The Journal of Economic Perspectives is a truly wonderful thing. Halfway between an annual review and a general interest journal, JEP provides up to date literature reviews intended for a broad audience and often with a minimum of mathematical frippery. The most recent issue has some fascinating pieces, including a pair about genetics and economics and one that rates to be an instant hit in the policy blogosphere by Diamond and Saez* on the case for a progressive income tax. Most interesting to me, however, is a short and very readable piece by Uri Gneezy, Stephan Meier and Pedro Rey-Biel titled: “When and Why Incentives (Don’t) Work to Modify Behavior.”

Casual readers of economics blogs or pop economics (e.g. Freakonomics) are probably familiar with such banal statements as “Incentives Matter” or “People Respond to Incentives.” Greg Mankiw makes the later one of his 10 Principles of Economics in his best-selling textbook of the same name, and the stand-up economist Yoram Bauman has a field day with it. The problem with this principle isn’t the text itself – people do respond to incentives – but the naivete with which we often assume people respond. Specifically, we tend to assume (based on tried and true rational economic actor models) that if you pay people more to do something than they were getting before, they will do more of it, and if you tax something or raise its price some other way, people will do it less.

Of course, that does govern an awful lot of behavior, especially behavior relevant to most of traditional economics: when the cost of beef goes up, we eat less of it, and so on. But what happens when this economics logic expands outward to areas of life that have traditionally been somewhat separated from direct pricing? It turns out the story is a lot messier. Sometimes people do what you’d expect, and sometimes they do the opposite: they do less of the thing you were paying them to do, or more of the thing you were paying them to avoid doing.

A classic example discussed in Gneezy et al (2011) concerns an Israeli childcare provider that was attempting to reduce the number of late pickups of children. Initially, there was no specific punishment attached to picking up children late, simply an admonition not to do so. The provider instituted a small fee (about 3$ US equivalent). What happened? Parents were more likely to pick their kids up late. Why? It’s not 100% obvious, but the review essay (and the underlying research) offer a few hypothesis that work more generally: one concerns tradeoffs between intrinsic and extrinsic motivations, another concerns audience effects, and a third concerns information. In this case, for example, the parents might have thought the true cost to the provider was much higher than the small fee and thus on learning that it only cost 3$, they adjusted their behavior accordingly – it wasn’t really that big a burden, so I’ll worry less about imposing it.

The tradeoff between intrinsic and extrinsic motivation shows up frequently in studies where students are paid for their performance or attendance in school. Depending on the circumstances, the size of payments, and what they are tied to (attendance, completion of assignments, grades, etc.) the effect can vary substantially. Usually, if the payments are large enough, the short run effect is as expected: kids do more work, show up more often, etc. But, worryingly, these short run effects are sometimes outweighed by long-run negative effects on intrinsic motivation, especially if the incentive program is eliminated (or simply stops after a certain grade).

The conclusion offers a nice summary, and hints at the complicated ways economists are coming to grips with some of sociology’s core interests (norms, trust, and the pervasive and sometimes contradictory consequences of rationalization and monetization):

When explicit incentives seek to change behavior in areas like education, contributions to public goods, and forming habits, a potential conflict arises between the direct extrinsic effect of the incentives and how these incentives can crowd intrinsic motivations in the short run and the long run. In education, such incentives seem to have moderate success when the incentives are well-specified and well-targeted (“read these books” rather than “read books”), although the jury is still out regarding the long-term success of these incentive programs. In encouraging contributions to public goods, one must be very careful when designing the incentives to prevent adverse changes in social norms, image concerns, or trust. (206)

Rather than review the rest of the paper, I’ll simply suggest that you check it out (it’s <20 pages, and easy to skim). The next time someone asserts that people respond to incentives, especially in a traditionally non-monetized context, you can readily agree – and then ask, why exactly do you think they'll respond the way you want them to?

* Saez is probably most famous right now for his work on historical trends in the incomes of the top 1%, research which I think heavily influenced the framing of the "We are the 99%" folks.