The effect of grade retention on secondary school dropout: Evidence from a natural experiment

Maria Ferreira Sequeda

Research output: Working paper / PreprintWorking paper

199 Downloads (Pure)


This paper analyses the effects of grade retention on secondary school dropout by evaluating a retention policy reform introduced in 2010 in Colombia. The reform ended the restriction that the annual number of retained students at a school could
not exceed 5 percent of the total school population. Using administrative data at the school level, we estimate a difference-in-differences model that exploits variation in schools’ retention rates before and after the reform. We distinguish dropout rates
by grade (grade 6 to 11). Moreover, we distinguish between retained students who dropped out of school by the end of the year of their retention and the dropout effect on all students enrolled in school the year after retention. Our robust estimates
reveal that higher retention increases the rate of students dropping out of school the same year of their retention, that means not enrolling to repeat the failed grade.
However, there is little, if any, causal effect of grade retention on the dropout rates of all other students enrolled in the school one year after retention. We find that the latter effect is stronger when retention takes place at the earlier grades whereas the
effect for retained students is strongest when retention occurs at grade 9 and grade 11, when students would be entitled to receive the lower secondary school certificate and the high-school diploma respectively.
Original languageEnglish
PublisherMaastricht University, Graduate School of Business and Economics
Number of pages57
Publication statusPublished - 2020

Publication series

SeriesGSBE Research Memoranda

JEL classifications

  • i20 - Education and Research Institutions: General
  • i21 - Analysis of Education
  • j24 - "Human Capital; Skills; Occupational Choice; Labor Productivity"


  • retention
  • school dropout
  • difference-in-differences

Cite this