Paul Glewwe (Minnesota), Nauman Ilias (Competition Economics), and Michael Kremer (Harvard), in a new paper, Teacher Incentives (a non-technical summary is here), find evidence from Kenya that incentives for teachers, based on rewards for teachers for improved student test results, had pretty much no effect.

In 1997, a Dutch charity sponsored a two-year program of prizes for the top-scoring and most-improved schools based on 4th through 8th grade student scores on district exams in two Western Kenyan districts. Fifty schools were randomly selected out of the 100 that applied. To discourage schools from manipulating the pool of students taking the exams, students who did not take the exam were assigned low scores. Prizes were comparable to merit pay programs in the United States, ranging in value from 21 to 43 percent of teacher monthly salaries.

Students in the incentive program schools were more likely to take the exams and had higher test scores, primarily in geography, history, and Christian religion, the subjects most susceptible to gains from extra coaching and memorization. The authors conclude that "Teacher attendance did not improve, homework assignments did not increase, and pedagogy did not change. There is, however, evidence that teachers increased effort to raise short-run test scores by conducting more test preparation sessions." The test score gains evaporated after the end of the incentive program.