Hunt Allcott: Taxing Our "Sins" at the Frontier of Behavioral Public Economics

Hunt Allcott is a Senior Principal Researcher at Microsoft Research visiting the Economics Department at Harvard University in 2020-2021. An applied microeconomist who studies topics in behavioral economics, environmental economics, public economics, and industrial organization, Prof. Allcott is also a Co-Editor of the Journal of Public Economics and a Scientific Director of ideas42, a think tank that applies insights from psychology and economics to business and policy design problems. 

Prof. Allcott is a leading scholar in the emerging field of behavioral public economics, which asks questions such as: How can we do welfare analysis if choice does not necessarily identify utility? How do we empirically measure consumer bias? How do we set socially optimal policies in the presence of bias? Are nudges a good idea? 

As Prof. Allcott explains in detail, the standard approach to policymaking relies on the “revealed preferences'' assumption – people always choose what is best for them – which implies that interfering with consumers’ decisions always reduces welfare. In practice, this assumption often does not hold. Behavioral economics extends the study of public economics in three ways: 1) new welfare implications of standard policies; 2) new policy tools like nudges; and 3) better predictions about how people respond to policy. 

One of Prof. Allcott’s recent notable studies is on payday loans, a type of short-term, high-interest loan that is typically due on the borrower’s next payday. Critics of payday lending argue that the practice is predatory and can trap vulnerable borrowers in debt they can’t afford, and a number of laws have been passed over the years to regulate the practice. In Prof. Allcott’s study, he seeks to answer the question of whether or not borrowers act in their own best interest (in which case, restricting payday lending would decrease welfare). 

He found that the most inexperienced borrowers underestimate their likelihood of borrowing again, whereas more experienced borrowers are accurate in their prediction of future borrowing – sophistication (the ability to predict future preferences) is gained with experience. Therefore, the truth is likely to be in the middle – that neither an outright ban on payday lending nor an utter lack of regulation would improve consumer welfare. How can policymakers take into account this heterogeneity when developing policy? Is it even possible or optimal to do so?  

Another fascinating experiment evaluates behaviorally motivated policy (Allcott and Taubinsky 2015). There are tremendous financial costs saved if one uses CFL (compact fluorescent lamp) light bulbs instead of incandescent ones, but many still purchase a large amount of incandescent ones. What explains this? Does it mean that consumers are biased and don’t know what’s best for themselves? How would one quantify the alleged bias in units of dollars? 

This is where the topic of “sin tax” comes into play. To impose any sin tax or regulation, it is implicitly assumed that “consumers are unable to make market decisions that yield personal savings, that the regulator is able to identify these consumer mistakes, and that the regulator should correct economic harm that people do to themselves” (Ted Gayer 2011). A key point in Prof. Allcott’s work, however, is that behavioral public economists try to “debias” imperfectly informed or inattentive consumers; they’re not trying to convince them to buy one kind of product or the other, hence minimizing the “demand effects.”

As Prof. Allcott explains in detail, behavioral public economists try to avoid a counting estimate of what one should be willing to pay for something. Rather, they try to elicit a demand curve and allow that revealed preference demand curve to show all kinds of normatively valid preferences that one may have between different goods. Then, they focus on one specific modification to the demand curve that arises from a bias that one identified. The meta-point is that they’re respecting revealed preferences up to specific biases that they measure. 

We end the interview with Prof. Alloctt’s recent work “The Welfare Effects of Social Media,” in which he and Stanford economist Matthew Gentzkow and others attempt to show how Facebook can impact well-being negatively, both socially and individually. Surveyed subjects were more likely to reduce their awareness of and attention to politics and be less polarized when they’re cut off from social media. It is interesting to note that participants were being paid to temporarily leave Facebook, and many asked for high prices that wouldn’t be sustainable in the long run. 

Hunt Allcott