Economists have long noted that the price mechanism can be effective at modifying human behavior. Psychologists classify this aspect of behavior motivation as extrinsic motivation, meaning that the behavior is induced by external pressure. If I increase my hours worked due to an overtime premium then I can be said to exhibit extrinsic motivation - I am responding to the price schedule offered me. In contrast to extrinsic motivation, psychologists posit intrinsic motivation as arising from within the individual. If the overtime premium is not needed for me to put in maximal hours because of my enjoyment or satisfaction in completing the job then I am intrinsically motivated to work long hours.
These long-standing concepts are often attached to discussions of public sector pay reform and, more recently, results-based financing in the health sector . A common concern voiced in discussions of public sector pay reform and pay-for-performance in the health system is that incentivizing the provision of care at the provider level will result in less care or lower-quality care as the extrinsic motivation from the incentives “crowds out” the intrinsic motivation to do good or be respected by peers. To the extent that intrinsic and extrinsic motivations co-exist and respond to changes in the regulatory environment such as varying incentive levels, this is largely an empirical question.
There is much heat in these discussions but little (although emerging) direct evidence, especially evidence from impact evaluations. I’ll return to this topic in future posts with a focus on how we might measure individual motivation, and what specific evidence exists as it relates to health workers. However there are some “classic” studies on this general topic published within the last 15 years that raise suggestive questions. (Frey and Jegen provide an excellent review of the early literature ).
A well-known example from Lazear  finds that switching from salary to piece-rate at an auto-glass installation firm increased average worker productivity by 44% - half of this gain was due to unproductive workers leaving the firm and the other half due to a direct productivity effect of the incentives. On net there was no “crowding out” effect of intrinsic motivation in the switch to piece rate, at least for the workers that remained with the firm. However I do wonder about the degree of intrinsic motivation for this worker population at the start of the study – if there is a low degree at baseline then there is not much intrinsic motivation to “crowd out” in the first place.
In contrast to Lazear’s study, readers may be familiar with two studies by Gneezy and Rustachini that experimentally manipulated: (i) the compensation paid to high school students  to solicit donations for (a widely-perceived) socially worthy charity and (ii) the fines paid by parents  who are late to pick up their children from daycare.
In this first study students were randomly assigned to one of three groups. The first group did not receive any compensation for collecting for charity (but did receive a motivational speech), the second group was allowed to keep 1% of collections for themselves, while the third was allowed to keep 10% of collections. When looking at the two groups that received a monetary incentive, as expected, higher payment results in higher total donations collected. The high incentive group collected an average of 220 Shekels while the 1% incentive group collected 155 Shekels. However the group that actually collected the most donations of all was the group not compensated at all for their efforts. This group collected 240 Shekels.
In regards to the parents who pick up their children at daycare, the control group did not have to pay a fine if they were late while the treatment group was charged a monetary penalty. After the fine schedule was introduced, the late arrival rate of treatment parents doubled. The fines transformed the parental motivation to arrive on time and actually resulted in an increase of sanctioned behavior. Perhaps the fines absolved the parents of a sense of obligation now that they felt teachers will be compensated for the disutility of having to stay late.
In a study of the siting of a nuclear waste depot  in Switzerland, Frey and Oberholzer-Gee find that 50% of respondents in two communities targeted for possible facility location would accept such a facility in their community. However when the same respondents were then offered cash compensation for the siting of the facility, only 25% would accept the facility placement in their community. Furthermore, the value of the randomized offer of compensation did not affect the likelihood of acceptance.
The three studies above clearly suggest that the introduction of monetary incentives can affect behavior in a contrary way to the intention of the policy maker.
An important distinction to bear in mind, however, is that in these last three examples (and unlike the Lazear example) all involve the introduction of a monetary incentive where one did not exist before. In these contexts, the motivation for and the interpretation of the now incentivized task may undergo a fairly extreme transformation in the minds of the subjects. Will this transformation occur in the minds of health workers offered a performance incentive? Remember these workers are already in a career for which they are compensated.
I tend to believe that it is quite demotivating to work in a remote clinic with broken equipment, intermittent salary payments, no easy access to clean water, and perhaps lackluster district supervision. Income is not merely the means to consume goods but also perceived by many as a marker of social worth. Thus I would not be surprised if the additional extrinsic motivation of a performance incentive complements, and not “crowds out”, any pre-standing intrinsic motivation. But the jury is out so let’s see what we learn.