No description available.

No description available.

This thesis examines two topics related to the prevalence of unequal developmental outcomes among children of different socioeconomic backgrounds in Peru. The first topic focuses on the origins of these inequalities by measuring the relative importance of school and early childhood influences for the emergence of cognitive development gaps between urban and rural children. I use rich longitudinal information on cognitive achievement, household and school characteristics, and a novel decomposition strategy. Results reveal that observable school influences occurring between ages 6 and 8, account for a significant share (around 35% and no less than 28%) of the difference in cognitive skill. The share attributable to differences in the early childhood environment is important but no larger than 50%. I also discuss how the proposed decomposition strategy compares to other linear decomposition methods and why it is less prone to biases than those employed so far in the literature. The second topic is related to the effects and mechanisms behind early childhood development interventions that seek to improve the home environment through a change in parental behaviour. I use data on parenting practices and caregiver beliefs regarding the importance of parent-child interactions collected from the control and treatment groups of families randomized for the evaluation of a home-visiting programme recently launched at scale in rural Peru. I find that this intervention produced a positive effect on the quality of the home environment (d = 0.5; p < 0.01) and shifted caregiver behaviour increasing their participation in educational play activities with the child (d = 0.3; p < 0.01). I also explore the constraints that limit caregivers' behavioural change. I find that treatment effects are not significant among the poorest caregivers and that they exhibit a positive wealth gradient. Further, I find evidence suggesting that caregiver beliefs condition their responses to treatment.