Health Policy$ense

Taking the Science of Financial Incentive Programs to a New Level: An Interview With Scott Halpern

Last week, Scott Halpern, Kevin Volpp and colleagues published a groundbreaking study in the NEJM demonstrating that financial incentives work in helping employees quit smoking, and that the design of the incentives matters. The article prompted widespread media coverage, as did the announcement by study partner CVS Health that, next month, it will launch a financial incentive program for all its employees who use smoke or use tobacco of any kind. You can read an excellent summary of the study by Aaron Carroll on The Incidental Economist, and watch Dr. Volpp explain the study on Knowledge@Wharton

I posed a few questions to Dr. Halpern about design and implementation of the study, as well as its real-world implications. His responses follow.

Why did you choose the specific forms of financial incentives to test?  

We were particularly interested in addressing two key questions regarding the design of optimal incentive programs: first, whether approaches that require people to commit up front, or have some skin in the game, truly work better than approaches that provide pure rewards; and second, whether group-oriented programs are more effective in promoting behavior change than individual-oriented programs. There are strong conceptual bases to support these two hypotheses - the insights that people are loss averse suggest the superiority of commitment contracts, and the findings that people are strongly motivated by social comparisons suggest the superiority of group-oriented incentives. However, neither of these established ideas had ever been put to a rigorous test in a study this large. As an example, thousands of people make commitment contracts on StickK.com each year, and the reported results are quite promising. However, the people who choose to make such commitments are almost certainly different from those who don't, so the reported successes say virtually nothing about the overall effectiveness of such programs. We wanted to take the science of incentive program design to a new level by disentangling how acceptable programs are, and how efficacious they are.

How did you decide on $800 as an expected value, and $150 as deposit contract?  Might a smaller deposit have gained larger acceptance to the more efficacious incentive?

We chose the $800 reward to provide a roughly equivalent intervention, after adjusting for inflation, to that shown by Dr. Volpp and colleagues to roughly triple quit rates in a 2009 trial among General Electric employees. In truth, there's no magic to this amount, and we certainly could have chosen a higher incentive while still having the programs be cost saving for employers. But at the same time, we don't know of any employers that are currently offering or thinking of offering larger incentives than this, and we wanted our programs to reflect real-world approaches. We chose the $150 deposit as an initial effort to balance acceptance of the programs (which would presumably be higher with lower deposits) and efficacy of the programs (which might be lower with lower deposits). In retrospect, I think we underestimated just how averse working people would be to deposit $150. Thus, based on our findings, we advised CVS Health to implement a deposit program using a smaller deposit to achieve greater acceptance, hopefully without losing too much efficacy because people will still have some skin in the game.

How did you (or CVS) deal with possible resentment from non-smoking employees about offering financial rewards only to smokers?

People often claim that providing incentives for people who partake in unhealthy behaviors disadvantages those who avoid those same behaviors for free. But the truth is that in the context of a large employer like CVS, people making healthier behavioral choices are already subsidizing those making less healthy choices through higher insurance premiums. So everyone - smokers and non-smokers alike - has aligned incentives to curb smoking rates among a company's workforce. Perhaps for this or other reasons, we did not observe any true resentment from non-smoking employees about our offerings to smokers.

The CVS workforce median income is fairly high at $60,000. Given the disproportionate burden of smoking among low-SES populations, how might your results differ among lower-income employees?  Would you design the incentives differently for low-SES groups?

This is indeed a key question given the growing gap in healthy behavior patterns among higher and lower-income individuals. And this is one of several key subgroups analyses we'll be focusing on in a follow-up manuscript. What we've found is that overall, lower-income people quit at considerably lower rates than higher-income employees, regardless of what type of incentive program you offer them. This was true almost entirely because lower-income people were less likely to accept the assigned programs in the first place. At the same time, we found that there was no different in the relative effectiveness of the programs among higher or lower-income individuals; lower-income people just had lower total quit rates across the board. This is a bit surprising from a traditional economic perspective because one might imagine that for lower-income people, $800 should provide a more substantive incentive than for higher-income persons. But clearly, for these approaches to work among poorer people will require efforts to make up-front engagement with the programs easier. This is exactly what CVS is banking on by requiring smaller but still substantive deposits.