By Linda Thunström, Ph.D.


Harvard professor Cass Sunstein defines nudges as “liberty preserving approaches that steer people in particular directions. ” These policy instruments have become so popular with law makers and organizations that they impact our daily lives, our lives tomorrow and occasionally even our afterlife: I was nudged today to choose a low calorie meal, I am nudged to save money each month, and I have been nudged to donate my organs when I die. It is easy to see the appeal of using nudges to change people’s behavior in socially desirable directions. They are often low in implementation cost, high in public support, and may result in substantial behavioral change. However, like other policy instruments, they might come with unexpected and undesired consequences.

Such adverse effects are observed in a recent body of studies that measures the behavioral and welfare impact of nudges. A study by Amy Wilson and her colleagues finds that information intended to encourage low-calorie milk consumption has the undesired effect of increasing both low- and high-calorie milk consumption. In another study, my collaborators and I at University of Wyoming and Copenhagen University find that calorie menu labels fail to encourage low calorie choices, given people may ignore the labels to avoid guilt from calorie consumption. Similarly, people have been found to avoid the ask to make charity donations. James Andreoni and his colleagues argue that a reason for such avoidance may be that the ask encourages people to sub-optimally donate. Amanda Agan and Sonja Starr examine the impact of withholding job applicants’ criminal records from human resources managers. They observe that doing so increases racial discrimination.

People differ in behavioral biases (“judgment errors”), socio-economic characteristics (which often affect preferences), knowledge and beliefs. The impact of nudges is therefore likely to differ across people. For instance, my colleagues and I find that spending reminder nudges entirely fail to affect those who would benefit from spending less  (“spendthrifts”), while it reduces spending by those who spend too little (“tightwads”). Further, Anders Anderson and David Robinson show that people who wrongly believe they are financially literate are the ones responding the most to a nudge that encourages people to take charge of their retirement savings. As a result, this group of consumers loses out on retirement savings compared to if they had stayed with the default savings option.

Jointly, these studies suggest that nudges may have adverse effects on consumers and their behavior, on average, or for subgroups. This is important knowledge, but we might want to take the analysis a step further, and measure the welfare impact of nudges. A few recent studies do exactly that – they measure welfare effects from informational nudges. Hunt Allcott and Judd Kessler show that home energy conservation reports increase welfare for most consumers, but has a negative effect on some. Mette Damgaard and Christina Gravert find that the welfare effects of charity donation reminders are substantially impacted by the psychological cost people experience from receiving the reminders. My own research suggests that calorie menu labeling positively affects the average consumer, but reduces welfare for some. Specifically, people with low eating self-control are harmed by the nudge – learning about the calorie content in food only makes them feel bad, without inducing behavioral change.

Policy makers are likely interested in who is benefited and who is harmed by a particular nudge. For instance, regressive policies (policies that hit low-income earners the hardest) are often unpopular, amongst law makers and citizens alike. There are other reasons too we might want to document distributional effects (differing impacts across people) of nudges. Modern technology offers up entirely new possibilities to diversify nudges across the population. With apps, web contracts and web based information, it is feasible to personalize nudges, such as default options for savings, health information nudges, and so on. Marketers have already figured this out, and have used this technology for decades. Retailers use purchase information from loyalty cards to personalize coupons and advertisements, algorithms lurk in the background of social media, connecting our interests with advertisements, and so on. Maybe this is the next step for policies: to be personalized, based on their welfare effects. Either way, understanding the incidence of nudges is important, and the research on this topic has only just started. Personally, I am convinced that I am better off from any default nudge that encourages savings, but stubbornly stick my head in the sand to avoid the nudge that informs me about the calorie content of my afternoon latte – I want to wholeheartedly enjoy my latte experience.


Linda Thunström
Linda Thunström is an assistant professor in the Department of Economics at University of Wyoming. She has a Ph.D. in economics and her research interests include behavioral economics and public economics.
Linda Thunström

Latest posts by Linda Thunström (see all)