Development that Works
  • About

    This blog highlights effective ideas in the fight against poverty and exclusion, and analyzes the impact of development projects in Latin America and the Caribbean.
  • Incentives in education



    Surveys and Impact Evaluation

    Participants and researchers interested on education held a session to discuss evidence on incentive schemes in education at the International workshop on Surveys and Impact Evaluation (#Taller_EEIPP) workshop taking place in Santiago.

    A fundamental question in education is how to introduce incentives for teachers to improve student outcomes. The teacher is the characteristic on the supply side that impacts more the quality of education (Vegas and Petrow, 2008).

    The evidence from the three studies we discussed shows that incentive schemes improve outcomes but are sensitive to the design and the context in which they are implemented.

    In the first section, Francisco Gallego from Chile´s Catholic University presented his work on the evaluation of the National System to Evaluate Performance (Sistema Nacional de Evaluación del Desempeño) which operates in Chile since 1996.

    The program introduces a group tournament scheme where 25% of the best schools earn a 2 year bonus equivalent to about 5% of its annual income. The index to determine rank is a weighted average that includes standardized test scores at levels and changes, enrollment and others.

    Using a regression discontinuity approach, he finds that this scheme improves the probability of graduation in 0.05 standard deviations. The effects are heterogeneous with a largest effect for schools with a smaller number of students, low socioeconomic status and poor performers before the program.

    In the second section Leonardo Costa from the World Bank presented evidence on the evaluation of the Bonus to School Performance (Bonus de Desempenho Escolar) in Brazil implemented in 2008.

    This individual piece rate incentive is directed to teachers in grades 4th, 8th and 11th which teach mathematics and Portuguese. Those teachers that achieve at least 50% of the goal based on a quality index receive a bonus.

    The index is composed by test scores and promotion rates. The effects of the program were estimated by comparing changes in results in schools in the state of Pernambuco where the program was offered to schools in the northeast of the country.

    Evidence shows that schools respond to incentives but these seem to decrease in time. Incentives do not seem to work in large schools. Students benefited the most are those with a low socioeconomic status and with poor performance before implementation.

    In the third session Venkatesh Sundararaman from the World Bank presented results on an evaluation in which he worked in India on individual and group piece rate incentive scheme implemented in 2005 for rural primary schools. The group incentive is a bonus proportional to the school average percentage improvement above 5% for mathematics and language.

    For the individual scheme, the bonus is calculated as a function of the average test scores for the group the teacher teaches. The evaluation shows that two years later the program improves student performance. There is no evidence on adverse consequences such as teaching to the test. Individual incentives always show better results than group incentives for several school sizes.

    The sample is not very homogeneous in size and therefore differences are not always significant. He and his coauthor do not find significant differences in impact magnitude across several demographic student characteristics.

    The three studies show that incentive schemes can improve the quality of education. Improvements depend on design and context. Some aspects in incentive design that influence on results include the magnitude of the bonus, whether the bonus is based on individual or group results, whether the bonus is given in a tournament or piece rate scheme, the formula, etc.

    The context includes population characteristics in which the program is implemented such as school size and the socioeconomic composition of the students. The next question to pursue is how these factors affect its effectiveness and how to better design a tailored formula for a given population.

    If you want more information on the topic, I suggest you start with any of these papers where other aspects are discussed and where you will find multiple references to other studies on the topic.

    Comment on the post