Search Results

The current literature has demonstrated the strong association between the co-occurrence of illicit drug use and sexual behavior. However, tantamount to this relationship are, psychosocial factors which, when examined concomitant with health risk behaviors grouped by maternal educational attainment, will help elucidate differences between categories of youth at risk for compromised mental and physical well-being. The Bronfenbrenner ecological framework is utilized to substantiate the relevance of health risk behaviors, environment and the importance of studying psychosocial factors in multivariate models.

The data selected for analysis to both demonstrate these relationships and identify risk profiles originate from the National Longitudinal Survey of Youth (NLSY), Young Adult 1998 cohort. Partitioning the NLSY 1998 cohort by mother's education tests how risk proneness as a mediator differs by maternal highest grade completed, as it affects adolescent deleterious behavior. These data are renowned for an oversampling of African Americans and are nationally representative of other ethnic groups such as Hispanics and Asians, requiring the application of an algebraic weight to normalize against the US population. Therefore, the key findings discovered in this study are: (1) the mediational effect in the pathway to health risk behaviors is risk proneness; (2) reported depressive illness symptoms are the underlying mechanism of risk proneness; (3) the path model is robust when tested among different groups using the Bronfenbrenner ecosystem paradigm; and (4) the weighting technique is vital to preserving the original distribution of the population, since the study sample needs to reflect the actual proportion of racial/ethnic groups in the US population.

Risk tolerance and liquidity constraints are widely believed to be key determinants of self-employment, but their independent effects have proved difficult to identify. In the second essay, I specify a theoretical model that illustrates how individual risk tolerance and liquidity constraints affect the decision to become self-employed. I then tackle the empirical identification problem by constructing a measure of risk tolerance that is corrected for reporting error, varies with age and assets, and allows for the endogeneity of assets. In contrast to previous studies that use regional variation in housing prices as an instrument for assets, I address the fact that housing appreciation affects homeowners and nonowners differently. I find that risk tolerant workers are more likely to be self-employed than are their less risk tolerant counterparts. However, net asset levels have an insignificant effect on self-employment entry once absolute risk tolerance is properly taken into account.

The absence of successful businesses owned by minorities, and by blacks in particular, is a concern for policy makers. In the third essay, I exploit detailed work history data in the NLSY79 to provide new evidence on the reasons behind the race gap in self-employment. My analysis of an "age uniform" sample of men, all of whom are observed from age 22 to age 40, reveals that racial differences in cross-sectional self-employment rates are largely due to the fact that minority workers' self-employment spells are relatively short-lived. Moreover, I find that minority workers' relatively high exit rates from self-employment are caused primarily by transitions to nonemployment. Estimates from a multinomial logit model of self-employment exits suggest that minority workers' weak attachment to the labor market prior to entering self-employment is an important determinant of their self-employment to nonemployment transitions, while lack of prior industry and self-employment experience contributes to minorities' transitions to nonself-employment.

Methods: Multiple GEE models were generated utilizing demographic, pregnancy-related, educational, poverty, and wealth-related variables. Substantive comparisons across GEE models were done based on an estimate of the percent reduction in the baseline odds ratio for low birth weight in comparing non-Hispanic Blacks to non-Hispanic Whites associated with individual and collective socioeconomic related adjustments.

Results: Despite documenting a substantial socioeconomic disadvantage among non-Hispanic Black births relative to non-Hispanic White births, accounting for differentials in wealth-status produced only minimal reductions in the elevated odds of low birth weight observed among non-Hispanic Black births when compared to non-Hispanic White births.

Conclusion: Differentials in current wealth status appear to have little impact on observed disparities in low birth weight between non-Hispanic Black and non-Hispanic White births independent of traditional socioeconomic indicators such as income and maternal education.

The second essay takes a different approach to investigate the intensive margin effects of SNAP on adult obesity. To mitigate the severity of endogenous participation and misreporting biases, I employ a strategy that examines only individuals who report participating in SNAP. I utilize a quasi-experimental variation in SNAP amount per adult due to the timing of school eligibility for children. The identification examines intensive margin changes SNAP benefits due to changes in household composition. This study finds that increases in SNAP benefits available to adults actually reduce BMI and the probability of being severely obese.

Among the general sample of youth, the results suggest that long-term exposure to neighborhood unemployment increases the odds of being obese while moving to affluence decreases the odds of being obese. Among Black youth, long-term exposure to neighborhood unemployment also increases the odds of being obese while Latino youth were sensitive only to a more recent exposure to neighborhood unemployment.

Among the general sample of youth, gentrification increases achievement scores. Among poor and urban Black children and poor and urban Latino children, exposure to neighborhood unemployment and neighborhood poverty generally reduces achievement scores. Moreover, the negative effects of neighborhood unemployment and neighborhood poverty are stronger among disadvantaged youth compared to the general sample of youth. Surprisingly, Black neighbors increase the achievement scores of poor and urban minority youth while neighborhood unemployment increases achievement among poor and urban Latinos. Among poor and urban minority youth, gentrification has no effect on achievement scores while moving severely disadvantaged Latinos to affluence decreases their achievement scores.

The second chapter, which is work performed jointly with Tatiana Baron and Pengpeng Xiao, studies the drivers of the differential wage growth between men and women. We document the widening gender wage gap and the gender differences in the transition rates into and out of employment and human capital dynamics over the life cycle in the National Longitudinal Survey of Youth from 1979 (NLSY79). We then develop an equilibrium search model to assess the extent to which labor market frictions, human capital accumulation, and motherhood contribute to the differential wage growth between men and women. In our framework, men and women have exogenously different transition probabilities and diverging human capital processes. Furthermore, women's careers can be impacted by maternity leave coverage policies. Firms respond to these gender differences when posting wages in market equilibrium. Results are in progress.

The second chapter uses a life cycle model to evaluate the role of occupational study in providing information about occupational preferences. The model is estimated using a sample from the National Longitudinal Survey of Youth 1979. The results suggest that the opportunity to gain preference information is an important advantage of occupational courses of study. Furthermore, consistent with Chapter 1, the productivity gain from occupational study is greater than the gain from general study.

My research reveals a complex picture. On average, mothers who reenroll attain more educational credentials, work more, and earn more than similar mothers who do not reenroll. However, mothers who reenroll are less likely to be married, family income declines, and they experience some negative physical and mental health effects.

Mothers' reenrollment relates to small, short-term gains in children's vocabulary and reading scores, but also more behavioral problems that persist into early adulthood. In the long term, children of mothers who go back to school have better academic outcomes, but this does not manifest in earnings gains in their early careers. There is no effect on children's probability of getting married, but children have some negative health impacts on average. Results vary somewhat by subgroup.

Policy and practice can help women and their families balance education, family, work, and personal responsibilities. This is particularly important for women of color, who are more likely to experience education disruptions and return to school later in life and who represent an ever-growing share of college students.

The first two chapters of this dissertation examines the labor supply effects of fertility incentives by making use of two major policy changes that occurred in Canada over the past 25 years: the Quebec Parental Insurance Program which provides generous parental leave benefits and the series of cash-transfer fertility incentives introduced in Quebec in the 1980s. The empirical work for these projects was conducted using confidential versions of the Canadian Census and the Labour Force Surveys on-site at Statistics Canada. I find that while increases in the generosity of parental leave benefits substantially increase the birth rate and induce increases in labor supply among women of child bearing age, cash-transfer fertility incentives only slightly increase birth rates and decrease female labor supply. The cost of each additional birth due to an increase in the generosity of parental leave programs is $15,828 in 2008 Canadian dollars, whereas the cost of an additional birth due to cash-transfer fertility incentives is $223,625 in 2008 Canadian dollars. Therefore, paid parental leave is a low-cost way to increase fertility whereas the price per additional birth due to cash-transfer fertility incentives is quite high.

The third chapter studies the effect of parental income changes on students who have already matriculated into college. From my empirical analysis using the National Longitudinal Survey of Youth 1997, I find that individuals who experience large year-to-year declines in parental income while they are in college respond mainly by increasing their term-time labor supply and do not adjust their use of financial aid significantly. Students who encounter large parental income losses are also more likely to experience periods of non-enrollment prior to receiving their degrees, but by the end of the survey period studied are also more likely to have completed their baccalaureate or associates degrees. Responses to parental income declines do not vary substantially by parental income levels at age 18.

The second essay, "Fringe Benefits and Job Satisfaction?" uses waves 1996 - 2004 of the National Longitudinal Survey of Youth (NLSY) to study the impact that fringe benefits have on job satisfaction. Fringe benefits stand as an important part of compensation but confirming their role in determining job satisfaction has been mixed at best. The theory suggesting this role is ambiguous. Fringe benefits represent a desirable form of compensation but might result in decreased earnings and reduced job mobility. In this second essay fringe benefits are established as significant and positive determinants of job satisfaction, even after controlling for individual fixed effects and testing for the endogeneity of fringe benefits.

The final essay discusses union exposure and its impact on job satisfaction, specifically that of former union workers. The paper uses all waves of the NLSY to study how length of exposure to a union negatively impacts current job satisfaction for former union members. Furthermore, the length of accumulated time after a worker leaves a union job is positively related to job satisfaction. Therefore, unions have a lasting negative influence on the job satisfaction of workers, even after they leave the union workplace.

The results obtained from the duration models suggest that unobserved heterogeneity and duration dependence are important in explaining substance use. The results from the structural model reveal that the impact of past consumption on substance use is large, that alcohol and cigarettes are contemporaneous gross substitutes, that cigarettes and marijuana are contemporaneous gross substitutes, and that alcohol and marijuana are contemporaneous complements. Policy simulations suggest that individuals react rationally to price changes and that alcohol use and cigarette smoking act as gateways to marijuana use.

Data for this dissertation were drawn from five waves (2002, 2004, 2006, 2008, and 2010) of interview data from the National Longitudinal Survey of Youth (1997) consisting of 16,868 person-waves constructed from 6,392 adolescents enrolled in the study. Using a temporal ordering data analysis technique, stressful events from a previous interview wave were utilized as explanatory variables in predicting current depression and substance use. Other variables in the analysis, like social support, were believed to be acting contemporaneously to reduce depression and substance use. Using generalized least squares regression (GLS) for panel data for depression and generalized estimating equations (GEE) for panel data in STATA for the dichotomous substance use outcomes, results indicated that stressful events measured in the past were significantly associated with current depression, and with current substance use, controlling even for prior depression and substance use. Results also indicated that social support exerts a protective effect against the strain-depression and strain-substance use relationship. Race-specific and gender-specific modeling of each outcome demonstrated marked differences among relevant factors, with gender-specific models better explaining depression, and race-specific models better predicting substance use. Moderation analysis of relevant predictors and these key social statuses indicated that several salient and significant differences existed among the effects of the explanatory variables. Theoretical and policy contributions from this study are related to empirical support for the inclusion of depression as a negative affective state in general strain theory, while also reflecting important social structural conditions, like poverty, in predicting these relationships.

In the second chapter, which is a joint work with Tino Sanandaji, we study the effect of taxation on entrepreneurship, taking into account both the amount of entry and the quality of new ventures. We show that even with risk neutral agents and no tax evasion progressive taxes can increase entrepreneurial entry, while reducing average firm quality. So called "success taxes" increase startup of lower value business ideas by reducing the option value of pursuing better projects. This suggests that the most common measure used in the literature, the likelihood of entry into self-employment, may underestimate the adverse effect of taxation.

This dissertation is one attempt at an answer. Using two sets of publicly available survey data, I examined U.S. wage differentials for males and females of the 1983-84 birth cohort shortly after their completion of a bachelor's degree in the late-mid to early-late 2000s and for several years thereafter. The effects of individual choices on employment and wage outcomes are relevant to young adults at the start of their careers, especially those who invested time, money, and effort to improve their employment and career prospects through higher education. Empirical analyses were guided by a conceptual framework using a life course approach to integrate the human capital model of wage-setting (epitomizing the individual choice argument) and social determinants theory (representing the believers' position on discrimination). Results showed college educated females earned lower wages and experienced less wage growth than their equally qualified male counterparts despite conditioning or controlling for measures of individual choice and notwithstanding a labor market environment more equitable to women than any that came before. These divergent trajectories constituted a pattern that transcended the relationship between high levels of human capital and small wage gaps at early career outset and suggest that even for recent cohorts of educationally advan taged young adult female workers, individual choice is an inadequate explanation for the wage disparities they experienced in the aggregate.

This second essay within this dissertation estimates how narrowly-defined mismatches between employees and occupations affect young adult job mobility. I construct new measures of academic skill mismatch, technical skill mismatch, and edu cational mismatch using employee level data from the National Longitudinal Survey of Youth 1997 (NLSY97) and occupation-level data from the Occupational Information Network (O*NET). Skill mismatch measures capture the distance between an individual's tested abilities (in percentile terms) and the reported skill requirements of an occupation (in similar percentile terms). Estimation of a Cox proportional hazards model of job tenure reveals that a ten point increase in academic skill mismatch increases a worker's monthly hazard of leaving a job by 2.3%, yet technical skill mismatches have no effect. An employee is also 1.7% more likely to leave a job in any given month if ten percent fewer employees within their occupation possess the same level of education. The main benefit of my mismatch measures is their derivation from underlying individual and occupational characteristics. I demonstrate that these characteristics not only influence employee-occupation mismatch, but independently affect job turnover. Thus, their inclusion significantly alters estimates of how employee-occupation mismatches impact young adult job mobility.

This project uses seven decades of census data and 21 years of the National Longitudinal Survey of Youth--1979 to identify the relationship between veteran status, spatial mobility, and social mobility in the late 20 th century. I focus on three questions: (1) do veterans and nonveterans vary in their migration behavior?; (2) do AVF veterans enjoy greater intergenerational occupational mobility or earn higher incomes than similar men who did not join the military?: and (3) do AVF veterans and nonveterans differ in their ability to use spatial mobility to access communities with greater economic opportunity or specific social characteristics?

I find that veterans exhibit higher rates of migration across the life course, and that this effect persists regardless of changes in military staffing policy and human capital differences between veterans and nonveterans. AVF veterans do not generally earn more money than do similar men without military experience, and rates of intergenerational occupational mobility for veterans and nonveterans are quite similar. Finally, black and white male veterans appear to be unable to leverage their elevated rates of spatial mobility to facilitate enhanced levels of locational attainment. While veterans are not disadvantaged in their ability to migrate into communities with more positive economic and social attributes, any returns to increased migration for veterans likely result from greater mobility and not from a disproportionate benefit from that movement.

This dissertation has three main findings. First, I find that young women's early career goals are influenced by women's disadvantaged position in the labor force more generally, as manifested in relationships with their mothers and women's status in the broader community. Young women with mothers who have lower occupational earnings, lower occupational prestige, and work with more women are more likely to plan to work in occupations with lower earnings, prestige, and more women themselves. Second, part of the influence of community context and mothers' attainments is indirect through young women's beliefs about gender. Third, early career goals and gender beliefs have lasting and cumulative effects on women's attainments in later life. Young women with less ambitious career goals and more traditional gender beliefs complete fewer years of education and are less likely to work full-time in later life. In turn, less education and fewer work hours are associated with employment in occupations with more women, lower median weekly earnings, lower occupational prestige, and lower hourly wages. Overall, the results provide evidence of the social embeddedness of women's career goals, and the cumulative impact of early career goals and gender beliefs on women's mid-life attainments. In addition, the results suggest that women's disadvantaged position in the labor market persists partly because the career goals of each generation are influenced by the constraints and opportunities experienced by their predecessors.

The population values for the covariance matrix and mean vector for the latent trajectory parameters for the normal population come from McArdle and Bell (2000). The parameter estimates were obtained using a subset of data from the National Longitudinal Study of Youth (NLSY; Baker, Keck, Mott, & Quinlan, 1993; Chase-Lansdale, Mott, Brooks-Gunn, & Phillips, 1991). The data include 233 children who participated in the NLSY in 1986 when the children were ages 6 to 8. The data were selected because this subset of children had measures of their reading ability 4 times over 8 years. McArdle and Bell provide more information on how the reading scores were calculated. This study used the linear LGCM parameter estimates presented in Table 5.4 on page 83 of McArdle and Bell (2000)

A group-level analysis is performed using linear probability models on aggregate national-origin groups to ascertain whether the welfare use of an immigrant group affects the average level of high school graduation, college enrollment, and welfare participation of the second generation, net of immigrant groups education level. An additional analysis assesses the relationship between prior parental welfare legacy and subsequent outcomes at the micro-level of the individual using binary and multinomial logit models.

Results from the CPS analysis provide no evidence of an intergenerational correlation in welfare receipt across immigrant generations, but do provide descriptive evidence of a positive correlation between immigrant first generation welfare receipt and the young adult second generation educational attainment. The NLSY97 analysis shows a persistent negative association between welfare legacy and high school graduation; a negative association that is most pronounced for children of natives. Results of this study also show the largest effect of welfare receipt among the most disadvantaged group, the young adult children of immigrants from Mexican and Central American countries. The main finding of this study suggests that the negative impacts of welfare receipt might be lessened and in some cases reversed among the young adults from immigrant families. Such findings challenge the common notion that immigrant families use welfare as a crutch across generations and raise serious concern about U.S. immigration and welfare policies.

While it was clear that race and gender were significant attributes in identifying career choice and job satisfaction in this study, workforce training exhibited significant effects on work-life and and occupation type specifically for our target groups. Workforce training significantly impacted the variability of occupation type and work life specifically for minorities. Women showed minor changes in the variability of occupation type as an effect of workforce training. Both women and minority job satisfaction showed significant variability as a condition of workforce diversity. Job satisfaction was significantly less influential in determining work-life for minorities while White work-life was much more effected by job satisfaction. Workforce development and training show promising results as conditions to improve occupational choice for women and minorities, specifically in non-traditional industries.

The first essay estimates the effect of a one year delay in entering kindergarten on academic performance. To deal with the endogeneity of school entrance age, we exploit the variation in month of birth and state kindergarten entrance age laws. The results suggest that older entrants have higher test scores compared to younger entrants in the same grade and are less likely to repeat grades. However conditional on age, older entrants perform worse because they have completed less schooling. scores, educational attainment and maternal labor supply.

The second essay uses Two Sample Instrumental Variables (TSIV) to estimate the effect of delayed school entry on educational attainment. We show that previous studies that have used instrumental variables to deal with the endogeneity of school entry age are severely biased. In addition, we propose a structural model of optimal kindergarten entry age and use indirect inference simulation methods to estimate the parameters of the model. Further, we use our simulated data to obtain true Local Average Treatment Effect estimates of the effect of school entry age on educational attainment, scores, educational attainment and maternal labor supply.

The third chapter evaluates the effect of delaying kindergarten entry on long run maternal labor supply. I use an exogenous source of variation in maternal net earning opportunities, generated through school entrance age of children, to study intertemporal labor supply behavior. The estimates suggest that delayed school enrollment has long run implications for maternal labor supply. Results point towards significant intertemporal substitution in labor supply. In particular, rough calculations yield an uncompensated wage elasticity of 0.76, wealth elasticity of -0.37 and an intertemporal elasticity of substitution equal to 1.1.

The second chapter is a joint work with Devon Gorry. We analyze the effects of having a teen mother on child health outcomes from birth to young adulthood. Using data from the National Longitudinal Survey of Youth from 1979-2014, we use an empirical strategy that relies on miscarriages to put bounds on the causal impacts of teen childbearing. Results indicate that there is no evidence that teen childbearing has negative impacts on children, and there may be some cases of positive effects. In addition, children of teen mothers report fewer diagnosed disorders and conditions requiring medical attention. There is no clear evidence that insurance coverage or utilization are driving the suggestive health improvements in children of teen mothers, but we cannot rule out this avenue.

In the third chapter, I evaluate the differential impact that children's chronic health conditions have on mothers' labor market outcomes across child gender. I employ the method of event study approach using multiple data from the National Longitudinal Survey of Youth. Results indicate that although gender balance across health conditions shows male children to be more likely to suffer from health conditions, mothers of girls who have chronic health conditions are less likely to be employed and work for fewer hours than mothers who have healthy girls. However, I do not find any gap in maternal outcomes in the case of male children. These results may suggest that the marginal girl that gets diagnosed has more severe conditions compared to the marginal boy child. The preliminary analyses open the door for future exploration of the possible mechanisms.

I use the model and estimated parameters to measure how workers value the better non-wage characteristics and on-the-job wage growth potential that result when unemployment insurance enables them to search longer. Accounting for these previously unmeasured benefits roughly triples my estimate of the program's value to workers. These results suggest an important role for non-wage characteristics in other applications throughout labor economics.

My dissertation research findings confirm that stratification research stands on a shaky foundation in ignoring family complexity. I show that models of intergenerational mobility which include family-level measures of class or educational background do a much better job of representing the mobility patterns in the data than conventional models do. Moreover, the choice of family background measure affects research conclusions. For example, I show that the apparent leveling of a long term trend toward greater equality in social class outcomes among recent male birth cohorts is actually due to the increasing but unmeasured effect of mothers' class resources on sons' outcomes. Rather than leveling off, inequality has increased noticeably compared to older cohorts. I also show that the lower predicted educational attainment of children raised in single parent families results from a comparatively weak relationship between absent parents' educational resources and children's educational success. This weak relationship exists only when parent-child involvement is low---when children spend sufficient time with absent parents, their achievement levels are close to those of young people who live full time with both parents.

The first chapter focuses on the skill accumulation of workers who have completed their formal education. Skills are acquired through work experience in the learning-by-doing (LBD) model. This model predicts that once hours of work are accounted for, there should be no systematic variation in wage growth. I use this prediction to test the LBD model by estimating, conditional on hours worked, the correlation between wage growth and variables affecting the incentives to accumulate skills. This correlation is found statistically significant and suggests a rejection of the LBD model for a sample of male and female workers from the National Longitudinal Survey of Youth (NLSY) 1979.

The second chapter is co-authored with Lance Lochner. We use the NLSY 1979 and NLSY 1997 to estimate the effects of family income on educational attainment in the early 1980s and early 2000s. The effects of family income on college attendance increase substantially over this period. We develop an educational choice model that incorporates both borrowing constraints and a "consumption value" of schooling. The model cannot explain the rising effects of family income on college attendance in response to rising costs and returns to college without appealing to borrowing constraints.

The third chapter is co-authored with Marc Frenette and Lance Lochner. We conduct a parallel empirical analysis of the effects of parental income on post-secondary (PS) education attendance for recent high school cohorts in both the U.S. and Canada using data from the NLSY 1997 and Youth in Transition Survey. We estimate smaller post-secondary education attendance gaps by parental income in Canada relative to the U.S., even after controlling for family background and adolescent cognitive achievement. We develop an intergenerational schooling choice model that sheds light on the role of potentially import ant determinants of the family income - post-secondary education attendance gap. We document Canada - U.S. differences in financial returns to PS schooling, tuition policy, and financial aid, discussing the extent to which these differences contribute to the stronger family income - attendance relationship in the U.S.

Six subgroups of workers from a random sample of youth who entered the labor market in 1979 were studied: white women, black women, Hispanic women, white men, black men, and Hispanic men. Using an intersectionality framework, I establish that occupational injuries differ by race-gender. Race-gender differences in the initial odds of injury, time trajectories, and relationships between substantive predictors support an intersectionality framework. Differences in injury risk across time were documented and modeled using a Hierarchical Generalized Linear Modeling framework. Non-Hispanic white men began the study with the greatest risk of injury, while minority men had the lowest risk of injury. Across the 1990's differences between race-gender subgroups diminished. Non-Hispanic whites and black women have the greatest risk of injury in later waves.

In each chapter we used two U.S. nationally representative data-sets (the Health and Retirement Survey RAND data-set, and the National Longitudinal Survey of Youth) to statistically estimate the probabilities of survival and transitions between states at each age with regression analysis. These probabilities were incorporated into discrete age and discrete state matrices. We examine age-specific state structure, the average remaining life expectancy, its variance, cohort simulations, dynamic heterogeneity and individual trajectories.

Two broad questions motivate this research. First, how do the patterns of offending over the life course differ across immigrant and native-born groups? Second, what factors explain variation in offending over time for immigrants and does the influence of these predictors vary across immigrant and native-born individuals? These questions are examined using two separate datasets capturing information on immigration and crime during two distinct waves of immigration in the United States. Specifically, I use the Unraveling Juvenile Delinquency data and subsequent follow-ups to capture early 20th century immigration and crime, while contemporary data come from the National Longitudinal Survey of Youth, 1997.

Three particularly salient conclusions are drawn from this research. First, patterns of offending (i.e., prevalence, frequency, persistence and desistance) are remarkably similar for native-born and immigrant individuals. Second, although differences are observed when examining predictors of offending for native-born and immigrant individuals, they tend to be differences in degree rather than kind. That is, immigrants and native-born individuals are influe nced similarly by family, peer, and school factors. Finally, these findings are robust and held when taking into account socio-historical context, immigrant generation, immigration nationality group, and crime type. In sum, based on the evidence from this research, the simple answer to the question of whether immigrants are crime prone is no.

Using the National Longitudinal Survey of Youth 1997, this dissertation compared the transition outcomes of foster and non-foster care youth. Youth were matched using propensity score methodology, which models the likelihood to be in foster care based on a set of pre-existing characteristics, and traditional matching methods based on demographic characteristics (i.e., gender, race, age, parent's education, income, and having a stepparent). Bivariate and multivariate analysis revealed that foster youth fared worse than unmatched youth in all transition domains. However, outcomes were similar for foster youth and youth matched using propensity scoring in all domains except housing. This suggests that the pre-existing characteristics that put youth at risk for foster care may contribute to their difficulties in emerging adulthood rather than the experience of foster care itself. Findings using the traditionally matched samples showed some group differences suggesting that a larger set of attributes than those matched on contribute to foster youth difficulties. Analyses examining interaction effects between foster care and characteristics associated with negative transition outcomes did not identify any particularly vulnerable subgroups of foster youth. These findings are a substantive departure from past research, which suggested that foster youth were falling behind youth in the general population and comparison youth. Findings from this dissertation are used to provide policy and practice recommendations for foster youth and other vulnerable youth.

We further focus on the US higher education market. Using a two-stage least squares model, we build a macroeconomic model of supply and demand for US higher education as measured by enrollment. We find that college education benefits (e.g. relative earnings and employment levels), credit factors (e.g. student loan amounts and household debt), and financial aid shift demand. Higher tuition prices increase appeal of higher education but credit constraints put a barrier on demand growth. Tuition prices and debt levels are correlated, suggesting that students respond to higher tuition prices by borrowing. School's operating costs, government aid, and tuition and non-tuition revenue drive supply. Schools can use tuition prices to signal quality, and relative demand side price in-elasticity allows them to raise prices.

Finally, we narrow in on individuals. We compare post-college income across differing groups of student ability, school quality, and major choice. We condition on intrinsic student ability based on the quality of the top school to which a student was admitted. We find that attendance at an elite institution increases post-college income. For students not admitted to an elite school, income is primarily driven by student ability and major choice. Students majoring in engineering and business earn higher salaries than students focusing on humanities and arts even after adjusting for ability. School quality plays a key role in the college attendance and school choice decisions, even though we do not find a significant effect of these decisions on post-college earnings in most cases.

The second essay, "Adolescent Substance Use and Intergenerational Transfers: Evidence from Micro Data," provides empirical evidence for the use of pecuniary incentives by the parent to influence child behavior. Using the first seven waves of the National Longitudinal Survey of Youth, 1997 (NLSY97), I measure the effect of child alcohol consumption on parental transfers. Owing to the plausible endogeneity of the child's alcohol use in the regression equation of transfers she receives from parents, I estimate this relationship using an instrumental variable which utilizes variation in the price of alcoholic beverages over time and across states as a source of exogenous variation. The main finding of the paper is that after accounting for the possible endogeneity of substance use, the incidence of alcohol consumption among youths significantly reduces the amount of parental transfers they receive. Given the robust evidence for a negative correlation between youth substance use and their discount factor in the economics and psychology literature, this result provides an empirical basis for the tough love model of intergenerational altruism. The existing literature on joint retirement suggests that married couples tend to coordinate their retirement decisions which seem to be largely explained by the complementarity in their preferences for leisure. However, the recent trend toward increased labor force participation of older married women may make synchronization of retirement decisions more difficult as more recent cohorts of women become more strongly attached to the labor force and build their own careers, a fact that has been overlooked in the literature.

My third essay, "Cross-Cohort Differences in Joint Retirement: Evidence from the Health and Retirement Study," uses the Health and Retirement Study (HRS) data from 1992 through 2006 to document that the likelihood of a married couple jointly exiting the labor force (given that both were employed in the previous period) decreases across successive birth cohorts of wives. I then estimate a discrete choice multinomial model of labor force transition for married couples and find that, while economic factors have substantial power in explaining variation across married couples in retirement behavior, trends across cohorts in these factors do not contribute significantly towards explaining the observed cohort trend in joint retirement. This result suggests that non-economic factors, such as changes in social norms and attitudes towards work, are likely to be more important explanations for this observed trend. From a policy perspective, an implication of this finding is that the bias in the estimated effect of a policy aimed at influencing older workers' labor force behavior (caused by ignoring potential interactions in retirement decisions of spouses) can be mitigated if recent cohorts are less likely to retire together.

In the first study, which was ecologic in nature, using data on reported cases of gonorrhea provided by special request from the Centers for Disease Control and Prevention (CDC), I demonstrated that certain dimensions of segregation were associated with rates of gonorrhea among blacks in the United States.

In an attempt to determine whether segregation impacted sexual risk by impacting sexual risk behaviors, the remainder of the dissertation examined whether segregation could help explain black-white differences in sexual risk behaviors in a population-based, 11-year prospective study of adolescents in the United States (NLSY97). Specifically, for the second study, we used 2-level hierarchical survival analysis to simultaneously examine whether MSA-level residential racial segregation is associated with age at sexual initiation, after accounting for other area-level covariates, such as area socioeconomic position, and individual-level covariates, such as gender and family income. We determined that segregation was not associated with early age at sexual initiation overall but that it did help to explain the racial disparity in this outcome. In more segregated areas, blacks were at higher risk than whites, whereas no racial disparity existed in less segregated areas.

In the third study, we performed 3-level hierarchical linear regression to examine whether residential racial segregation was associated with a sexual risk index over 11 years of follow up. In this study, we did not find any evidence that segregation was associated with the sex risk index or that it modified the trajectory of the race-sex risk association.

The first chapter presents a model in which young men find employment to enhance their value as marriage partners. When the effect of employment on marital value declines, young men's employment declines as well, in preparation for a less favorable marriage market. Taking this prediction to U.S. data, I estimate that fewer young men sought employment after 2 interventions that reduced the value of gender-role-specialization within marriage: i) the adoption of unilateral divorce legislation, and ii) demand-driven improvements in women's employment opportunities. I then use a structural estimation of the model to investigate interaction between the marriage market and male labor market shocks. Simulations find that the indirect effect of a negative shock to wages on young men's employment, operating through the marriage market, is nearly as large as the direct effect that operates purely through the labor market. These findings highlight the changing marriage market as an important driver of secular decline in young men's labor market involvement.

To address the measurement problems demonstrated in chapter two, in chapter three we relate test scores to adult outcomes. Using data from the CNLSY, we perform order-preserving scale transformations on reading and math test scores to maximize their ability to predict completed education. We find that the black-white achievement gap grows during the early years of education when measured in terms of test scores' economic value. Classical measurement error is ins ufficient to explain the growth in the gap.

The first chapter of this dissertation investigates the long-term effects of parental incarceration on children. I utilize detailed intergenerational data and a variety of empirical methods to provide evidence that individuals who report resident parental incarceration during childhood experience depressed levels of educational attainment and earnings as an adult. These effects appear to vary by parent and child gender. The second chapter is concerned with estimating the returns to education attained after an incarceration spell. I analyze longitudinal individual histories of incarceration, education, employment, and earnings for a sample of former prisoners using regression and propensity score matching techniques. My results suggest that education has a positive effect on post-release labor supply and earnings, but this benefit is largely confined to the completion of four-year college degrees. The third and final chapter reevaluates the negative relationship between incarceration and earnings found in the current empirical lit erature. I extend this literature with a battery of quantile regression models. My results clarify incarceration's effect on subsequent low earnings and suggest that the incarceration wage penalty is smaller in magnitude for low-skill, low-earnings employment. In total, this dissertation extends the current understanding of incarceration's effects on individuals and households, particularly with respect to performance on the market for labor. Each essay also provides some insight into the effectiveness of American criminal justice policy.

The first empirical chapter utilizes panel data from the National Longitudinal Study of Youth (NLSY), a nationally representative survey, and growth curve models to examine racial/ethnic differences in wealth trajectories between ages 21 and 45. Findings reveal that relatively small wealth gaps between Whites, Blacks and Hispanics exist in their early 20s, but these initial inequalities are magnified with age. In the second substantive chapter, data from the Health and Retirement Study (HRS), a nationally representative, longitudinal dataset is used to examine whether racial/ethnic wealth gaps narrow, remain stable, or widen between during the years leading up to retirement (ages 51 and 73). Results show that Whites experience more rapid rates of wealth accumulation than their minority counterparts during middle and later life, resulting in accelerating wealth disparities with age, consistent with a process of cumulative disadvantage. At age 73, the average White household has a net worth of approximately $122,000, whereas both Hispanic and Black household have less than $5,000. Substantial racial/ethnic disparities in wealth trajectories persist after controlling for group differences in life course capital suggesting that other factors such as racial/ethnic differences in portfolio composition, financial transfers, and exposure to discrimination may contribute to wealth disparities. The third substantive chapter uses HRS data to examine racial/ethnic differences in health trajectories. Results indicate that there are dramatic racial/ethnic disparities in both the levels and rates of change in health. Overall, findings from this study show that racial/ethnic inequalities result in divergent aging experiences for Black, Hispanic, and White Americans.

In the third chapter, I estimate and calibrate a marriage matching model to quantify the share of returns to education that is realized through marriage. In the model, more educated agents earn higher wages in the labor market, and are more productive in housework. Men and women who marry benefit from the presence of household public goods, complementarities in household production, and the division of labor between spouses. The predictions of the model are matched with NLSY data on sorting in marriage, and data on the allocation of time from the American Time Use Survey. Counterfactual analysis for men and women at age 40, suggests that better marital outcomes generate in the order of 35 percent of the return to education for women around middle age, and in the order of 10 percent of the corresponding return for men.

While the model generates levels of mismatch broadly consistent with the data, the degree of positive sorting is underestimated for higher-skilled workers. The second chapter, "Targeted Search in Heterogeneous Labor Markets'", extends the theory of targeted search by introducing continuous skill heterogeneity among workers and firms in frictional labor markets. Workers are unable to fully direct their search, but instead pay an information cost to reduce the variance of the job offer distribution. A lower variance increases the worker's expected match quality but decreases the offer arrival rate. Results show higher-skilled workers target their search more intensely, decreasing the expected level of mismatch among higher-skilled workers and allowing the model to better fit the data on skill mismatch and sorting.

I. The Effects of Affirmative Action Policies in University Admissions on Human Capital Development of Minority Children: a Test of the Expectations Hypothesis.
It has been well documented that minority children leave primary school with lower levels of acquired skill than do their white counterparts. The causes of this "skill gap", however, are not well understood. This paper attempts to analyze one possible cause: the impact of perceived labor market discrimination on the human capital development of minority children. Using the CNLSY79 data, I take advantage of recent changes in affirmative action laws regarding university admissions in California and Texas as a natural experiment. I employ both difference-in-difference-in-difference and individual fixed effects methodologies to test for changes in achievement test scores among minority children between the ages of 7 and 14. The results show a significant drop in test scores among thirteen and fourteen year old African-Americans in the affected states relative to whites, but no significant impact among Hispanics. Younger age groups show negative, but insignificant effects. These results suggest that expectations do play a role in the human capital investment of minority children and further research in this area is warranted.

II. The Effect of Expectations about Future Opportunities on Human Capital Investment by Minority Parents and Children.
Given the large skill gaps that exist between minority and white children, and the degree to which these skill gaps are correlated with labor markets outcomes, it is important to understand why these skill gaps exist. Of particular concern is the fact that these skill gaps are present in children prior to entering kindergarten. In this paper, I analyze the impact of racially disparate future expectations among minority parents and children on their human capital investment decisions. A labor-leisure choice model is used to show that a perceived difference in future opportunities for minority children relative to whites should result in a reduction of human capital investment. This model is tested using changes in affirmative action laws to determine if parental and child human capital investments are affected by the changes in the policy. This paper utilizes the CNLSY79 data, which contain a number of parent and child input variables that are highly correlated with achievement test scores. The empirical results will be used to determine if the significant drops in minority test scores found in chapter one can be explained by changes in parental or child behavior. Additionally, these results will provide evidence on whether parental responses to negative expectations can be considered a plausible explanation for the large skill gaps that exist at very young ages.

Using rich household data from the National Longitudinal Survey of Youth (NLSY-97) and the Fragile Families and Child Wellbeing Study, I first show that cohabiting couples, relative to married couples, have higher separation rates and on average worse cognitive outcomes among their children.

Regression analysis revealed that that time preference measured using an additive scale of intertemporal behaviors was significantly associated with net worth. The incentive choice questions as a measure of time preference were not significantly associated with net worth. The respondents with a high rate of intertemporal discounting as measured by the time preference scale accumulated less net-worth than respondents with a lower rate of intertemporal discounting. In addition, in the regression model when individual behaviors involving intertemporal tradeoffs such as smoking, drinking, and not taking physical exams were added as individual behaviors, the model was the preferred predictor of net worth.

What explains differences in pre-market factors? Three types of inputs are believed to determine the skills agents take to the labor market: ability, family inputs and school inputs. Therefore it is crucial to understand first the importance of each of these inputs. The literature on the production of achievement has not been able to provide an estimation that can take the three factors into account simultaneously at the student level. Chapter 1 attempts to fill this gap by providing an estimation of the production function of achievement where both types of investments (families and schools) are considered in a framework where the inputs are allowed to be correlated with the unobserved term, ability to learn. I do this by applying Olley and Pakes' (1996) algorithm which accommodates for endogeneity problems in the choice of inputs for the production of achievement and by using parents saving for their child's postsecondary education to control for the unobserved component (i.e. ability to learn) in the production of skills. What makes this saving measure informative is the fact that parents decide it at the same time they choose the home and school inputs that will affect the observed test score (the current outcome). However those savings will not affect the current outcome, but instead will affect future labor market outcomes through college choices. The estimates for the role of family inputs are in line with previous findings. Additionally, the estimates of school inputs show that they are also important for the formation of students' skills even after controlling for ability to learn. The estimates of the production funct ion are used to compute counterfactual exercises. In particular, this paper evaluates what would happen if the inputs for black students are reassigned so that their inputs are the actual amount they receive plus the differential that white students receive. This exercise shows that equalizing home inputs would reduce the achievement gap by 15.6 percent while equalizing school inputs would do it by 9.2 percent. If instead inputs are altered only in 12th grade, house and school inputs have a similar impact on students' achievement: school inputs would reduce the gap by 7.2 percent while home inputs would do it by 7.4 percent.

Chapter 2 explores a further area that Chapter 1 does not discuss: whether parents substitute or complement families and school inputs. Parents may alter the investment in their child's human capital in response to changes in schooling inputs. If substitutability between parental and school inputs in the production of achievement is prevalent, then increases in school inputs could crowd out parental inputs. If instead there exist couplementarities between school and parental inputs, then increases in school inputs might increase parental involvement. Chapter 2 studies whether parents react when their child's school inputs decrease by studying out-of-school suspensions and their effect on parental involvement. Because out-of-school suspensions are chosen by the class teacher or the principal of the school and not by the parents, they are a good candidate for exogenous (to parental choice) variation in the level of school resources across students. Out-of-school suspensions are a consequence of student misbehavior, and thus do not occur randomly across students. Therefore, in order to capture the effect of how parents react to the decrease in school inputs, I instrument the number of out-of-school suspensions with measures of "principal's preference toward discipline." The identification comes from the fact that students in schools wi th stricter principals are more likely to be suspended. The estimates show that without controlling for selection, out-of-school suspensions are negatively correlated with the level of parental involvement. Once selection is taken into account, the effect disappears.

Earnings depend not only on pre-market factors but also on the agent's experience in the labor market. That is, it is important which job he gets and how his earnings evolve within and across employers. In their seminal paper, Topel and Ward (1992) estimate that nearly a third of total wage growth in the first 10 years of labor market experience is due to wage jumps at the time of changing a job. Unfortunately, the job ladder model, the workhorse for this literature, cannot explain the big number of wage cuts for workers that change employers (as opposed to those who remain in their job). An extension of the job ladder model that has been proposed to ameliorate this failure is the introduction of a shock to the existing employer-employee match. But such a process has not been identified empirically in the literature yet. Chapter 3 uses a particular feature of the National Longitudinal Survey of Youth (NLSY79) to provide a convincing identification strategy for the wage shock process: two measures of workers' compensation, wages and labor earnings. The first part of the chapter shows that although the dynamics of wages are consistent with a job ladder model, the same is not true for the dynamics of earnings. While relatively large wage increases follow job-to-job transitions. we observe that job-to-job transitions are negatively correlated with hourly earnings. We speculate that this is due to the fact that job-to-job transitions are more likely to follow a large reduction in wages. We find that this result is robust to mis-measurement in the labor supply and disappears for workers paid by the year. The rationale for this last finding is that. workers paid by the year are much less likely to be hit by wage shocks than other workers. Using the multiple measures of workers' compensation and data on employment transition, we calibrate a modified job ladder model that allows for shocks to the employer-employee snatch. We show that the model fits the data well and that a model that does not include this feature would fail to match the data.

In this study I focus on one under explored set of factors - individuals' life course expectations - which may condition these associations. Although a large body of research and theory in developmental psychology emphasizes the importance of expectations in shaping identity development and consequently, well-being, sociological and demographic research on family status and mental health has, with a few exceptions, largely ignored this perspective. I integrate these two research traditions to develop and test a theoretical model which argues that the mental health effects of role acquisition or absence, and the timing and sequencing of role entry, likely depends on expectations for role acquisition, timing, and sequencing.

Using data from the National Longitudinal Survey of Youth 1979 (NLSY79) I find substantial variation in individuals' ability to meet their expectations for the occurrence, timing, and sequencing of marriage and parenthood. For the most part, recent increases in permanent singlehood, age at marriage, childlessness, age at first birth, and pre-marital childbearing were not expected. OLS regression on respondents' reported depressive symptoms at age 40 indicate that expectations for the occurrence, timing, and sequencing of marriage and parenthood generally moderate mental he alth outcomes associated with the transition into marriage and parenthood, producing significant variation not only in mental health outcomes but also in mental health differences across marital and parental status. The fact that most of the demographic changes in family formation were not expected indicates that for those at their forefront these changes resulted in a substantial degree of structural strain, leaving few in this cohort psychologically unaffected.

The results of the current study indicate that there are stark similarities in the number and patterns of offending trajectories that emerge across race and ethnicity. Additionally, the current study finds support for both general and race-specific effects regarding the ability of risk and protective factors to distinguish offending trajectories. The finding that some risk factors have race-specific effects has implications for DLC theories which predict racial invariance in the causal processes that influence offending throughout the life-course. Additionally, the current study finds little evidence of trajectory-specific etiologies across the full study sample. This finding supports general over developmental theories and is consistent with prior research which indicates that risk factors are best able to distinguish between offenders and non-offenders rather than between offenders who follow divergent developmental trajectories. Overall, the current study findings contribute to the growing body of empirical research examining key DLC issues in the context of race and ethnicity.

Employer-provided health insurance and labor supply of married women. This work presents new evidence on the effect of husbands' health insurance on wives' labor supply. Previous cross-sectional studies have estimated a significant negative effect of spousal coverage on wives' labor supply. However, these estimates potentially suffer from bias because wives' labor supply and the health insurance status of their husbands are interdependent and chosen simultaneously. This paper attempts to obtain consistent estimates by using several panel data methods. In particular, the likely correlation between unobserved characteristics of husbands and wives affecting labor supply--such as preferences for work--can be captured using panel data on intact marriages, and potential joint job choice decisions can be controlled using fixed-effects instrumental variables methods. The findings, using data from the Current Population Survey and the National Longitudinal Survey of Youth, suggest that the negative effect of spousal coverage on labor supply found in cross-sections results mainly from spousal sorting and selection. There is only a small estimable effect of spousal coverage on wives' labor supply.

Health insurance tax credits and health insurance coverage of low-income single mothers. The Omnibus Budget Reconciliation Act of 1990 introduced a refundable tax credit for low-income families who purchased health insurance coverage for their children. This health insurance tax credit (HITC) existed during tax years 1991, 1992, and 1993, and was then rescinded. We use Current Population Survey data and a difference-in-differences approach to estimate the HITC's effect on private health insurance coverage of low-income single mothers. The findings suggest that during 1991-1993, the health insurance coverage of single mothers was about 6 percentage points higher than it would have been in the absence of the HITC.

Once a model is selected, the model parameters are needed to be estimated. The expectation-maximization algorithm (Dempster et al., 1977) and the data augmentation using MCMC (Hastings, 1970; Tanner and Wong, 1987a) are widely-used techniques to draw statistical inferences of the parameters for the LCPA model. As a number of measurement occasions increases in the LCPA model, however, the computation cost of expectation-maximization or MCMC will become exponentially intensive. On the contrary, if one adapts recursive scheme in the update steps, calculations will be simplified and become generalized to more time points. In light of this, we formulate each update step with recursive terms which are directly analogous to forward-backward algorithm (Chib, 1996; MacKay, 1997).

The parameter estimation for the LCPA model benefits from recursive formula, but the recursive algorithm still requires careful examination for the existence of multiple local modes of the objective function (i.e., log-likelihood). Applying the recursive formula, we implement deterministic annealing EM (Ueda and Nakano, 1998) and deterministic annealing variant of variational Bayes (Katahiral et al., 2008) in order to find parameter estimates on the global mode of the objective function. Both methods are based on the deterministic annealing framework, in which ω is included as an annealing parameter to control the annealing rate. By adjusting the value of ω, the annealing process tracks multiple local modes and identifies the globalized optimum as a result.

At last, we are interested in analyzing the early onset drinking behaviours among the young generation. We apply latent class-profile analysis to alcohol drinking behaviours as manifest in self-reported items drawn from the National Longitudinal Survey of Youth 1997, which was a survey that explores the transition from school to work and from adolescence to adulthood in the USA. To unveil the stage-sequential bevaviroal progressions, we adopt dynamic Dirichlet learning process to characterize the probable progressions in a discrete manner and then identify patterns in which similar progressions are grouped. For the parameter estimations, we conduct deterministic annealing approaches with predetermined annealing schedule.

The second essay concentrates on how health affects labor market outcomes. Past empirical work establishes a wage penalty from being overweight. In this essay, I exploit variation in an individual's weight over time to determine the age when weight has the largest impact on labor market outcomes. For white men, controlling for weight at younger ages does not eliminate the effect of older adult weight on wage: being overweight as a young adult only adds an additional penalty to adult wages. However, for white women, what they weigh in their early twenties solely determines the existence of an adult wage penalty. The female early-twenties weight penalty has a persistent effect on wages, and differences in marital characteristics, occupation status, or education cannot explain it. It also is not a proxy for intergenerational unobservables.

For simplification and motivated by the data set from the National Longitudinal Survey of Youth (NLSY79), we only consider a bivariate linear regression model with β = (β 1 , β 2 ), where β 1 and β 2 are both p × 1 vectors. We show that if β 1 ≠ β 2 , then it degenerates to a univariate linear regression model which is classified as case 1 in here. On the other hand, we show that if there is a linear restriction on β, for instance, the two column vectors in β are the same, that is β 1 = β 2 , it is a bivariate linear regression model which does not degenerate to a univariate linear regression model and is classified as case 2.

In this thesis, we propose to estimate the regression coefficients by a modified Buckley-James estimator and show that the modified BJE is consistent and has asymptotic normality under certain discontinuous regularity conditions for both case 1 and case 2. Moreover, in case 1, various non-normal asymptotic distributions of the BJE are presented when the regularity conditions are violated. In case 2, we further carry out simulation studies to compare the asymptotic properties of the modified BJE under different sample sizes and various continuous underlying distributions. We also perform data analysis to the NLSY79 data.

We further consider the estimation problem in case 1 with missing covariates involved, which is denoted as case 3. The extension of the BJE based on the generalized maximum likelihood estimator (GMLE) of the underlying distribution is discussed. The Newton-Raphson (NR) method is not feasible in computing the GMLE due to the large sample size of the real data example. We propose a self-consistent algorithm to bypass this difficulty.

Chapter two studies the impact of housing wealth on college enrollment in the housing boom and the housing bust. I take advantage of the recent housing boom and bust as an exogenous source of variation. I find that a $10,000 increase in home equity increases the probability of initial college enrollment by 0.19 percentage points. Housing wealth has a larger impact on college enrollment during the housing bust than during the housing boom. The asymmetry is only economically and statistically significant for families with lower annual incomes. According to my estimates, the decline in home equity during the housing bust would have caused a drop in college enrollment of 3.5 percentage points, or 9.6%, for families with income less than $70,000, other things equal. My results provide important implications for government financial aid policy. If the goal of the government is to maximize the college enrollment impact of a given level of financial assistance, it is useful for the government to implement a need-based counter-cyclical financial aid policy.

Chapter three studies the long-term effects of kindergarten enrollment on individuals' educational and social outcomes.

In the second essay, I use an index based on Roemer [1998] to study the effect of education policies on health inequality in two exercises. First, I use the British Household Panel Survey to estimate the effect of compulsory schooling reform on health inequality. Second, I use NLSY79 to study the effect of the court-ordered education financing reforms in the United States. My results indicate that the British reform does not have any significant effect on equality of opportunity for health. The U.S. reform, however, appears to be equality enhancing under our framework. In the third essay, I discuss the optimal level of cigarette tax under the equal opportunity framework. Using cigarette taxation as policy instrument, I compute the level of health inequity under different tax rates. I estimate an empirical model in which return to smoking is allowed to vary across individuals. Policy simulations are performed to estimate the optimal cigarette taxation to equalize opportunity for health.

The first chapter explores the relationship between workers' wages and the gender of their supervisor, conditioning on the occupational gender composition. It develops a theoretical model suggesting that supervisors' task assignment accuracy is affected disparately in occupations of different gender types, leading to varying degrees of skill mismatch among workers. This leads to average wage differences between workers with female supervisors and those with male supervisors in occupations of different gender types. Consistent with the theoretical predictions, the empirical evidence suggests that workers have better occupation-skill matches and higher average wages if they work with female supervisors in predominantly female occupations, compared to those with male supervisors; the opposite is true for workers in predominantly male occupations. Although not significant at the early career stage, supervisor wage effects emerge as a worker’s career develops. These findings emphasize the importance of supervisors' task assignment accuracy in workplace gender wage disparity, and underscore the necessity of integrating minority managers to the "gendered" organizational contexts.

The objective of this study is to examine women's and men's decisions to participate in demanding physical activity and attain a healthy weight. To achieve this, a productive household model of investment in health is first derived. Second, both trivariate probit and seemingly-unrelated-regression models of decisions on physical activity and BMI or obesity are developed. These outcomes are hypothesized to be related to health attitudes, prices of food, drink and health care services and products, the respondent's personal characteristics (such as education, adjusted family income, opportunity cost of time, occupation, marital status, race and ethnicity) and the respondent's BMI or being overweight at age 25. Third, data from the 2004 round of the National Longitudinal Survey of Youth (NLSY79) are used to fit the models.

Due to basic physiological differences in men and women, separate analyses are undertaken for men and women. Also, two physical activity equations, one for participating in moderate physical activity and the other one for participating in vigorous physical activity, are fitted. Findings include: an individual who ha s a higher adjusted family income has a lower current BMI or a lower likelihood of being obese; females with higher education are more likely to be obese or have higher BMI, while males with higher education are less likely to be obese or have lower BMI; older males within our cohort have higher BMI or higher likelihood of being obese; higher prices for fresh fruits and vegetables and non-alcoholic drinks increase BMI and likelihood of obesity for females but not for males; and higher prices for processed fruits and vegetables reduce BMI and likelihood of obesity for females but not for males. In a joint test of the null hypothesis of no food and drink price effects on the possibility to be obese, the hypothesis was rejected for women but not for men. When exercise is measured in minutes and weight as BMI, the hypothesis of no effects of the prices of food and drink on BMI is rejected for women but not for men. When individuals are classified as over-weight or not over-weight at age 25 and exercise is measure in minutes and weight is measured as BMI, the null hypothesis of no impact of food and drink prices on these outcomes is rejected for early non-overweight females, but not for males or early overweight females.

First, significant relationships were found between participation in job training and race, education, occupation, and typical length of work duration each year among women at mid-career. Whites were more likely than non-Whites to participate in both on-the-job (OJT) training and other training courses or educational programs (Other training). Women with education beyond high school were more likely to receive OJT and other training than those who were high school graduates or less. Women who were employed in professional, managerial, or technical occupations were more likely to receive both OJT and other training than those who worked in other occupations. In addition, women who typically worked for longer durations each year had greater odds of receiving job training than those who worked a shorter duration each year.

Second, a significantly positive relationship was found between mid-career women's participation in OJT over their working life and their hourly wage rates. However, participation in other training showed no significant relationship with wage rates. Findings from this study also revealed that women who were White, had a higher education degree, worked in the manufacturing industry, worked in a professional, managerial, or technical occupation, and typically worked a longer duration each year, tended to receive a higher hourly rate of pay. Based on these findings, several recommendations may be offered. (1) There is a need to close the gap in post-school training acquisition for women who tend to receive less training. (2) These findings support the need to continually promote continuing education programs and training services for mid-career women. (3) The goal of training policies should be to improve women's job placements. Also, training policies should serve to enhance women's employability and job mobility after the usual schooling age, when they may work longer and thus remain longer in the labor market.

In chapter 1, I provide a background and literature review of the body of work that forms the basis for my questions. In chapter 2, I investigate whether there are differences in the job search strategies used by the formerly incarcerated compared to those never incarcerated in the NLSY97, a nationally representative longitudinal data set. I then explore whether an incarceration spell changes the job search strategies of the formerly incarcerated by comparing their pre-incarceration job search strategies to their post-incarceration job search strategies. Finally, still using the NLSY97 data, I examine how specific job search strategies are related to employment outcomes for the formerly incarcerated compared to those never incarcerated.

Chapter 3. While more risk averse individuals are less likely to become entrepreneurs, theory predicts that more risk averse entrepreneurs pick ventures with higher expected returns and so they should survive in business longer than their less risk averse counterparts. Using successive entry cohorts of young entrepreneurs in the NLSY79, we find contrary to theory that the most successful entrepreneurs are the least risk averse. This surprising finding suggests that commonly used measures of risk aversion are not indicators of taste toward risk. Instead, measured risk aversion signals weak entrepreneurial ability--the least risk averse are apparently those who can best assess and manage risks. Indeed, our interpretation is consistent with recent experimental evidence linking cognitive ability with a greater willingness to accept risk.

Chapter 4. The fourth chapter investigates the stability of measured risk attitudes over time, using a 13-year longitudinal sample of individuals in the NLSY79.

Chapter 4 turns its attention to the family, and examines the marital transition of remarriage, an outcome unexplored in previous work. Using data from the National Longitudinal Survey of Youth 1979, I find that the gap in remarriage rates found between Black and White men in the United States disappears in the military context. Such results suggest that socioeconomic prospects are the culprit for differences in remarriage rates found at large, and that the relatively equal socioeconomic standing of service members, irrespective of race and ethnicity, help promote remarriage rates. Taken together, my research helps update and expand our understanding of the impact of military service on the life course specifically for racial and ethnic minorities, and what these differences might imply for social stratification and inequality.

From the National Longitudinal Survey of Youth 1979 cohort, 3,280 injured workers involving 5,204 events during 1988 through 2000 were followed up. Longitudinal analyses with 29,520 observations were conducted using SAS 9.1.

Different types of job insecurity at different level (e.g. at macro local economy level, job characteristics level, company level, and social structure level) were explored. Moreover, how different job insecurity measures at different levels are empirically related to workers' compensation outcome in the NLSY79 data was investigated.

Many covariates including education, income, occupation (measured by either injury risk or psychosocial factors such as decision latitude, job satisfaction), industry, and type of injury were considered. Both GEE logistic regression results (OR predicting filing by insecure contract = 0.55, 95% C.I. = 0.34_0.89, OR recent unemployment experience = 0.79, 95% C.I.= 0.65_0.96, OR combined job insecurity =0.75, 95% C.I. = 0.63_0.90) and panel data analysis results (beta = -0.18, p<0.0001) showed that workers in the low job security group are less likely to file for workers' compensation when they were hurt on the job.

Out of the filed cases, 53.5% were denied. The results showed that workers who have severe injury (OR = 1.85, 95% C.I. = 1.71_2.01), job insecurity (OR=1.82, 95% C.I.=1.52_2.17), low income (OR = 1.94, 95% C.I.=1.60_2.36), and manual jobs (OR =1.84, 95% C.I.=1.51_2.24) were more likely to lose wages when they filed claims.

The study found that there exist negative effects of workers' compensation filing such as lost earnings, employment disadvantages, and under-compensation by workers' compensation insurers. The study suggests that workers with low job security need to be protected from any reprisal action against their employment when they were filing for workers' compensation.

In the second chapter, I study the effect of alcohol consumption on youth fertility. Alcohol consumption is often believed to be a cause of risk-taking behaviors. Despite a well-established correlation between alcohol intake and various risk-taking sexual behaviors, the causality remains unknown. I attempt to establish a causal effect of alcohol use on the likelihood of pregnancy among youth using a variety of models ranging from a fully parametric to a semi-parametric discrete factor approximation method. Using data on 17-28 year-old women from the National Longitudinal Survey of Youth, I find that even after controlling for unobserved heterogeneity alcohol consumption increases the likelihood of pregnancy by 4.7 percentage points. This positive effect was observed in the semi-parametric model where the cumulative distribution of heterogeneity was approximated by a 4-point discrete distribution. Quantitatively similar but statistically weaker estimates were obtained from the two-stage least squares model and the bivariate probit model. Finally, models that ignore the effect of unobserved heterogeneity failed to establish this relationship.

The second essay in the dissertation estimates racial wage differentials among college graduates across academic majors. I examine the combined effects of positive attitudes towards hard work and the selectivity of respondents' undergraduate institutions on the wages of graduates and the racial wage differentials. I use data on respondents from the 1997 cohort of the National Longitudinal Survey of Youth (NLSY97) and Kitagawa-Oaxaca-Blinder decompositions to determine the magnitude of unexplained wage differentials between white graduates and Latino, mixed-race, and Black (grouped together as "marginalized") graduates. I find substantial wage gaps between the wages of the white and marginalized graduates, especially when decomposing differences between the wages of white men, marginalized graduates (both men and women), and marginalized women. The interaction between positive attitudes towards hard work and college selectivity contributes the most toward the unexplained difference in returns to characteristics between the two groups analyzed. This essay contributes to the extensive literature on wage differentials by providing another explanation for the existence of wage differentials between white and marginalized college graduates: differences in the returns to positive attitudes towards hard work and attendance of selective colleges.

In chapter 3, I use a dynamic model to analyze how changes in major-specific tuition levels would affect college and major choice. In my model, students face borrowing constraints; therefore, relatively small changes in tuition can potentially affect college and major choice despite large differences in lifetime earnings across majors.

Data for this dissertation come from the National Longitudinal Study of Adolescent to Adult Health, Fragile Families and Child Wellbeing Study, and the National Longitudinal Survey of Youth. These data have several attributes that make them ideal for this study: they include, for example, large adolescent populations, males and females, and, importantly, comprehensive measures of family size, delinquency, and controls.

In order to quantify the wage gap between heterosexuals and lesbians, gays and bisexuals (LGBs), a Mincer-style wage equation is estimated via Maximum Likelihood interval regression using data from the General Social Survey 1988-2002. I find that gay men face a 15% gap compared to heterosexual men, while the wage differential between lesbians and heterosexual women is very close to zero and not statistically significant.

It is often claimed that for discrimination against LGBs to take place, the employer needs to know that the employee is LGB. I estimate an endogenous switching model in which wages and the decision whether to disclose sexual orientation to the employer are simultaneously determined. When making the disclosure decision, the employee takes into account both expected discrimination and the psychological costs of non-disclosure. Using data from the Urban Men's Health Survey, I show that gay men who do not disclose their sexual orientation to their employers would face a 16% penalty for doing so.

Discrimination against racial minorities too presupposes that the employer has a perception of the employee's race. Using data from the National Longitudinal Survey on Youth (NLSY) 1979-2002, I compute for each self-reported racial group (blacks, Hispanics and whites) a measure of the times in which they are perceived as white by NLSY interviewers. Blacks who are perceived as white even only occasionally face a wage gap (with respect to whites) that is less than half of the wage gap that black men who are always perceived as black face. For Hispanics, being always perceived as white eliminates the Hispanic-white wage gap completely. I also show that more educated people, people working in some white collar occupations, and people who are married are more likely to be perceived as white.

Results from the latent class growth analyses resulted in two to three classes characterized by an abstainer group, an adolescent-limited group, and a stable moderate-level chronic group. In general, being male, increased substance use, and increased delinquency were consistent predictors of class membership. Regarding race and sex differences, being a minority was moderately related to class membership in males but was not significant for females. Being male was a very strong predictor of class membership for Black and Hispanic participants but a relatively weak predictor for White participants. Overall, results supported a general risk factor perspective over a gender or race specific risk perspective. Across race, sex, and cohort, self-reported delinquency was the strongest risk predictor of class membership, suggesting that differential arrest probability is predominantly explained by differential involvement in delinquent behavior.

Results of structural equation modeling suggest that economic pressure fully mediated the negative association between household income and women's depressive mood. However, no moderating effects were observed. For both white and black women, the effects of economic pressure on depressive mood did not vary according to women's locus of control or receipt of financial instrumental support from their parents. Additionally, women's locus of control was not associated with higher-order moderation of the effects of receiving financial instrumental support from parents.

The results suggest that approximately one-third of women and one-fourth of men belong to trajectories, or latent classes, characterized by chronic moderate levels of underemployment over the adult life course. In addition, various trajectories revealed the multifarious nature of stress by demonstrating unique patterns of severity, duration, timing, and sequencing of underemployment. Findings confirmed past positive associations between underemployment and being African American or having lower education levels. And, of principal interest in this dissertation, results demonstrated that membership in worse underemployment trajectories was generally associated with higher levels of depression and worse self-rated health, but not binge drinking or chronic conditions at mid-adulthood for women and men.

Furthermore, the results suggest that a cumulative process is at play whereby adverse childhood conditions, including lower parental education and several measures of family and other disadvantages, operate by impeding educational attainment and/or increasing the odds of membership in a higher-risk underemployment tra jectory, which ultimately harms health. There is also evidence that those who have higher levels of baseline depression are disproportionately selected into worse underemployment trajectories.

Ultimately, these findings indicate that the underemployed must be disaggregated from the employed in future research, whether employment status is the focus of interest or a control variable. They also highlight the need to theoretically and methodologically engage the dynamic nature of stress, as well as situate stress in a life course framework whereby heterogeneous pathways over the life course originating in childhood are examined.

Results indicated that an overall parenting construct characterized by parenting variables from both theories was associated with self-control and deviance. Further evidence indicated that parenting and self-control additively explained more variance in the engagement of deviance rather than self-control mediating the link. Finally, results indicated that deviance was best explained when three measures of self-control (i.e., ages 8-9, 12-13, and 16-17) were added to the model along with effective parenting.

Overall, results allude to the importance of examining parenting constructs as described by both the general theory of crime and the coercion theory. Further, self-control was found to be important in the explanation of deviance.

Health insurance mandates have become increasingly popular with policy makers as an alternative to public provision of health insurance benefits. In this paper I analyze the effects of state mental health parity mandates in the labor market and in the insurance market. Theory suggests that health insurance mandates could increase employers' costs and potentially reduce employer provided health insurance, and/or lower wages. States passed parity mandates throughout the 1990s and early 2000s, however, self-insured health insurance plans are not subject to these state regulations. Therefore, I estimate the effect of state mental health parity laws by assigning probability values of being subject to these state parity mandates - that is, probability values of being in a non-self-insured health plan - to each respondent in the Current Population Survey based on firm size, industry and year. These probability values are constructed using information on eligibility and, self-insured plans in the Employer Health Benefits Survey. Results provide no evidence that parity mandates increased the number of people uninsured. Findings on the impact on wages are not conclusive since coefficients seem to be sensitive to changes in the sample.

Essay one examined the determinants of money arguments within marriage utilizing Lundberg and Pollak's (1994) theory of non-cooperative game theory. Respondents' negative financial behaviors, higher income, and birth order (being laterborn) were found to influence a greater frequency of money arguments.

Essay two examined the predictors of individuals' financial behaviors, specifically socialization characteristics and gender role attitudes (traditional versus non-traditional). Using a theoretical framework of gender role theory (Eagly, 1987), younger age, not being married, being non-Black, non-Hispanic, being males, and having higher income were all found to be predictive of at least of one of the three financial behaviors used in this study.

Finally, using a theoretical framework of Becker's (1993) theory of human capital, essay three explored the intergenerational transfer of attitudes and human capital across two generations and their possible link to the respondents' financial behaviors. Results showed that mothers' enhanced human capital, endowed and attained, and nontraditional gender role attitudes have a significant positive impact on the children's financial behaviors. Respondents' income was also found to be significant.

In light of the results provided in Chapter 1, in Chapter 2 I study the contribution of the main determinants of wage gains from job changes to the early-career gender wage gap among highly-educated American workers.

The first chapter, "Explaining Income Inequality and Intergenerational Mobility: The Role of Fertility and Family Transfers," studies the effect of fertility differences across income groups on inequality and mobility. Poor families have more children and transfer less resources to them. This suggests that family decisions about fertility and transfers increase income inequality and dampen intergenerational mobility. To evaluate the quantitative importance of this mechanism, we extend the standard heterogeneous-agent life-cycle model with earnings risk and credit constraints to allow for endogenous fertility, family transfers, and education. The model, estimated to the US in the 2000s, implies that a counterfactual flat income-fertility profile would---through the equalization of initial conditions---reduce intergenerational persistence by 13% and income inequality by about 4%. The impact of a counterfactual constant transfer per child is twice as large.

The second chapter is an empirical analysis of public expenditure composition and economic growth. Many countries have adopted across-the-board cuts in the course of reform programs that are not desirable from the efficiency and equity perspectives, nor are they sustainable over time. For these countries, releasing resources for critical public programs and securing fiscal adjustment is absolutely necessary. We do a cross-country comparison to find out the productive and unproductive components of government expenditures and assess whether a change in the composition would have positive growth effects. We conclude that the composition-mix has different effects on growth for countries at different stages of development. There is a threshold effect for some categories of expenditure and so management of these resources is very important.

In the third chapter, we discuss issues whether eliminating the digital divide and providing home computers is a way for human capital formation for countries. Although computers are universal in the classroom, nearly twenty million children in the United States do not have computers in their homes. There is a big role of home computers in the educational protect and thus human capital formation. Home computers might be very useful for completing school assignments, but they might also represent a distraction for teenagers. We use several identification strategies and panel data from the two main U.S. datasets that include recent information on computer ownership among children--the 2000-2003 CPS Computer and Internet Use Supplements (CIUS) matched to the CPS Basic Monthly Files and the National Longitudinal Survey of Youth 1997--to explore the causal relationship between computer ownership and high school graduation and other educational outcomes. Teenagers who have access to home computers are 6 to 8 percentage points more likely to graduate from high school than teenagers who do not have home computers after controlling for individual, parental, and family characteristics. We generally find evidence of positive relationships between home computers and educational outcomes using several identification strategies, including controlling for typically unobservable home environment and extracurricular activities in the NLSY97, fixed effects models, instrumental variables and including future computer ownership and falsification tests. Home computers may increase high school graduation by reducing non-productive activities, such as truancy and crime, among children in addition to making it easier to complete school assignments.

Whites and different minority groups approach their education in different ways due to their upbringing and differences in culture. The aim of this research is to investigate the role of family income and family background characteristics on educational attainment of individuals. This analysis focuses on grade 9 completion, high school graduation, and college enrollment behavior for both males and females by ethnicity/race. In particular, this study compares the schooling attainment behavior of Asians with the other three ethnic/racial groups. Variables related to family income, family background, cognitive ability, and income returns to education are studied to see their effect on the schooling choices of Asians and how that compares to Whites and the other minority groups over time, as it is important to determine and understand the factors that primarily contribute to the outstanding educational success of Asians. The educational attainment models are estimated using Probit method with the data taken from three different sources namely NLSY97, CPS (1980 and 2000) and Census (1980 and 2000).

The results show that the influence of Asian mothers' schooling on their child's education is substantially lower than the mothers from the other three ethnic/racial groups and this could possibly be due to their higher overall expectations for their children compared to the other three groups, as the sole interest of Asian parents lie in their child's education and academic accomplishments. Family income has a positive and significant influence on achieving education for all racial/ethnic groups in 1980 and 2000. The effect decreases in most cases from 1980 to 2000 for all groups which could suggest less reliance on family income and availability of other sources of financing education and also increased importance of attaining higher levels of education. Income returns to education has a greater impact on grade 9 and high school completion for both Blacks and Hispanics in most cases compared to Whites and Asians, suggesting higher educational and earnings aspirations for the latter two groups.

Chapter three examines the effect of housing wealth changes on private school enrolment. I use data from The National Longitudinal Survey of Youth's child supplement to examine the relationship between housing wealth and private school enrolment. I use a multinomial logit model and find that self-reported housing price changes increase the likelihood that respondents switch from private to public school. Heterogeneity analyses reveal that house price increases have a positive relationship between switching from private to public school across income, gender, race, and religion. Finally, a rise in house prices increases the likelihood that a child moves from public school to private school when transitioning from middle school to private school.

The first chapter provides an introduction to the health behaviors and other maternal influences discussed in this dissertation. It is important to model the subject’s childhood accurately to account for childhood factors that could affect future decision making. After controlling for the subject’s childhood environment, the researcher can see the direct impact of maternal health behaviors on her child’s future health decisions. The health behaviors at interest in this study are cigarette smoking, alcohol use, and binge drinking. There is an intuitive link between parental, here only maternal, behaviors and child behaviors. The real question is if the child, the subject, sees this influence as a positive or a negative in future decision-making. The unique design of this study measures maternal health behaviors during the subject’s childhood, aged birth through 15, and subject health behaviors later in life, aged 16 and older.

Chapter 1 looks at cohort and generation differences in levels and predictors of early fertility and sex among first, second, and third-plus generation young adults in the U.S. during the late 1970s and 1990s. The analyses seek to determine (1) how this behavior has changed by generation and arrival cohort and (2) what family and community factors predict these outcomes for immigrants. The study finds that first-generation immigrant youth exhibit the lowest levels of early fertility and sex, negatively assimilating by the third-plus generation. The findings suggest a positive association between arrival cohort early sex only. Further, the research fails to uncover significant predictors of these outcomes among first and second generation immigrants although the measures used have long been associated with them in the general population--suggesting that protective effects of generation reflect unmeasured cultural and social mechanisms.

Chapter 2 examines whether welfare policy changes occurring in the 1990s altered fertility outcomes for low-educated and unmarried foreign-born women. The study also looks at whether welfare reform had unintended consequences on immigrant women, like reducing prenatal care or decreasing the fertility of welfare-eligible or non-welfare dependent women through "chilling effects." The results show no consistent evidence that welfare policy changed fertility behavior among any immigrant group and raises doubt concerning the usefulness of targeting women's childbearing decisions with economic incentives.

In the second paper, I consider three extensions to the original self-selection model presented in Chapter 3. Workers are assumed to accumulate human capital and experience dismissals that are exogenous to their behavior. Results suggest that adding human capital accumulation to the original model enhances the model's fit to the white collar wage data but worsens its fit to observed dismissal rates in both occupations. The model that includes both human capital accumulation and exogenous dismissals explain the observed wage and dismissal dynamics the best. Human capital accumulation is responsible for most of the wage growth observed in the white collar occupation whereas the direct effect of human capital on the blue collar wage growth is not substantial.

We decompose the causal effects of maternal education into direct and indirect channels. Ours is the first study on intergenerational returns to education to do this decomposition. We utilize child fixed effects to account for endogeneity of maternal schooling and other inputs due to unobserved time invarying characteristics. Hausman tests indicate that fixed effects are indeed needed. Our results show that an additional year of mother's schooling causally increases Math and Reading test scores and reduces a child's risk of being overweight. Additionally, direct effects of maternal education are more important for Math, Reading and probability of a child at risk of being overweight; while indirect effects are more important for Behavioral problems. Lastly, we find that role of the father figure in the family differs by family structure. In families where father is the biological father, mother's education is equally important as the father's. However, when father isn't biological, mother's education is more important for behavioral outcomes.

In this dissertation, I use both Current Population Survey data and National Longitudinal Survey data to compare black women's and white women's wages between 1980 and 2002. I examine general trends using summary statistics, then employ linear regression and relative distribution methods to examine the relative effect of human capital and job characteristics on women's wages. In addition, I develop synthetic wage trajectories to examine the wage growth of women over their twenties based on educational attainment.

Taken together, the analyses in this dissertation show that broad industrial restructuring resulted in an "office economy," providing women with greater opportunity that coincided with their increased labor force participation and educational attainment. Women disproportionately moved into managerial and professional occupations relative to the total population, but white women reaped greater benefits from this move. Although black women experienced wage growth, their overrepresentation in the service industry and worsening relative wages as professionals, managers, and sales workers increasingly disadvantaged them. In addition, NLS data indicate that black degree holders had flatter wage trajectories over their twenties in 1991 relative to 1980, thus even the most advantaged black women suffered losses during the 1980s.

To date, most analyses regarding wage inequality among women have limitations including focusing on estimating the effects of changing selection into the labor force and restricting the sample to young women. This broad, descriptive analysis provides a foundation for future research examining the accumulating disadvantage of black women workers relative to white women.

In the second chapter, we explore one aspect of U.S. education that has not garnered a lot of attention until fairly recently that is occupational choice. We add an education sector to an otherwise standard Hsieh et al. (2019)-style model to explore the extent to which changes in career opportunities in other occupations affect the selection of workers into teaching careers. In our model, changes in the allocation of teaching talent have implications for the evolution of class size as well as quality of instruction and hence the accumulation of human capital during the workers' formative years. This gives rise to a trade-off between static and dynamic efficiency, which we quantify by way of a structural model.

The second essay tests whether the theoretic model that explains the racial wage gap can also explain the gender wage gap. Specifically, I test whether the correlation between AFQT and wage is stronger for a employer-employee couple that shares the same gender than for a couple with opposite genders. I find that the data does not support this hypothesis.

The data used for this study were from the 1979, 1980, 1981, 1993, 2000, 2002, 2004, and 2006 waves of the National Longitudinal Survey of Youth 1979. Well-being in each dimension, investment in risky financial assets, and investment in education are jointly modeled using simultaneous equations models.

In the first equation of the simultaneous equations model, years of schooling which was an indicator of education investment was predicted by one's risk tolerance and a set of preference characteristics. Objective risk tolerance under the drug use and sales domain was found to have a significantly negative impact on years of schooling, which means less risk tolerant persons invest more in themselves via education. Consistently, females, Hispanics, and Blacks who were commonly considered less risk tolerant or more economically disadvantaged were found to have more years of schooling. The findings provide evidence for the argument that investment in education may be viewed as one type of insurance, rather than as risky investment, by the young baby boomer generation. Young baby boomers who were not married or did not have any children also had more years of schooling, reflecting the fact that investment in education involves considerable opportunity costs. Therefore, the young boomers who were comparatively short of time tended to invest less in education. Notice that parental education and education of the oldest sibling affected education attainment of the young baby boomers, implying a lasting and extensive effect of education investment that spreads not only from generation to generation but also within generations.

The second equation in the simultaneous equations model was to examine potential factors that explained any discrepancy in risky financial investments across the young boomers. Empirical results supported the hypothesis that investment in education is a significant predictor variable, which means the young baby boomers who have more schooling allocate their money in the financial market in the form of holding more stocks or possessing a higher ratio of risky financial assets to financial assets, including stocks, corporate bonds or government securities, and mutual funds. Economic indicators such as net worth and total family income had a positive effect, regardless of the definitions of risky financial investments, which implies risky financial assets are normal goods. Home ownership was also a significant predictor for both risky financial investment measures, suggestive of a mechanism of diversification provided by home ownership that helps to minimize the risk from any investment. Objective measures and survey-based measures of risk tolerance, however, failed to account for multi-dimensional nature of risk.

The last equation in the simultaneous equations model was to examine the determinants of individual well-being in particular dimensions. Empirical evidence demonstrated the importance of risky financial investment in determining the difference in economic well-being and the significance of education investment in predicting the discrepancy in economic well-being and psychological well-being, namely self-esteem. Total family income partially explained the disparity in psychological well-being. Note that preference characteristics played different roles across well-being dimensions, accounting for a larger proportion of variation in economic, physical, and psychological well-being. These results are suggestive of multi-dimensional well-being that cannot be predicted by one single common factor, implying the necessity of diversity of government policies and programs.

Based on the empirical results, theoretical and policy implications were drawn. These findings suggest that the original Capital Asset Pricing Model partially explained demand for risky financial assets. Deviations from the model emphasize the importance of adding human capital and other types of assets to the explanation of demand for risky financial assets. The current study has several implications for practitioner, policy makers, and researchers. These implications pertain to allocation time and money on investments in different domains, appropriate guidance to meet clients' financial goals in terms of their time-varying, conditional risk tolerance, instant transfer and tax policies for subsidizing education investment, and development of assessment instruments including multi-dimensional situations and scenarios in each specific domain to assess risk tolerance of individuals.

The third chapter studies the contemporaneous effects of family income and maternal employment on the cognitive and noncognitive skills of children ages 5-16. By using legislative changes associated with income tax liabilities and interstate banking as exogenous sources of variation in family income and maternal labor supply, this study makes important improvements to the methodologies existing in the literature. It shows that on a child's cognitive achievements family income has a significant positive effect, but maternal labor supply has a negative effect. Family income has no significant effect on a child's noncognitive development, whereas maternal work has a significant effect although the direction of effect varies among different sub-scales. These findings confirm the existence of the trade-off between time and money that mothers face. The trade-off arises because maternal working hours, with its negative direct effects, may yield positive indirect effects through income. Furthermore, our extensive investigation on mechanisms suggests that that the available sources of care as an alternative to parental care may not be as conducive to producing non-cognitive skills as they are to producing cognitive skills. With an increase in income, families could afford a better school or a better after-school activity but these alternative sources may be better only at producing cognitive skills. Thus, even with an increase in income families have little scope to switch to a more productive alternative care conducive to non-cognitive skills, resulting in an insignificant effect of income.

The second essay examines the effects of a disabled child on the mother's marital durations. This is the first study that provides a complete picture on how child disability affects different types of marital durations by examining whether child disability has a disruptive effect on the mother's marriage duration, whether child disability deters divorced mothers from being remarried, and how child disability affects never-married mothers' likelihood of starting a marriage. This study also investigates multiple married or unmarried spells to obtain more insight of how a child in poor health affects the mother's marital status in the long term, beyond the first marriage and the first divorce. Using data from NLSY79, the vast majority of individuals of the sample are followed for more than 20 years. The study uses a duration model, which has the advantage of taking into acco unt of time-varying covariates and censored observations. The estimation procedure also allows unobserved heterogeneity. The results show that a single mother with a disabled child has a significantly lower likelihood of starting a new marriage than her counterparts. Simulations show that child disability deters divorced and never-married mothers from a (re)marriage by an average of 15 months or 14 months, respectively, compared to the situation when the child is in good health. However, the results do not show any evidence that a disabled child has a disruptive effect on the parents' existing marriage.

This dissertation is intended to contribute to our understanding of the decisions made by adults to return to school. Using information from the National Longitudinal Survey of Youth 1979 (NLSY79), the following primary questions are addressed: Does labor income, age, or tuition decrease the likelihood that an adult will return to school? Do expected gains from further education increase the likelihood that an adult will return to school? Do other factors such as family income, race and gender affect the likelihood that an adult will return to school?

The findings of the empirical analyses indicate that if a less-educated adult belongs to a racial minority, is male, is relatively older, lives in a poor family, or earns lower-than-median income, then he/she is less likely to return to school. Among persons from poor families or earning low incomes, an increase in income would actually discourage investment in further education instead of making it more affordable for them to return to school. Those who return to school tend to have higher-paying jobs or belong to relatively richer families. Females are more likely to return to school the longer the time that has elapsed since their high school graduation, while minorities are more likely to enroll in college immediately after or within a relatively short period after graduation from high school.

Using NLSY79, the first chapter shows that return to training is higher in small than in large establishments. This new empirical finding is not explained by the existing theories which are based on competitive assumptions. An equilibrium search model is constructed to interpret this empirical regularity. With suitable parameters, the model generates an equilibrium characterized by wage dispersion and one in which (1) large firms pay more to their workers, (2) they train a higher proportion of their workers, and (3) wage return to training is lower in large firms.

The second chapter analyzes the determinants of wage growth for blacks and whites and decomposes the difference in their growth rates. The average difference in black-white wage growth can be almost entirely attributed to the differences in endowments, with pre-market factors---schooling and AFQT---explaining two-thirds and general labor market experience accounting for another third of the differential. The role of labor market discrimination appears to be small.

Chapter three discusses changes in black-white inequality in relation to establishment size. Important aspects of the employment relationship are examined, including job initiation, fringe benefits, on-the-job training, wages, and job separations. The chapter shows that overall racial discrimination in the U.S. labor market does not diminish as establishment size increases, except in the case of hiring. In addition, relative preferential treatment for blacks in terms of hiring is solely responsible for the overrepresentation of blacks in large establishments.

The last chapter summarizes main findings in the dissertation in the unifying framework of equilibrium search theory. It is argued that perfect competition assumptions do not accord with the U.S. labor market well and search is an effective alternative to model the decentralized market.

Additional analyses link these differences to ways SES modifies within-family developmental dynamics. Disadvantaged families invest more in developmentally advanced than less advanced children in early childhood, while advantaged families invest more equally across children. Such dynamics may exacerbate socioeconomic inequality among children who face early developmentally challenges. As children mature, these differences either disappear or reverse, reinforcing socioeconomic disparities among more skilled children.

Decomposition analyses using these same data trace a substantial degree of inequality in educational outcomes to the fact that high-SES children not only have more of the skills, resources, and experiences that promote educational success, but also derive greater benefits from these factors. The same holds in an experimental analysis of a social capital-building developmental intervention, wh ich primarily benefitted the most socioeconomically advantaged families and children in the study.

In sum, efforts to understand and address intergenerational inequality must account for the fact that socioeconomic disparities between families alter integral aspects of the processes that shape children's developmental and educational trajectories. Whether the goal is to promote upward mobility, reduce inequality, or make educational and developmental interventions more efficient or effective, it is important to consider how socioeconomic background modifies children's environments and experiences.

Although the model I developed in the second chapter provides a good fit to the wealth distribution, it was not able to replicate the vast wealth held in the top 5 percentiles of the distribution. Thus, in the third chapter of my dissertation I generalize the income possibilities to include returns on entrepreneurial ventures. I construct and simulate a model where individuals choose between wage work and entrepreneurship. In combination with capitalist spirit preferences the possibility of high returns on entrepreneurial ventures allows the model to generate all of the skewness of the wealth distribution.

The final chapter of this dissertation focuses on the relationship between wealth accumulation and the decision to become an entrepreneur. Empirical studies find a positive relationship between being male, white, older, married, having more assets, and self-employment (see Dunn and Holtz-Eaken 2000). Yet, most theoretical research focuses on the importance of unobservable factors such as risk aversion, entrepreneurial ability and the desire to be one's own boss (see Evans and Jovanovic 1989). Not surprisingly, there exists little empirical evidence on how important these characteristics are in the decision to become an entrepreneur, in particular, whether they play major or minor roles relative to assets and wage opportunities. Further, there is no consensus on whether aspiring entrepreneurs are liquidity constrained (Gentry and Hubbard 2000; Hurst and Lusardi 2003). Using a panel of data from the 1979 National Longitudinal Survey of Youth, I show that not only are liquidity constraints unlikely to be the major factor in the decision to become an entrepreneur, the characteristics pointed to in the theoretical literature are far more important for this decision.

In my second chapter I seek to test the hypothesis that suburban living environments lead to a higher body weight compared to central city environments. I use the geographic information from the confidential NLSY79 files in conjunction with census tract population density data for the years 1981 to 2004. The effect of population density on the body weight of individuals who move is estimated conditioning on individual and area level characteristics. The chapter concludes that the actual effect of population density on the weight of individuals is not as large as the public health literature would want people to believe, but it is neither as insignificant as the urban economic literature to date estimated.

In my last chapter I survey the literature on labor market effects of obesity and then estimate the effect of obesity on labor market outcomes in Germany. To date, only one study briefly considers the effects of weight on labor market outcomes in Germany. This chapter argues that, apart from individual earnings, other labor market outcomes in conjunction with additional data sheds more light on the relationship between weight and labor market outcomes in Germany. This chapter examines the relationship between body weight and labor market outcomes in Germany. While no definite conclusion is reached, this is the first study that attempts to study the relationship between body weight and labor market outcomes in Germany in this detail.

While many studies use parental socioeconomic status and health to predict in part their children's future prospects, the first empirical chapter investigates the dynamic relationship of child health with maternal health and socioeconomic factors over time. Using a series of latent growth curve models, this study examines the association between trajectories--or intra-individual models of stability and change--of child activity limitations and trajectories of maternal limitations in work due to health, labor force participation and household poverty status. The second analytic chapter examines and accounts for potential methodological biases in the relationship of child health with maternal health and socioeconomic factors in three ways: autoregressive cross-lagged models that address the reciprocal relationship between child health and maternal outcomes, fixed effects analyses that control for individual characteristics that are constant over time to limit omitted variable bias, and a latent class analysis to address in part both omitted variable bias and measurement error. The third analytic chapter examines the concordance of mothers' and children's reports of children's general health status, the concurrent validity of both mothers' and children's reports of children's general health status, and whether the association between child and maternal health depends on who reports children's health status.

This dissertation unifies research on stratification in socioeconomic and health outcomes with the life course and stress process perspectives, by pushing traditional stratification research to take seriously the idea that individual lives are linked lives and applying unique methodological approaches that account for the linked lives of mothers and children at the point in the life course when the child lives at home.

This study differed from previous studies in that it is archival and longitudinal. The participants came from the National Longitudinal Survey of Youth 1979, a nationally representative data base (NLSY79, 1997). The sample of the NLSY79 included 12,686 young adults first surveyed in 1979. These individuals are presently in their forties and have been surveyed for over twenty years. Participants' school and life experiences were recorded. The subsequent data allowed this researcher to study predictor variables most related to college entrance.

Multiple correlations and path analyses were utilized in this study. The correlations which were found to be statistically significant and accounted for the most variance included IQ and achievement scores positive correlation to college entrance. The direct paths that were statistically significant toward the achievement measure (i.e., Armed Services Vocational Aptitude Battery) included family poverty, race, locus of control orientation and higher self esteem. The achievement variable (i.e., ASVAB) accounted for the most variance with regard to participants staying in school and entering college. Results are discussed in terms of psycho-educational settings.

The second chapter aims to understand the relationship between mental health and labor market outcomes. We bound the impact depressive symptom severity has on both the probability of employment and on earnings.

The first chapter of my dissertation investigates the relationship between union coverage and work-related injuries. Trade unions play a central role in establishing a safe work place at the firm level. Union representatives ensure the implementation of safety standards set by the Occupational and Safety Health Administration (OSHA) and provide employees in bargaining units with education on job-specific risks. As a consequence, the injury rates of workers covered by union contracts may be expected to be lower than those of uncovered workers, given similar demographic and occupational characteristics. However, the work environment is not the sole determinant of injury rates. One additional factor is the worker's effort to avoid and minimize work-related hazards. When the personal accident-prevention activities cannot be monitored, the employee needs to face a correct system of incentives. If union representation fails to provide these incentives, covered workers may experience higher injury rates than uncovered employees. In analyzing this relationship, previous studies have failed to include controls for one or both of the following factors: (1) individual unobservable characteristics, and (2) firm-specific hazards. Using data from the NLSY79, the goal of the first chapter is to address these major limitations and provide an estimate of the effect of union coverage on the probability of nonfatal work-related injuries. The results of this chapter indicate that covered workers are about 50 percent more likely to suffer an occupational injury than similar uncovered workers. The direction of this estimate is then interpreted within a moral hazard framework, where union and nonunion workers face different incentives when they choose their level of accident-prevention activities.

The second chapter examines the economic consequences of a work-related injury. Injured workers may experience reductions in earnings for multiple reasons. First, injured workers may experience temporary or permanent decreases in work hours. Second, the wage of the worker may decrease because of lower post-injury productivity caused by any loss in general, specific or health human capital. This outcome is more likely to occur when the injured worker is forced to change firms, occupation or industry. In this setting, union coverage at the time of the injury may play an important role in reducing the size of the earnings losses. Unions provide protection against dismissal without just cause and may favor the assignment of an injured worker to a different job or work schedule within the same firm, therefore decreasing the probability of quitting. Using data from the NLSY79, this chapter applies a A standard version of the difference-in-difference methodology is used to investigate whether union coverage at the time of the injury is able to reduce the earnings losses caused by the injury. The findings indicate that union coverage is able to preserve workers' earnings in the year of the injury.

The third chapter investigates whether gender discrimination affects the earnings in the arts occupations. Female artists generally earn less than male artists who work in the same occupation. This gender earnings gap can be due to a combination of differences in individual characteristics (explained portion) and differences in the returns to endowments (unexplained portion). The latter have been often interpreted as an estimate of discrimination, although they could be the result of gender differences in unobservable skills. Using the Five Percent PUMS from the 2000 U.S. Census, a modified Blinder-Oaxaca decomposition technique (Yun, 2005) is applied to the gender wage gap for eleven arts occupations. The results of this chapter show that a significant unexplained pay gap, which could be attributed to discrimination, exists for female artists employed in full-time jobs. The explained portion of the gap indicates that women are often segregated into low-paying industries. A second type of decomposition (Fairlie, 2003) is then used to investigate the reasons of gender segregation across industries. These additional findings suggest that discriminatory hiring practices are likely to exist for some arts occupations.

Using data from the ongoing biennial National Longitudinal Survey on Youth surveys, exploratory and confirmatory factor analyses were applied to empirically define parental monitoring using measures encompassing multiple facets of the construct. Parental monitoring was ultimately operationalized using a second-order confirmatory factor model, with four first-order factors (i.e. parental school involvement, communication, time involvement, rules/decision-making) supporting the definition of a 'set of correlated parenting behaviors' (Dishion & McMahon, 1998). Consistent with a transactional conceptual framework (Wills & Dishion, 2004), path analysis examined the direct and indirect longitudinal associations between parental monitoring, child disclosure, parental knowledge, and alcohol involvement among children and young adults.

Findings indicated that parental monitoring was a significant protective factor for females across a number of alcohol use measures, both directly and indirectly via child disclosure, maternal knowledge, and early alcohol initiation in the case of subsequent heavier alcohol use. In males, higher monitoring levels in middle childhood protected against alcohol-problem use in young adulthood. Child disclosure reduced the odds of binge drinking in females, controlling for negative peer pressure and maternal alcohol use.

Through proper monitoring practices, parents play an important role in reducing both short-term and long-term alcohol involvement in youth, particularly among females. Proper monitoring could help buffer the observed independent effect of negative peer pressure in early childhood on later youth alcohol use. Child disclosure was an important mediator that warrants further attention. The study provides further support for parenting influences on youth alcohol use and will help guide existing family-focused evidence-based programs aimed at reducing youth substance use and misuse.

Little evidence exists on the relationship between working hours and future wage growth. In the first chapter I examine this relationship and show a well-defined convexity: the positive effect of hours is stronger at high levels of labor supply. A four-wave panel survey of men and women who registered to take the GMAT between June 1990 and March 1991 is used to show that this relationship is especially strong for young professional workers, but it is also present in the more broadly representative 1979 cohort of the NLSY. As the underlying mechanism for this empirical regularity, I propose a job-ladder model in which workers differ in their preferences for leisure. The findings can be used to account for up to half of the gender gap in wage growth.

The second chapter links a worker's propensity to change jobs to the educational choices she makes. A model of the choice of graduate management program type based on job search theory predicts that more mobile workers are more likely to enroll in a full-time Master of Business Administration program. The chapter also adds to the literature on employer-provided general training; the model predicts that employers are more likely to provide tuition assistance to workers who find quits costly. I use the survey of GMAT registrants to show that these predictions hold true empirically.

In the third chapter I focus on the effect of student debt on marriage and fertility. Using the GMAT Registrant Survey, I find that higher amounts of business school loans decrease the probability of observing marriage or births, particularly for men. Other sources of financing like employer-provided tuition assistance do not have such an effect. I also show evidence that expectations about fertility do not influence observed student debt.

In the second chapter entitled "Self-esteem, Educational Attainment and Wages: A Question of Selection," I use the National Longitudinal Survey of Youth 1979 again to explore the relationship between self-esteem and wages found in the first chapter more closely. I find that self-esteem partially estimates selection into higher levels of education. Conditional on this selection, the remaining direct effects of self-esteem on wages are negligible. This evidence suggests that self-esteem affects wages indirectly through educational attainment.

I find that gender intersects with race, class, and occupational status and leads to a larger fatherhood wage premium for relatively advantaged men (married, whites, professionals, or the college educated). For married white men, one child is associated with a $7,300 increase in annual earnings. For married black men, one child is associated with a $3,100 increase in annual earnings. Although married white men earn a premium for each additional child, married black men pay a penalty for having more than two children. Married white men also spend more time at work on the birth of a child, whereas married black men do not. These outcomes remain robust in fixed effects and instrumental variable models. They likely reflect the gender division of labor as well as employers' preferential treatment of fathers over childless men.

Over the past three decades married white men have experienced a significant reduction in their fatherhood earnings premium. Married black men have not experienced any change in the small premium that they earn for having one and two children, and they have experienced an increase in the penalty that they pay for having a third child. The rise of racial inequality between married white and black fathers parallels the erosion of black men's employment stability. As gender inequality between mothers and fathers subsided over the past thirty years, racial inequality among married fathers has increased.

I address this question through a mixed methods study of the experiences of Black and Latino young adults in labor market intermediaries (LMIs). Employers and jobseekers often rely on LMIs to make job matches. I combine qualitative data on four non-profit intermediaries that target Black and Latino jobseekers with national-level, quantitative data on intermediary use and job outcomes.

I found that specific LMIs--e.g. schools and public agencies--help these ethnic groups get jobs. Intermediaries can also influence earnings and job satisfaction. Additionally, I found that the effect of using intermediaries does not always accrue equitably. Using schools to find jobs yielded cross-ethnic gains in earnings while using family and friends only harmed Blacks. Next, I conducted an in-depth qualitative study of an organization serving low-skilled ethnic minorities. I found that through an LMI, this group acquired work-relevant cultural capital, and I specify in this dissertation how that socialization process occurred. Finally, I compared the job placement experiences of Blacks and Latinos across four minority-targeted intermediaries. I found that the programs gave elite college graduates an edge in competing for elite jobs, placed non-elite college graduates in the running for elite jobs, and brought high school graduates into the race for good jobs. These stratified transitions revealed that low-skilled workers faced relatively more turbulent pathways to career entry and needed continued support beyond the program period. I concluded that some labor market intermediaries facilitate upward mobility for minorities even if that mobility may not be enough to produce broader racial and social equality.

The first chapter investigates the relationship between wages and the size of previous employers. Despite the plethora of papers examining the firm-size wage premium, there is little attention to potential long-term effects of firm size on wages. This paper documents a positive association between current wages and previous firm size, using data from the National Longitudinal Survey of Youth. The magnitude of this relationship is larger among workers that are more able. These workers are also substantially more likely to be working in large firms. I develop a dynamic model where workers choose the size of their employer to maximize career earnings. Large firms offer the opportunity to specialize, which increases the rate of learning. The model induces self-selection based on human capital into large firms and higher career wages for those that start in large firms.

The second essay develops a framework where some firms employ up-or-out rules, while others do not. Up-or-out rules are optimal for firms when they must screen new associates to find partners and have an incentive to maintain partnership quality. I show that when workers have heterogeneous effort costs, the higher effort cost workers will self-select into firms that do not use up-or-out rules. A costly investment, in the form of effort as an associate, is required to reveal productivity as a partner. Workers with high effort costs do not find the investment worthwhile, thus, they will not make partner and select firms where they can retain their firm-specific human capital.

The final essay looks at the incentive effects associated with up-or-out rules. I describe a model with heterogeneous abilities among associates. Firms observe a noisy signal of output based on associate ability and effort. Using permanent associates as well as partnership, the firm is able to induce the optimal effort from all types; up-or-out rules alone cannot accomplish this. Use of a consolation prize requires the firm to increase the rewards of partnership, to prevent high ability workers from pooling with lower ability workers.

I then test these predictions empirically. Because worker safety investments are difficult to measure, I take an indirect approach. Previous studies find that body weight is positively associated with the risk of accidents (injuries), suggesting workers can invest in safety by maintaining a lower weight. Thus, using body mass index (BMI) as a proxy for such investment, and the 1979 National Longitudinal Survey of Youth (NLSY), I estimate the effect of high BMI on the risk of experiencing a workplace injury. I find that obesity increases the risk of injury by 22 percent. However, although non-obese workers have higher safety-related productivity, this is probably not (entirely) due to purposeful investment.

I then use the 1979 NLSY and job risk data from the Bureau of Labor Statistics to examine whether worker investments in safety (safety-related productivity) increase wages. Comparing wage differences in high and low risk jobs for low- and high-BMI workers, I find that obese workers earn a 5 percent smaller return to risk than their non-obese counterparts, which is consistent with the hypothesis.

Finally, in a similar analysis but focusing more on the effects of weight per se, I assess whether a correlation between weight and wages can be explained by safety-related productivity. I find that obese males in high risk jobs earn 5 percent lower wages than their non-obese counterparts due to their lower safety-related productivity. For females, between 49 and 78 percent of the wage effects of obesity yielded by cross-sectional estimates can be explained by lower safety-related productivity, and the remainder by individual heterogeneity.

Materials and methods . Two longitudinal datasets are utilized to address this potential causal association. The first is an ongoing longitudinal survey of 12,686 males and females beginning in 1979 from the National Longitudinal Survey of Youth (NLSY), a nationally representative sample from the United States. In the 1994 follow-up interview, 8,759 adults (age range 29-37 years) had complete data on past-year adult cannabis use and current depression. Individual's probability to use cannabis was predicted through a propensity score approach using over 50 baseline covariates. The second dataset is from an observational prospective cohort study of 2,311 first-grade children enrolled in 1985-1986 as part of the Johns Hopkins University, Prevention Research Center (PRC) randomized trial of classroom-based preventive interventions. In the young adult follow-up interview, 1,494 adults (age range 19-24 years) had complete data on early-onset cannabis problems and young adult major depression. Both studies utilized new causal inference statistical techniques, known broadly as propensity score techniques. Observed confounding covariate differences were controlled through the estimation and application of propensity score techniques. Numerous propensity score techniques exist, yet few guidelines are available to aid researchers in choosing the best technique for a specific dataset and research question. Propensity score estimation and application techniques are described and a taxonomic approach is proposed to guide selection of the best techniques for these data.

Results . In the NLSY dataset, after using propensity score adjustment, the odds of current depression among past-year cannabis users was only 1.1 times higher than the comparison group (95% Confidence Interval (CI): 0.8, 1.7). After applying the best propensity score technique to use for the PRC dataset, the risk of young adult depression for the recent-onset cannabis problem using females was not statistically significantly different from the risk among the comparison females (Odds Ratio (OR): 0.68, 95% CI: 0.20, 2.34). The risk of young adult depression among the early-onset cannabis problem using males was positive, but still not statistically significantly different from the comparison males (OR: 1.72, 95% CI: 0.77, 3.60).

Discussion . Similar inferences may be drawn from both studies testing the potential causal link between cannabis and depression. After adjusting for differences in baseline confounders of cannabis use and depression, past-year cannabis use was not a significant predictor of current depression among adults. For adolescents, although the estimated association was higher for males, the qualitative difference in risk for males and females and the lack of statistical significance for either gender did not support claims of a causal association.

Conclusion . These data do not support the hypothesis that there is a causal link between cannabis and depression.

The first chapter uses in-depth interviews with 61 junior and senior high school girls to show how social class shapes educational and occupational aspirations and plans through the availability and use of social networks. These interviews reveal that middle class adolescents are embedded in resource-rich social networks that facilitate high educational and occupational attainment while limited social ties, family instability, and parental disengagement produce disadvantages for working class and poor youth. The second chapter uses survey data from the National Longitudinal Study of Youth 1979 (NLSY79) and National Educational Longitudinal Study (NELS) to explore the relationship between events in the transition to adulthood and fulfillment of one's educational and occupational expectations. Findings reveal that the order and timing of family formation and dissolution events can disrupt young people's paths to attainment in early adulthood. The final chapter uses NLSY79 and NELS datasets to test the relationship between falling short of one's expectations and emotional and psychological outcomes in early adulthood. Results indicate that occupational expectations can serve as baseline standard with which to judge later accomplishments--falling short of these goals leads to lower emotional and psychological well-being in adulthood. These findings support the claims of relative deprivation theory, which argues that dissatisfaction arises from the gap between what one has and what one wants.

In the first part of this dissertation, I examine racial differences in SAH using the Older Men cohort of the Dept. of Labor's National Longitudinal Surveys. I find that blacks are more likely to report poor SAH than whites, but much of this difference is accounted for by socio-economic differences. For both blacks and whites, functional health is the primary correlate of poor SAH. However, leisure time and rural roots are significantly associated with poor SAH in blacks but not in whites. Still, the pervasive influence of functional health on poor SAH leads me to conclude that there is a high degree of comparability between black and white assessments of poor SAH. In the second part of the dissertation, I use poor SAH to characterize difference in the healthy worker effect by race. I find that the difference in the proportion with poor SAH between workers and non-workers is about twice as large in the black cohort than in the white cohort. However, much of the difference is accounted for by socio-economic status. However, I also find that after adjusting for SES, whites unable to work are about 1.4 times as likely to report poor SAH than blacks unable to work.

The use of SAH as a response in epidemiological studies of men where race is a factor appears to be justified when results are adjusted for SES. The use of SAH as a response in occupational studies of men where race is a factor is also justified when results are adjusted for SES as long as the cohorts studied exclude those unable to work. Among those unable to work, there are differences in how white males and black males report poor SAH.

Findings revealed that GED recipients' contexts and experiences in adolescence were characterized by greater risk exposure (e.g., gang involvement, teen parenthood) in comparison to high school graduates. Collectively, these differences revealed that more risk factors were present in GED recipients' lives long before they dropped out of high school. Even after accounting for a range of factors from multiple ecological domains in adolescence, GED recipients still had significantly lower household income-to-poverty ratios and lower rates of postsecondary educational attainment than their high school graduate counterparts. However, their general and behavioral health (e.g., alcohol misuse, exercise behaviors) outcomes were either equivalent or better than high school graduates' outcomes. Looking within the GED recipient group, recipients who eventually completed a postsecondary credential differed in adolescence from those who did not on a number of factors, such as engagement in risky behavior and parental education.

Overall, findings suggest that differences between GED recipients and high school graduates existed in a range of ecological domains in adolescence and accounting for these differences attenuated differences in these two groups' outcomes in adulthood in some cases but not in others. Findings also suggest that GED recipients with a postsecondary credential had different adolescent experiences than recipients without a postsecondary credential. These results signal points for prevention, early intervention, and support for recipients post-GED receipt.

The second paper, "The Influence of Height on Academic Outcomes," explores the relationship between height and educational achievement. Taller students earn higher grades, are more likely to graduate from high school, and attain more years of schooling. However, this height effect varies across school size: height has a stronger correlation with outcomes for students from bigger schools. These results imply that intelligence alone does not drive the association between height and outcomes. Differential access to activities such as school sports plays a role in driving these results.

Finally, the third paper, "The Effect of Menarche on Education: Explaining Black-White Differences," looks at the relationship between developmental timing in girls and their educational outcomes. A girl's developmental timing, marked by age at menarche, is strongly correlated with educational attainment. However, the direction of this correlation varies by race. Among the white population, delayed menarche is associated with improved educational outcomes while among the black population, it is associated with worse outcomes. Different responses to pregnancy driven by variance in marriage market quality across race can explain part of this relationship. Absent pregnancy, differing marriage patterns across race may also contribute to the trends between menarche and achievement. Evidence suggests that endogenous factors such as health, which influence both development and outcomes, can not explain the varying relationship between age at menarche and education across race.

The second chapter furthers the first chapter's analysis by using the National Longitudinal Survey of Youth (NLSY) in order to reconcile the varying estimates of estimated wage gaps in the previous literature using the NLSY. Results suggest that differences in sample design as well as control variables used have a large impact on the size and sign of wage gap estimates. Estimates that take into account differences in test scores are typically positive in previous studies implying a Hispanic wage advantage. However, results presented show how accounting for state fixed effects produces negative estimates even with test score controls, further suggesting the importance of location in examining Hispanic wage outcomes.

The last chapter analyzes the effect of selection bias on estimated Hispanic-white women's wage gaps. Previous studies generally find the wage disadvantages experienced by Hispanic women are explained by relatively low levels of human capital. These results are based on observed wages of working women, however, and there are ethnic differences in who selects into labor force participation. Using the NLSY, results presented find that accounting for the wages of non-participant women widens the disparity in wage offerings between Hispanic and white non-Hispanic women than what is normally estimated using the wages observed by working women alone.

This study was conducted to evaluate the performance of stMLE in the estimation of LGM through a Monte Carlo simulation. Application of stMLE was also illustrated through estimation of LGM with math achievement test data from the National Longitudinal Survey of Youth. Results confirmed that nMLR can still produce biased parameter estimates when data are non-normally distributed. On the other hand, stMLE resulted in many estimation issues. Although stMLE presents a theoretically appropriate framework to estimate LGM with non-normal data, more research is needed to determine the conditions under which it performs well.

Six waves of data are derived from the National Longitudinal Survey of Youth 1997. Analyses employ a subsample ( n = 4753) of adolescents who were between the ages of 12-14 and engaged in delinquent activities at the first round of data collection. Logistic regression results indicate that the odds of engaging in delinquency increased if the individual is older, male, involved in a gang, disengaged from school, and exposed to more violence. A series of growth models focusing on the subset of adolescents who engaged in some level of delinquency suggest that individual, neighborhood, and family processes predict adolescent trajectories of delinquency. Specifically, females begin at higher initial levels of delinquency and have a slower rate of change, while minority status indicates starting at a lower initial level and a slower rate of change in delinquent activities. Measures of negative community influences suggest higher initial levels but slower growth in delinquency. Positive family processes indicate lower initial levels of delinquency. Overall, these predictors of delinquent trajectories are similar for males and females. Experiencing violence at an early age was the only characteristic that significantly differed between males and females, predicting increases in delinquency over time for males. Results suggest numerous avenues for research, intervention, and policy. Broader implications for policy, such as providing adolescents with opportunities to serve in crime prevention efforts in their communities, are also discussed.