interested in using incentive programs to address a range of problematic health behaviors, including exercise. Approximately 40 percent of large employers in the US offer some form of incentive-based wellness program (Hewitt Associates) and recent health-care legislation in the U.S. contains language explicitly authorizing firms to spend up to 30% of the cost of employee insurance on wellness programs. While employers and insurance companies are primarily motivated to offer these programs in attempt to deal with the externality imposed on their bottom line by the poor health behaviors of individuals, This study reports the results of a large-scale field experiment examining the effect of financial incentives to encourage exercise among employees at the headquarters of a Fortune 500 company. The treatment group for the experiment received a one-month incentive program, during which time they could earn $10 per visit to the on-site company gym for up to 3 visits per week. A secondary treatment offered half of the incentivized group at the end of the treatment period the opportunity to incentivize themselves by creating a self-funded commitment contract for continued exercise at the company gym. In these contracts, employees put money on the line that was forfeited to charity if they models of time inconsistency suggest that incentives may interact with self-control problems in important ways that could improve individual welfare in addition to social welfare. 1 Regular exercise contributes to a healthy lifestyle, with a wide range of benefits including reductions in heart disease, obesity, diabetes, and cancer (Pate et al. 1995). 2 Baicker et al. (2010) provide a review of workplace wellness programs and estimate (cross-sectionally) that on average firms see returns of around $5 for every dollar invested in wellness programs. Preliminary and Incomplete: Please do not circulate 2 failed to regularly use the gym. The primary outcomes of interest for the experiment are a) whether the temporary incentive program had lasting effects on exercise at the company gym and b) whether the degree of any such habit formation was stronger for the group offered commitment contracts in addition to their incentives. Models of time inconsistency suggest a number of reasons that a temporary incentive program could have a lasting effect on behavior. First, there are large start-up costs to beginning an exercise routine, including deciding on the level and type of exercise, gaining familiarity with a gym or other exercise setting, and changing existing daily routines to make time for exercise. In the face of these start-up costs, time-inconsistent people may procrastinate on starting to exercise even though they think the long-run benefits of exercise exceed the costs. Second, the physiology of exercise is such that exercise typically becomes more pleasurable the more one exercises, and as such a person may need to go through an investment period before reaching a steady state where the daily costs of exercise are not too high. Individuals with present-bias may find it particularly difficult to motivate themselves to exercise through this investment period. The incentive program, with its month-long per-use incentives and fixed start date that precludes procrastination, could be expected to help overcome both of these issues. On the other hand, even after establishing an initial exercise routine, people with present bias are more likely on a daily basis to be tempted to avoid the immediate costs of exercising. One-time incentives are unlikely to help with that daily self-control struggle. The availability of a commitment contract, though, may be an effective way to deal with this problem for those who are sophisticated about and self-aware of their time inconsistency. During the 4 weeks that the incentive program was in place, the incentives led to approximately a doubling of the use of the company gym by those in the treatment group relative to the control group. The subjects in this experiment are working adults in the Midwest, with an average age of 40 and self-reported rates of overweight and obesity (69% combined) matching the national average reported by the Centers for Disease Control. As such, while it is perhaps not surprising to economists that people exercise more while paid to do so, these results nonetheless provide a useful benchmark for understanding how this policy-relevant population responds to incentives for exercise. The treatment groups continued to use the gym at higher rates than the control group even after the incentive program ended. Among those offered incentives only there was a modest, Preliminary and Incomplete: Please do not circulate 3 statistically-significant increase of 3.5 percentage points in the fraction of employees regularly exercising at the company gym in the weeks after the incentive program ended (a 16% increase relative to the baseline of 22% regularly exercising in the control group). These results relate most closely to seminal work by Charness and Gneezy (2009) and a recent working paper by Acland and Levy (2010). The effect for this group largely dissipates by two months after the program. The post-incentive effects were much stronger for the treatment group offered the ability to create a self-funded commitment contract at the end of the incentive program. On average this group saw an increase of 10 percentage points (45%) relative to control in the fraction regularly exercising over the first two months after the incentive program ended. This long-run effect also dissipates more slowly than for the group only offered incentives.