<p>This dissertation consists of two separate essays on economics of education. First, the role of teacher-student interactions is analyzed. Teacher effectiveness is generally characterized by a single effect that is common across students. However, educators are multi-task agents that choose how to allocate their efforts among pupils. Some teachers may target their courses towards the top students in the class while others to the bottom, leading to different complementarity effects. Moreover, the introduction of accountability programs, such as No Child Left Behind (NCLB), could induce a reallocation of teacher's efforts, affecting the dynamics of student-teacher interactions. This study shows that the role of complementarities is key from a policy perspective. In this regard, an analytical framework and a novel iterative algorithm are implemented in order to characterize and quantify these effects. Results indicate that interaction effects played a crucial role in shaping the distribution of student achievement, especially after the implementation of NCLB. While more than half of the total gains in test scores experienced by the bottom third of the student achievement distribution post NCLB are due to adjustments in teacher-student complementarities, those with the very highest abilities have seen decreases in their performance.</p><p>In the second essay, gender disparities in educational attainment are explored across races. The sizable gender gap in college enrollment, especially among African Americans, constitutes a puzzling empirical regularity that may have serious consequences on marriage markets, male labor force participation and the diversity of college campuses. For instance, only 35.7 percent of all African American undergraduate students were men in 2004. Results show that, while family background characteristics cannot account for the observed gap, proxy measures for non-cognitive skills are crucial to explain it.</p> / Dissertation

<p> This dissertation examines the intergenerational effects of maternal early health, the intergenerational effects of maternal education and the distributional effects of school size. </p><p> Chapter 1 is an introduction that summarizes the contributions made in this dissertation. Chapter 2 examines a new question with important implications: Does a mother's early health affect her child's human capital development? My coauthor and I use two extremely different and established methodologies to identify variation in mothers' early health: variation in early life disease environment and variation in early life economic environment. We connect children to the environments experienced by their mothers using the state, month and year of maternal birth that appears on each child's birth record. To identify children's outcomes later in life, we connect their birth records to their 3^rd through 10^th grade school records using a high quality algorithm that relies on first and last names, exact dates of birth and social security numbers. We find that a one standard deviation improvement in maternal early health improves 10^th grade test performance in the following generation by .07 to .08 standard deviations. </p><p> Chapter 3 examines the intergenerational effects of maternal education. My coauthor and I use variation in compulsory schooling laws across states and over time to identify exogenous variation in maternal education; we estimate local average treatment effects using Two-Stage Least Squares instrumental variables estimations. We connect children to the environments experienced by their mothers using the state, month and year of maternal birth that appears on each child's birth record. To identify children's outcomes later in life, we connect their birth records to their 3^rd through 10^th grade school records using a high quality algorithm that relies on first and last names, exact dates of birth and social security numbers. We find an additional year of maternal education improves 3^rd grade test performance in the following generation by .31 standard deviations on average and that this relationship is driven by children born to white mothers. </p><p> Chapter 4 uses state-wide, student-level data from Illinois to examine the distributional effects of school size. I apply two established strategies to identify variation in school size; I use population-level panels of data to identify year-to-year changes in enrollment within schools and I exploit variation induced by school openings. I find smaller schools simultaneously improve average ACT achievement in 11^th grade and close achievement gaps between more and less advantaged students. Specifically, a 20 percent decrease in school size improves students' ACT performance by 1 percent on average and improves ACT performance by 1.5 percent on average for African American students that receive free or reduced price lunch.</p>

While federal deregulation of the trucking industry had little impact upon the truck drivers serving the Acadian oilpatch, recent legislation deregulating intrastate transportation yielded vast changes in the structure of the occupation. In the past, success as a trucker in the oilpatch depended upon an individual's entrepreneurial drive, as well as the social and familial networks upon which that individual could rely. Intrastate deregulation allowed several truck companies to dominate the industry; these companies grew via a complex series of alliances between transportation companies, service companies, and large oil concerns. These alliances disrupted the process by which individuals transformed social capital into economic capital. The foremost impact of these changes is a rapid drop in trucker's income---many now exist on the brink of insolvency. At the same time, the period of crisis has opened the sector to previously inconceivable options, including forays toward unionization, as well as the entry of women, blacks, and outsiders into the labor pool.

This dissertation consists of three papers relating to the field of Law and Economics. The first two papers examine the impact of increased judicial discretion on both racial disparities and inter-judge disparities in the federal criminal justice system. The third paper analyzes the effects of OSHA programs on workplace safety, wages, and employment. The common thread throughout this work is a focus on how legal actors and institutions affect substantive outcomes of individuals. / Economics

This dissertation explores the interplay between industrial racial hiring practices and the following institutions and transitions characterizing the inter-war period: unionization, institutional change among unions, business cycle activity, government anti-discrimination policy, and high-wage policies. The degree to which industrial racial hiring practices differed across manufacturing and mining industries and the impact that this industrial segregation had on black workers is explored. During World War I, when many northern employers first hired black workers, there was a significant difference in how black and white workers were distributed across industry. However, the segregation decreased significantly over time and it was not a contributor to the black-white income differential among industrial workers. Black workers were not employed disproportionately by industries with low wages, with low capital-to-labor ratios, or that were disproportionately dangerous. However, industrial segregation exposed them to greater unemployment risk, explaining a portion of their disproportionately high unemployment rates. The third chapter identifies some of the forces that shaped and mitigated industrial segregation. The way that black workers were distributed across industries was a function of union density, union affiliation, and tight wartime labor markets. The craft based unions affiliated with the American Federation of Labor were notorious for discriminating against black labor. The industrial unions affiliated with the Congress of Industrial Organizations (CIO) actively promoted the organization of black labor. Consequently, the mean probability that a randomly selected employee in an industry was black was negatively associated with general unionization and positively associated with CIO affiliated unionization. A government agency explicitly created to aid black workers in obtaining employment in defense industries during World War II, the Fair Employment Practice Committee, did not have a significant impact on industrial segregation. The fourth chapter of the dissertation assesses the impact that the high-wage policies of the Great Depression had on black unemployment. During the inter-war period, increases in workers' share of company revenues and unionization increased black workers' share of cyclical employment. By successfully increasing these factors, the Great Depression high-wage policies caused a disproportionate share of the employment downturn to be allocated to black workers.

The purpose of this study is to improve understanding of the reasons women are less likely than men to choose to study collegiate-level physical science and engineering and why women have lower rates than men of working in the physical science and engineering occupations. The theoretical frameworks used to examine these questions are self-efficacy, as formulated by the psychologist Albert Bandura, and peer influence, as suggested by the anthropologists Holland and Eisenhart: It is important to note that self-efficacy and peer influences evolve throughout the lifetime, and differences in genders began to diverge dramatically at adolescence. This study, however, is primarily concerned with post-secondary outcomes and recommendations. The National Longitudinal Survey of Youth 1979-1993 (NLSY) with an N = 12686 was utilized to create the database for this study. The analysis used an econometric method, multinomial logit analysis, to infer which of 30 some independent factors affect the mutually exclusive outcomes of majoring or working in other than the sciences and engineering, majoring or working in the biological sciences, and majoring or working in the physical sciences or engineering. The independent factors were those suggested by previous readings of the literature, e.g., demographics and high school attainment variables, as well as those additional independent factors available through the NLSY that pertain to self-efficacy and peer influence. The findings indicate that strong evidence exists to support both self-efficacy and peer influence. The results suggest convincing linkages between self-efficacy and the eventual major and occupation in the physical science and engineering. This study also reveals that peer influences are especially important in developing college major and career aspirations of girls and women.

This project examines how changing organizational arrangements in a technological field affect individual level outcomes and processes of career formation. In the field of the life sciences, the biotechnology industry has emerged as an employment option with a fundamentally different organizational form. Three main research questions are addressed concerning the changing organizational setting of life science careers: (1) How are traditional stratification of science patterns affected by the option of employment in network rather than hierarchical, organizations? (2) Who enters a new, sought after, employment arena first? and (3) How does a new career path become legitimate? The data collected for this project are both quantitative and qualitative. The quantitative data were gathered from National Institutes of Health archives. Logistic regression analyses were performed on the sample of 3395 PhDs to estimate dichotomous career outcomes. The qualitative data come from interviews and ethnographic observations with scientists in a variety of settings--university laboratories, commercial firms, and government institutes. While traditional patterns of stratification in science--educational background and gender--were found to have effects in this sample as well, organizational context is very important to understanding how stratification may be mitigated. Gender inequality in the attainment of leadership level positions was consistently found in more hierarchical organizational settings, but did not appear in network organizations (biotechnology firms). In contrast, educational background had significant effects across all types of organizational forms. PhDs with elite educations were more likely to enter biotechnology both in earlier and later periods of industry history. Male and female PhDs were equally likely to enter the biotechnology industry, and this result also did not vary by time period. The common frames used by scientists in biotech and other science-based organizations to legitimate biotechnology work include: resources (scientific as well as monetary), networks (ties to respected scientists who endorse biotech), and analogies to academe. Biotechnology employment is retroframed as similar to yet different from academic work---indicating some interesting frame tension. This study has implications for scholarship particularly in the areas of organization theory, sociology of science, and gender and work.

This research questions two fundamental assumptions of established educational policies designed to promote gender equity. First, that the external labor market is the principal predictor of disciplinary salaries when all other factors are controlled, and second, that integration of women into these marketable disciplines will result in pay equity. This study describes the national trends in female participation and gender redistribution in academic disciplines, evaluates student gender composition as a proxy for "feminization" of academic fields, and examines the value of comparable worth and labor market variables in tandem to predict faculty salary increases by discipline. Correlation, chi-square and logit analyses were performed to determine the direction of gender redistribution among disciplines over time, and to address the relationship between the level of disciplinary gender composition change and the level of disciplinary salary change. The annual percentage change (logged) in full-time assistant professor salary by discipline and institution was regressed on the proportion of female students within disciplines, the distribution of male students among disciplines, NRC rank, and prior year salary (logged). Gender redistribution among disciplines is not arbitrary and changes in gender composition can predict the level of disciplinary salary increases. The full regression model was significant. The variables for female participation tended to have a significant negative influence, while the variables for male participation tended to have a significant positive influence on changes in faculty salary.

This dissertation explores on-the-job leisure choice in individual labor supply environments and in work teams. The individual static labor supply theory implies that income-compensated wage increases will increase hours of work--positive substitution effects on work hours. While positive substitution effects are the testable implication of the theory, numerous empirical studies estimate negative substitution effects and therefore question the empirically validity of that model. I present a framework for the individual labor supply decision that theoretically explains negative substitution effects as the result of substituting on- and off-the-job leisure. An experimental environment in which subjects perform a task for piece-rate wages provides the data to test this theory. When on- and off-the-job leisure are choice variables, the results show that average substitution effects on hours worked are positive (implying that the classical theory is robust with respect to the assumption of no on-the-job leisure) but that some individuals display negative substitution effects. This provides evidence consistent with the substitution of on- and off-the-job leisure. When only on-the-job leisure is a choice variable, the data support another theoretical extension which predicts positive substitution effects on work effort. The results have implications for employers who might try to induce more work effort (less on-the-job leisure) through income-compensated wage increases. Work team decisions are also examined in the context of the voluntary contributions mechanism for public goods provision. Expected utility theory predicts that free-riding will dominate more efficient social incentives, and that uncertainty with respect to the provision of the public good will cause even more free-riding. An experimental environment confirms the existence of free-riding in this "uncertainty" environment, but results are mixed as to whether the free-riding is worse than in situations without the uncertainty. When the probability of public goods provision increases in group contributions, higher marginal incentives promote higher contributions. These results have implications for work team managers. If uncertainty lowers contributions, compensation based on effort instead of outcomes may raise effort. However, since higher contribution levels raise marginal incentives, any way in which a team manager could raise effort would be beneficial since it would promote high future effort.

This dissertation examines the impact of network effects in two settings--the computer software markets and self-employment decisions by individuals. Although there have been strong developments on the theory of network effects, relatively little empirical work has been done to examine their importance. The first part of this dissertation focuses on network effects in the market for computer software. It has been hypothesized that the presence of network effects in this market might often lead to lock-in of an inferior technology. An indirect test of this hypothesis is devised by taking advantage of a natural experiment afforded by the introduction of the programming language, Java. Java made it possible for programmers to write a single program that would run on any operating system. It therefore had the potential to eliminate the indirect network externalities in the operating systems market. Hedonic price regressions with fixed time and firm effects are estimated to test for the effect of Java on the extent of competition in the software market. Results using data compiled from magazine reviews of graphics applications programs indicate that Java was successful in creating more competition in the market for software applications. The second part of this dissertation examines whether social networks might explain the persistent racial gap in Self-Employment (SE) rates in the United States. Self-employment rates in the United States fell dramatically for most of the twentieth century before starting to increase in the 1970's. The racial gap in self-employment rates however, remained constant throughout this period. Many theories have been proposed in the literature but none of them successfully explains the persistence of the gap. A multinomial logit specification is used to model individual decisions to become self-employed. The average SE rate in the neighborhood is used as a measure of the network effect. Results indicate that social networks played an important role in promoting self-employment among blacks since 1950. Given the initial conditions of lower SE rates among blacks, the role of social networks in promoting SE might be able to explain the persistence of the racial gap in SE rates.