Bringing Ulysses to Scale: A Tale of Persistence, Spillovers and Customer Loyalty

Financial incentive programs are an increasingly attractive way to improve health behaviors. While research from behavioral economics shows these interventions effectively improve a specific health behavior while in place, less is known about the extended effects for both the participants and the host organizations. Understanding these effects is essential to bringing behavioral science insights to scale. If an intervention has positive effects on a targeted behavior, but leads to negative consequences once the program ends, or shows negative spillover effects in other domains, it may be unsustainable.

We address these questions by examining the extended effects of a penalty-based intervention: a voluntary 6-month commitment contract that significantly improved healthy grocery purchases (Schwartz et al. 2014). Households receiving a 25% discount on healthy grocery purchases, put their discount on the line by precommitting to a 5-percentage-point increase above their household healthy grocery baseline. Those who met the goal kept their discount; those who did not forfeited it.

Since this precommitment was run within a comprehensive health rewards program, we could examine the commitment device’s impact beyond the specific time period and targeted behavior (nutrition). Specifically, we examined the penalty effect’s persistence, whether it lead to negative positive or negative spillover effects (e.g., healthier nutrition = less exercise) and, finally, the impact on customer loyalty and engagement.

Persistence: Precommitted households continued to buy healthier groceries during each of the 6-months in the post-intervention period (β = 2.84, SE = .95, p < .01). Persistence was strongest for the most loyal customers, as well as those who had the biggest discount on the line, thus demonstrating that penalty-based incentives can be catalysts to promoting long-term healthy change.

Spillovers: We see no evidence of negative spillover effects. Committed households exercised slightly more (though not significantly) than control households (β= 07, SE=.05, n.s.). Interestingly, there was a significant positive effect on exercise when examining participants who had put the most money on the line during the commitment period, (β=.23, SE=.07, p<.01). Our results reinforce the value of incentives as a tool for improving behavior, by suggesting that they do not lead to negative consequences in other domains, and may even lead to positive spillovers for some.

Loyalty: A penalty-only commitment contract may improve targeted health behaviors, but at the cost of customer loyalty—a trade-off many firms are unwilling to make. We test the effect of the intervention on customer loyalty by examining involvement with the health rewards program after the commitment contract ended. Committed households showed a positive (though non-significant) change in loyalty status (our measure of involvement) during post-intervention year (β=.07, SE=.12,n.s.). This is an important finding, because most committed households forfeited their discount at least once, which could have resulted in backlash and firm dissatisfaction.

These results highlight the scalable potential to implement behavioral science insights, including loss aversion, and alleviate some concerns over negative spillovers and consumer backlash.