This dissertation studies the childhood determinants of cognitive and non-cognitive development as well as the determinant of educational outcomes and the consequences of public policy. I examine how educational policies and negative shocks affect the cognitive and non-cognitive development on early and middle childhood in the short and medium-run. Also, I investigate how a public policy impacts health outcomes of the population.

In the first chapter joint with Carmen Quezada, we estimate the impact of school meals quality on student outcomes. We take advantage of a staggered implementation of a national program that improved the nutritional content of meals in public schools in Chile starting in 2015. Using a Difference-in-Difference approach and national student-level data over six years, we estimate a credible Intention-to-Treat impact of healthier meals on Math and Reading test scores. We find an average increase of 0.036 standard deviations in combined scores. The students from the poorest and rural households present the largest effects.

In the second chapter, I evaluate the effect of the 2010 Chilean earthquake on early life on children’s cognitive and non-cognitive development in the short and medium-run. I exploit the quasi-random spatial and temporal variation using a Difference-in-Difference approach. I find that seven years later children between in-utero and five years old at the moment of the earthquake decrease their Peabody test score by 0.06 standard deviations by each additional unit in the Mercalli intensity. I explore possible mechanisms. I find that mothers in most affected places by the earthquake increase their probability of behaving riskily, specifically, in smoking from in-utero up to the first six months of the child’s life and a reduction of the household income two months after the tremor equivalent to 11% decrease of it; however, the effect faded away two years after the shock.

In the third chapter, I analyze the effects of the Daylight Saving Time (DST) transitions on automobile accidents in Chile between 2002-2018. I use a Regression Discontinuity Design exploiting the discrete nature of the transition into DST, and a Difference-in-Difference approach, taking advantage of the different dates that the policy starts and ends over the years. I find a 2.7% reduction in the number of automobile accidents under the DST regime. I isolate the two main mechanisms: sleep disruption and the reallocation of light. I find suggestive evidence that the sleep disruption effect plays a relevant role in both transitions: it increases automobile accidents by 7.4% the first week following the transition into DST and decreases them by 4.5% the first week following the transition into Standard Time (ST). However, I do not find conclusive evidence for the light effect. Under the DST regime, in the evening the negative effect on automobile accidents supports the light mechanism, however, in the morning, also a negative effect does not support it.