Association for Public Policy Analysis & Management

The benefits and costs of grade retention policies have been widely debated. Promoters claim that it can benefit students by allowing them to restudy the material. Critics claim that it could negatively affect student’s motivation. In 2002, the Colombian government introduced a school-level policy that limited the share of students that could fail a grade. According to this reform, each year the number of failed students could not be more than 5% of the total students in each school. In the schools that exceeded this limit, some students were promoted to the next grade level with their peers, regardless of their performance. Later in 2009, this cap was removed and schools did not have any constraints on how many students could be grade retained.

In this paper, I assess the effect of the introduction and the discontinuation of this policy on educational outcomes such as grade failure, dropout and future academic achievement. The characteristics of the Colombian case have important implications for educational effectiveness. The existence of this cap is encouraging schools to practice social promotion but in a less extreme way than prohibiting repetition altogether. Also, from the point of view of the students, this is a decrease in the likelihood of failing a grade as opposed to the entire elimination of the threat of grade retention. The contribution of this paper is to provide evidence of the consequences of moderate grade failure policies on schooling outcomes, especially since implementing radical policies such as automatic promotion could sometimes be politically unfeasible.

I use administrative records from an annual census of urban schools from 1998-2012. I link this data with students' test scores from a secondary school exit exam for 13 cohorts that graduated between 2000-2012. For the empirical strategy, I exploit the variation in the years of exposure across cohorts and the bindingness of the policy across schools. Plausibly, schools affected by this policy were those with historic failure rates higher than 5%. To deal with potential sources of endogeneity, I estimate a fixed effects model that controls for school fixed effects, year fixed effects and school-specific linear trends. I perform robustness checks that test for mean reversion and for differences in schools before the policy was in place.

I find that the introduction of this policy is associated with a decrease in grade failure and dropout. In addition, I find evidence of reductions in language test scores for cohorts that were affected through their primary education. This suggests that the effect stems from more exposure to the policy during early years of instruction. Given the results on dropout, I argue that the effects on test scores could be driven by two mechanisms. First, there is a composition effect as more low achieving students are taking the exam since schools with binding caps are retaining less. Also, there is an incentive effect as less effort is exerted by all students now that the threat of grade retention is lower in schools with binding caps.