Association for Public Policy Analysis & Management

This paper provides evidence that parental incarceration increases children’s educational attainment. I collect criminal records for 90,000 low-income parents who have been convicted of a crime in Colombia and combine it with administrative data on the educational attainment of their children. I exploit exogenous variation in parental incarceration resulting from the random assignment of defendants to judges with different propensities to convict and incarcerate. My identification strategy differs from the usual judge IV application because I model incarceration as two stage decision problem: First conviction, and then incarceration. I exploit judge leniency along these two different margins. Intuitively, I take advantage of the fact that I can compare children of parents who faced similar judge conviction leniency but had different incarceration leniency. I derive a new expression that extends the Local Average Treatment Effect concept, to a setting with two sources of unobserved treatment heterogeneity. I find that conditional on conviction, parental incarceration increases education by 0.8 years for children whose parents are on the margin of incarceration. This positive effect is larger for boys, violent crimes, and cases in which the incarcerated parent is the mother.