Please use this identifier to cite or link to this item: https://hdl.handle.net/10419/227297 
Year of Publication: 
2020
Series/Report no.: 
IZA Discussion Papers No. 13770
Publisher: 
Institute of Labor Economics (IZA), Bonn
Abstract: 
This paper presents the first analysis in the literature of the effect of test-based grade retention on adult criminal convictions. We exploit math and English test cutoffs for promotion to ninth grade in Louisiana using administrative data on all public K-12 students combined with administrative data on all criminal convictions in the state. Our preferred model uses the promotion discontinuity as an instrument for grade retention, and we find that being retained in eighth grade has large long-run effects on the likelihood of being convicted of a crime by age 25 and on the number of criminal convictions by age 25. Effects are largest for violent crimes: the likelihood of being convicted increases by 1.05 percentage points, or 58.44%, when students are retained in eighth grade. Our data allow an examination of mechanisms, and we show that the effects are likely driven by declines in high school peer quality and educational investments that result in lower non-cognitive skill acquisition. We find little effect on juvenile crime. Using the method proposed by Angrist and Rokkanen (2015), we also estimate effects of grade retention away from the promotion cutoff and show that our results are generalizable to a larger group of low- performing students. Our estimates indicate that eighth grade test-based promotion cutoffs lead to nontrivial private and social costs in terms of higher levels of long-run criminal convictions that are important to consider in the development and use of these policies.
Subjects: 
grade retention
crime
grade promotion policies
regression discontinuity
education
discontinuity high school
middle school
JEL: 
I2
K4
Document Type: 
Working Paper

Files in This Item:
File
Size
713.38 kB





Items in EconStor are protected by copyright, with all rights reserved, unless otherwise indicated.