This study investigates the effects of adverse conditions in-utero and in early childhood on children’s physical and cognitive development and explores potential mechanisms through which these adversities affect children. I focus on violent conflicts that represent a significant shock to the well-being of many households in developing countries, using data from Colombia. Young children represent a particularly vulnerable population to violent conflicts since in the early stages of life a child develops the physical/cognitive/socio-emotional skills that will become the foundations for lifelong productivity (Barker, 1972; Cunha & Heckman, 2007; Almond & Currie, 2011).

I use a rich household survey (Hogares Comunitarios de Bienestar) that includes detailed information on child/family outcomes (N~21,000), and most importantly, the geographic location of a child while in-utero and in childhood that allows me to identify the exact exposure to violence in these periods. I measure sudden changes in violence using the occurrence of massacres (defined as the intentional killing of four or more people by another person or group) in each month-municipality. The identification strategy exploits the variation in cohorts exposed to the Colombian conflict and compares outcomes of children exposed to violence to those of un-exposed children, within municipality-trimester/year, controlling for a rich set of characteristics, fixed-effects, and trends. In further analysis I also account for sibling fixed-effects that provide more conservative estimates of violence on child outcomes. Results show that exposure to violence in early-life reduce child’s height-for-age Z-scores and cognitive test scores by 0.11 and 0.22 of 1 SD. Regarding potential mechanisms, I find that violence is negatively associated with breastfeeding, protein consumption, and maternal employment, and positively associated with having an absent father. Violence also deteriorates parenting: psychological aggression increases, and parental time investments and other investments that stimulate a child’s cognitive development fall. Although I do not directly test it, these results suggest that maternal stress is an important pathway through which violence harm children.

This work contributes to the literature in several ways. First, to my knowledge, this is the first paper to explore the effects of violence in early-life on cognitive skills. Previous studies have mostly focused on health (Camacho, 2009; Akresh et al., 2012; Miniou & Shemyakina, 2013), and whether violence affects other domains of human capital remains mostly speculative. Identifying such effects is important since child cognitive skills are strongly associated with future outcomes (Heckman, Stixrud, & Urzua, 2006). Second, it provides information on whether timing of exposures matter (in-utero vs. childhood) and whether the timings differ across skills (health vs. cognitive). Third, this study presents novel evidence on potential mechanisms at the household level, an important area of research that helps formulate policies to improve the health of children in affected areas.