Abstract Leon (2012) found that exposure to political violence in Peru during childhood negatively affected completed schooling using a 2 percent sample of the 1993 and 2007 censuses. When using the full sample in 1993, we find statistically significant effects of exposure pre-conception, violating the parallel pre-trends assumptions of Leon (2012). We also find that the long-run effects (measured in 2007) are smaller in magnitude using the full census. Finally, we show valid evidence of an actual adverse impact of political violence, but only for those exposed in utero.
Huaroto et al. (Wed,) studied this question.