1. This has always struck me as a potentially important critique of standard economic assumptions. The ever brilliant Chris Dillow notes this is a problem with using market mechanisms. As he says
This problem is exacerbated by another: motivation crowding out. Who is likely to be the better probation officer: the one drawn to the profession by a desire to rehabilitate offenders; or one who will earn a little more for hitting a contractual target? Who is likely to better look after vulnerable children: someone attracted to work in childrens' homes by a love of children; or one working for a profit-maximizing private equity firm? If the cash nexus comes to dominate, other motives such as professional pride recede not just because people change but because those with strong professional ethics simply leave the job.
2. The paper he links to is Bowles and Reyes. They nicely use a Lucas critique argument
"Here we extend the logic of the Lucas Critique to questions of framing, motivations, and social norms, in short, to preferences. To do this we modify the standard public economics and mechanism design assumption that taxes, subsidies, and other explicit incentives affect behavior only indirectly, that is by altering the economic costs and benefits of the targeted activities. In this conventional approach explicit incentives thus do not appear directly in the citizen's utility function and as a result, the behavioral effects of explicit economic incentives and social preferences are separable, the effects of each being independent of the levels of the other. We modify the citizen’s utility function so that this separability property need not hold and as a result the two kinds of motivations may be either complements -- social preferences being heightened by incentives appealing to self-interest -- or substitutes, when explicit incentives are said to crowd out social preferences."
As they say there may be other reasons
Incentives may have counter-intuitive and counter productive effects for reasons other than non-separability (Seabright (2009)).
Strong monetary incentives, for example, may overmotivate an agent leading to greater than the optimal level of arousal posited by the so called Yerkes-Dodson law. This appears to be the mechanism underlying the negative effects of high incentives found in three experiments by Ariely, et al. (2005).
Similarly, if agents have an income target, monetary incentives may allow target attainment with less effort. Camerer, et al. (1997) suggest that this may explain why New York City taxi drivers work fewer hours when they are making more per hour.
and they add the possiblity that just setting a target might signal how hard the achievement is:
the target may also infer information about the person who designed the incentive, about his or her beliefs concerning the target, and the nature of the task to be done (Benabou and Tirole (2003), Fehr and Rockenbach (2003)).
3. Dillow's earlier blog post on Ronnie O'Sullivan is a fantastic summary.
