No description available.

This thesis comprises three chapters on the study of investments in human capital. The first chapter analyzes the effects of Head Start, a preschool program for poor children, on adolescents' behaviors. As program is means-tested, its effects are identified using discontinuity in the probability of program participation induced by eligibility rules. Since there is a range of income thresholds, which vary with family size, state and year, the effect is identified for a large set of individuals in the neighborhood of each discontinuity. Participation in Head Start reduces the incidence of behavioral problems, grade repetition, obesity, depression and criminal behavior. Among others, the first chapter leaves two questions unanswered: the first concerns the low take-up of social programs; the second relates to changes in parents' investments in their children when they face income shocks. To first question is addressed in the context of a program launched in Chile in 2002 - the Chile Solidario. The aim of this program is to provide psycho-social support to indigent families. As Head Start, Chile Solidario is a means-tested program and its effects are identified using a RD design. The program increases the take-up of subsidies and employment programs. However, information provided by Chile Solidario about other programs increases take-up only among those with less (direct or indirect) previous contact with the welfare system. The last chapter studies the ability of parents to insure investments in their children's human capital against income shocks. Parental investments in children are central to their development over the entire period of childhood. During this period families are potentially hit by a variety of shocks to their resources, which can impact investments in children. Understanding parental reaction is relevant for the design of anti-poverty programs that target vulnerable families with children and it may shed light on the mechanisms behind the effects found for child care programs and income transfers. The approach used builds on a life-cycle model that accounts for within and across periods nonseparability in parents' problem introduced by the accumulation of human capital of children (Cunha, Heckman and Shennach, 2010). I account for the fact that parents have an array of possible investments available. In particular, time maybe a suitable substitute for some types of goods/expenditure investments. I implement my analysis using the Children of the National Longitudinal Survey of the Youth 1979 (CNLSY79). Ignoring the multiplicity of investment that parents have in each moment and the cumulative nature of skills may produce biased estimates. I reject the hypothesis that time use and consumption towards children are separable over time; time use is complement to child's expenditures when children are less than six (for children of worse o families), but it is substitute of expenditures for school-age children.

The thesis aims to expand economists’ general understanding of rural development by interpreting the formation, expansion and functioning of rural labour markets as being situated at the centre of development. For this, we start by setting up an analytical framework that is rooted in classical political economy and that highlights the importance of socio-economic relations of power and property, processes of socio-economic differentiation, as well as the centrality of gender and intra-household relations as critical parts and complements in any holistic analysis. Through this we explore the deep theoretical links between the labour market and any process of rural development. The principal part of the thesis then goes on to applying this analytical framework using empirical survey data collected in the West Usambara Mountains region in North Tanzania through primary fieldwork in 2008 exploring wider processes of labour market participation and capital accumulation. A major aspect of this research is to compare our results with the influential study by Sender and Smith (1990) conducted in the same region in 1986. Some important results of our study are a relatively clear trend towards production being predominantly governed by wage labour relations, a general increase of economic pressure on people’s land holdings and a resulting process of increased differentiation and separation from the means of subsistence. We furthermore observed – in contrast to Sender and Smith’s conclusions – a tangible reduction of men’s capacity to coercively appropriate women’s labour power within the household, which they diagnosed to be a major impediment towards the greater process of development.

Although the relationship between immigration and crime has been a very controversial subject in the UK, the empirical evidence is limited. This thesis intends to narrow this gap by providing a comprehensive investigation for England and Wales of immigrants’ both active and passive involvement in criminal activities. Before exploring the aforementioned relationship, Chapter 1 discusses and provides solutions to an identification issue that afflicts leading models for under-reported count data. It also provides some tips for practitioners who intend to use these models in applied research. These findings are important for this thesis, since estimators that deal with under-reporting are considered in Chapter 2. Chapter 2 studies the individual-level relationship between immigration and crime using self-reported crime data. Although this work focuses on property crime, violent crime is also considered. Both binary and count data models that account for under-reporting are used, since under-reporting is a concern in crime self-reports. Our findings suggest that, if anything, immigrants under-report by less than natives. Most importantly, these models predict that after controlling for under-reporting and basic demographics, immigrants are less involved in criminal activities, but the estimated difference is statistically insignificant. Nevertheless, an extensive sensitivity analysis indicates that this estimate is very robust, suggesting that this relationship exists, but data limitations and complexities of the considered models reduce the precision of the estimated coefficient. Finally, Chapter 3 comprehensively examines whether victimization experiences are different between immigrants and natives. Very interestingly, although observed demographic differences can explain the positive property crime victimization-immigration differentials, unobserved factors give rise to a negative association between immigration and violent victimization. All results suggest that this is due to immigrants’ lifestyle choices associated with lower victimization risks. As will be explained throughout Chapter 3, this finding is consistent with the findings of Chapter 2.

This thesis focuses on two dimensions of the child production function - the technology of human capital formation and the role of speci…fic family inputs into human capital. The …first two chapters explore the technology by which inputs produce child human capital. Speci…fically, for given parental lifetime income, these ask whether the timing of income matters for later outcomes of the children. Two methodologies estimate the effect at different margins. Firstly in a fully flexible model, the relationship between parental income at child ages 0-5, 6-11 and 12-17 and subsequent child outcomes is estimated nonparametrically, allowing for complementarity across periods. Income aged 0-5 is as important in general as income at age 6-11 for child human capital formation. Complementarities exist between 0-5 and 6-11 for households with low permanent income, which are those likely to be credit constrained. Similarly, very strong complementarities are found between early years income and income during adolescence (age 12-17) for the group of poor parents. Chapter 3 analyses the role of permanent and transitory income shocks at different ages, upon adolescent human capital. Empirical results suggest the effect of permanent shocks declines across age. This is intuitive, given that a permanent shock changes household wealth and hence a shock at age 1 drives more future income realisations than a later shock. Transitory shocks on the other hand, have an increasing effect upon child outcomes across child age. Further, there is evidence of intrahousehold insurance against paternal transitory income shocks. The fi…nal two chapters of the thesis look at parental inputs in the production function. Chapter 4 allows the life cycle of skill formation to begin pre-birth, by estimating the role of maternal smoking during pregnancy upon birth outcomes. Results suggest a large proportion of the correlation is explained by a maternal fi…xed effect. Finally, chapter 5 offers a cross country comparison of the similarities in child test score gaps, by a range of measures of family inequality. Despite wide institutional differences, this chapter estimates homogeneous correlates for maternal education, family size and child gender upon child achievement, but differences in the covariates of lone parenthood and ethnicity.

The objective of this thesis is to contribute to a strand of the empirical labor supply literature by advancing our understanding of the labor supply of relatively older workers. This is a topic of particular interest in developed countries, where due to current population trends older individuals comprise an ever growing share of the population. Chapter 1 provides a summary and overview of the thesis. Chapter 2 shows that husbands and wives have an incentive to coordinate their retirements due to the existence of leisure complementarities, which arise when one or both spouses enjoy retirement more if it is shared with their partner. Chapter 3 advances our understanding of older individuals' incentives to continued work by showing that, after accounting for selection into retirement and composition effects, there is no statistical evidence that wages of individuals who remain in their career job ever decrease with age. In other words, conditional on remaining on the career job, the individual wage profile does not have an inverted-U shape. Any wage decreases associated to the declining physical and cognitive abilities associated to the aging process would materialize only at the point where the individual transits from the career job into part-time work, usually referred to as semi-retirement. For individuals that transit directly from the career job into full retirement, no decrease in wages would be observed. Chapter 4 builds on the results obtained in chapters 2 and 3 to estimate the role of leisure complementarities in determining joint retirements. If finds that they account for 8% of the joint retirements observed in the data (those where husband and wife retire within a year of each other). This result underlines the importance of jointly modeling the behavior of husbands and wives. Confining the analysis to the study of men while taking the behavior of their wives as exogenous -the approach traditionally followed in the literature-, ignores a source of simultaneity in spouses' decisions. This may lead to inaccurate predictions of the effect of policy changes on men's retirement behavior.

In this thesis we study the impact of monitoring and heterogeneity on the set of equilibria of dynamic games. In Chapter 1 we show how heterogeneity in time preferences can help create new intertemporal incentives. Proving the folk theorem in a game with three or more players usually requires imposing restrictions on the dimensionality of the stage-game payoffs. Considering a class of games in which those restrictions do not hold, we show how to recover a folk theorem by allowing time preferences to vary across players. In Chapters 2 and 3 we show how a small degree of imperfection in the monitoring technology can have large effects on the set of equilibria of dynamic games. We study a dynamic voluntary contribution game with irreversibility and a game with an asymptotically finite horizon. In both settings, when monitoring is perfect, players can cooperate and obtain payoffs in the repeated game that are strictly greater than the payoffs from the unique inefficient stage-game equilibrium. We show however that introducing an arbitrarily small amount of noise in the monitoring technology can cause a complete breakdown in cooperation. Finally in Chapter 4 we investigate how information is transmitted in a revision game with one-sided incomplete information. Players aim to coordinate on an action which depends on an unknown state of the world and players can only revise their actions stochastically during a preparation stage, at the end of which the prepared action profile is implemented. Miscoordination arises from the possibility of no longer receiving revision opportunities until the deadline. We show that close to the deadline no information is transmitted and that far from the deadline the uninformed player prefers to be miscoordinated.

In 1981 Chile was the fi…rst country in the world to privitise its pension system moving from a pay-as-you-go scheme (PAYG) to a De…fined Contributions (DC) scheme. Individuals in the labour market at the time of the reform were given the choice to either stay in the PAYG system or to opt-out to the DC scheme. New entrants must join the DC system. Exploiting the wide differences in pension formulas across schemes, I …firstly fi…nd that the reform signi…ficantly increased expected pension wealth for most of those who opted-out. I then investigate the extent to which households substitute this increase by decreasing accumulation of other wealth. As the decision to either stay or to opt-out was not random, I gain identi…fication through an instrumental variable approach. I …find a pension offset of around 30%. Among the possible reasons for the incomplete offset are imperfect information, the desire to compensate for new risks faced and habit formation. Lastly, through a non-linear random effects dynamic model that allows for state dependence and unobserved heterogeneity, I estimate the effect of pension system design on individuals' formal/informal labour market decisions. Results indicate that individuals in the DC scheme are 23% more likely to be formal than those in the PAYG scheme at any one period. Further, simulations show that the boost in formality caused by the reform lasts throughout the life cycle. State dependence is even more important indicating that labour market past decisions do affect future ones. The unobserved heterogeneity is also high and signifi…cant but it is only a …fifth of the state dependence. The results on state dependence and initial condition suggest that there is scope for public policy to affect formality decisions.

Chapter 1 investigates the impact of weather-related income shocks on infant mortality in rural Ecuador. I find that favorable weather conditions during the growing season have a negative effect on infant survival rates when the harvest takes place during the first and third trimester of pregnancy, and the first trimester following birth. My results suggest that the negative effects of an increased ma- ternal labour supply, following a positive agricultural productivity shock, during pregnancy and the first trimester after birth outstrip the positive effects resulting from the consequent higher income when considering year-to-year weather fluc- tuations. I also find that favorable weather during the growing season reduces -via maternal time- prenatal care, skilled assistance at birth, and breastfeeding duration and frequency. Chapter 2 explores the presence of spillover effects on schooling outcomes from the Colombian welfare program, “Familias en Acción”, on ineligible households in rural areas. The program provides cash subsidies to poor families conditional on children school attendance. I find that ineligible chil- dren — those living in a household that has not been classified as poor—residing in targeted areas are more likely to stay in the school during the transition pe- riod between primary and secondary school. My results suggest that peer effects might play an important role in schooling decisions as the increased grade com- pletion rate of the peer group increases the individual completion. Chapter 3 uses a randomized experiment to examine the causal effect of improving hous- ing conditions on child health, and adult mental health. We find that replacing floors, upgrading toilets, kitchen, and play areas has no impact on child health but these results are subject to a high level of non-selective attrition on children. We also find that the program improves caregiver’s mental health as measured by the CES-D depression score.

This thesis contains three chapters on dynamic models with discrete and continuous outcomes. In the rest chapter, I focus on indirect inference estimation. Indirect inference is used to estimate parameters in models where evaluation of the objective function directly is complicated or infeasible. Indirect inference is typically formulated as an optimization problem nesting one or more other optimization problems. In some cases the solution to the inner optimization problems can be obtained in one step, but when such a solution is not available, indirect inference estimation is computationally demanding. I show how constrained optimization methods can be used to replace the nesting of optimization problems and I provide Monte Carlo evidence showing when this approach is bene cial. The second chapter uses panel data from the United Kingdom to estimate a model of wage dynamics with labour participation where the variance in wages is decomposed in a permanent and a transitory component. Most studies that estimate similar models ignore non-participation; individuals without a wage are simply removed from the analysis. This leads to biased estimates of the parameters if working individuals are di erent in their unobservable characteristics compared to people that do not work. I use a dynamic selection model to include a discrete labour participation choice in a simple model of wage dynamics and compare the results to a version of the model that does not include labour participation. In the third chapter, I show how some of the assumptions on the dynamics of the unobservables in the second chapter can be relaxed. High dimensional integrals have to be approximated to estimate the less restrictive models. I use sparse grids and simulation methods to approximate these integrals and compare their performance on simulated data.