It would seem natural that in a system where pay is linked to performance, better performers are rewarded more handsomely. Yet, there are plenty of examples where this does not work. If it is difficult to find a metric of performance, then the selected metric may give wrong incentives. The classic example is programmers paid by the number of code lines written, which leads to bloated and confusing code. Another one is with many civil servant jobs where the output is not related to a marketed good. The issue can be so bad that a negative correlation between performance and pay emergences.
Ola Kvaløy and Trond Olsen show that this could happen in another way that is not related to monetary and non-monetary rewards, yet has an impact on intrinsic motivation: weak enforcement probability. Suppose the latter is variable. Then high monetary rewards may be associated with a low probability that the scheme will actually be used. Workers may then just slack off. The above hypothesis is not completely insane, as one can imagine situations where there is moral hazard on the side of the worker, and the work contract tries provide the worker both with incentives and with some insurance. It works also without worker risk aversion if there is limited liability.
Ola Kvaløy and Trond Olsen show that this could happen in another way that is not related to monetary and non-monetary rewards, yet has an impact on intrinsic motivation: weak enforcement probability. Suppose the latter is variable. Then high monetary rewards may be associated with a low probability that the scheme will actually be used. Workers may then just slack off. The above hypothesis is not completely insane, as one can imagine situations where there is moral hazard on the side of the worker, and the work contract tries provide the worker both with incentives and with some insurance. It works also without worker risk aversion if there is limited liability.