Armed conflict violence has effects beyond physical destruction and direct victimization. This paper investigates whether exposure of pregnant women to violence in the area where they reside affects their newborns' health; that is, the effects that may occur in the offspring even when the women themselves are not directly victimized.The paper aims to strengthen the existing empirical evidence on the importance of the prenatal environment and shocks for early life outcomes. Early studies (Camacho, 2008;Mansour & Rees, 2012) found scarring of health of newborns exposed to violence during gestation but overlooked gender differences in the effects. Because boys are less adaptable to prenatal adversity than girls, they are at greater risk of mortality and morbidity, especially when facing early pregnancy shocks (Kraemer, 2000). Consistent with this, more recent work (Dagnelie et al., 2018;Valente, 2015) found higher fetal mortality among boys, and evidence of survival selection reflected as an absence of worse health outcomes for those alive. It is hypothesized that this selection effect dominated in more deprived contexts, such as in the less developed countries analyzed in the latter studies-Nepal and the Democratic Republic of Congo (DRC)-where baseline health is poor (Almond & Currie, 2011;Valente, 2015), but there is still a gap to assess whether this would hold in relatively more prosperous developing countries, or whether non-biological factors can shape this relationship. With this in mind, this study re-examines the case of Colombia, initially addressed by Camacho (2008), disaggregating the analysis by gender and checking for survival bias. The present study uses survey data instead of vital registration records, which has a smaller sample size and hence less precise estimates. Nevertheless, this data allows assessing the impact on fetal mortality and two additional innovative extensions. Firstly, a contribution is made by analyzing the heterogeneity in the
