This dissertation consists of three chapters. In the first chapter, I estimate wage effects of double majors and double degrees among a sample of college graduates in their early career, using the 1997 National Longitudinal Survey of Youth (NLSY97). I rely on selection on observables and control for individuals’ test scores, family background, and school characteristics when estimating the wage effects. I further consider whether wage effects of a double major/degree can be explained by two mechanisms: the “skill-enhancing” effect (increase in the depth of knowledge accumulated in college) and the “job-matching” effect (increase in the chance of working in an occupation that is more closely related to one’s college major). I examine whether estimated wage effects associated with a double major/degree (after controlling for confounding factors) decrease as a result of controlling for the depth of knowledge accumulated in college and the relatedness between college major and occupation. I find that having a double major does not make a significant difference in one’s early-career post-college wages. A double degree is estimated to be associated with a 0.088 increase in log wages after controlling for confounding factors. About a third of this effect can be explained by a combination of both the “skill-enhancing” and “job-matching” effects. In the second chapter, I use the NLSY97 to study whether being mismatched in the first job (meaning the individual’s occupation is not among the common occupations to which his/her college major typically leads) has a long-lasting effect on wages. I also investigate wage growth and job change patterns for different types of mismatched workers. I distinguish between demand-side mismatch due to job dissatisfaction and supply-side mismatch due to reasons other than reported job dissatisfaction. I find that both types of mismatched workers have significantly lower wages compared to matched workers, but that demand-side mismatched workers face a larger wage penalty than do supply-side mismatched workers. However, the wage penalty associated with demand-side mismatch reduces about 1.6 times as fast as does the penalty of supply-side mismatch as labor market experience increases. The result is that the estimated log-wage effect of mismatch virtually disappears in six years for both demand-side and supply-side mismatched workers, even though the former face a large wage penalty at the outset. Further, I show that demand-side mismatched workers tend to have more between-job mobility and between-job wage growth than matched workers, whereas supply-side mismatched workers tend to have more within-job mobility and within-job wage growth than matched workers. Overall, job mobility and subsequent wage growth contribute to the closure of the wage gap between matched and mismatched workers. My findings support predictions stemming from the job match literature that wage effects of first-job mismatches are not long-lasting. In the last chapter, I use NLSY97 data to determine the extent to which detailed measures of college-related factors, based on course credits and grades earned in different fields of study, explain the gender wage gap among college graduates in their early career. I start with a standard set of controls and then add my detailed measures of college-related factors to identify the increase in the explained gender wage gap. A decomposition of the gender wage gap reveals that the inclusion of detailed measures of college-related factors along with the standard set of controls increases the explained part of the estimated gender wage gap from 65.6% to 69.1%-77.8%. Among all the pre-market factors, detailed measures of college-related factors have the most explanatory power to the estimated gender wage gap (28.5%-39.1%). My findings imply that gender differences in credits and grades earned in different fields of study capture additional gender differences in skills that cannot be fully represented by gender differences in other factors such as college major and occupation. Compared to gender differences in college major and general academic achievement, gender differences in credits and grades earned in different fields of study are better pre-market measures for differences in skills between college-educated men and women.

Wage differentials across college majors are huge and have been increasing. The type of college education becomes important for college students in terms of future earnings. Understanding the treatment effect of major choice in a certain occupation is difficult because of the sorting behavior and the effect of occupation choice. In order to accomplish this, I provide a dynamic model that combines major choice with occupation choice. The simulation results illustrate that science majors earn 30% more if they choose jobs related to science. However, this high premium does not exist in all jobs. The major choice itself does not guarantee a high return. Occupation choice matters a lot in obtaining a higher premium. The second chapter proposes a dynamic model of college course and occupation choices, where individuals make human capital investment under imperfect information about the future return. Using simulation results based on this model, I investigate the role played by uncertainty in student choices. I contribute to the recent task-based heterogeneous human capital literature by adding choices made before individuals enter the labor market. By combining college transcript data and occupational knowledge requirement information, I match human capital with occupational tasks to better evaluate the labor market performance of college graduates. For tractability purposes, both human capital and occupational tasks are aggregated into two dimensions: STEM and non-STEM. Estimation results indicate that college courses have different returns at work, with STEM courses inducing relatively higher wages. When uncertainty is eliminated, individuals specialize more in STEM or non-STEM based on their comparative advantages. The change of specialization in STEM courses is bigger compared to non-STEM courses. Overall benefits of human capital specialization are more pronounced in top ranking colleges. Old-age medical expenditure risks have been documented to impose significant impacts on elderly savings. However, little is known about the consumption effects of elderly medical expenditure risks. In this study, we examine the effect of medical expenditure risk on elderly household consumption decisions. We identify the causal effect by exploiting the exogenous reduction in prescription drug spending risk as a result of the introduction of Medicare Part D in the U.S. in 2006. Using the Health and Retirement Study (HRS) data during 2004 - 2010, we find that declining medical expenditure risks had little impact on total consumption, regardless of nondurable or durable consumption.

In this dissertation, I use administrative databases from the state of Texas to address substantive issues in higher education policy. Chapter two, joint with Jesse Cunha, evaluates a novel program designed to increase college enrollment among students who are academically prepared for higher education but unlikely to enter on their own. The program sets up "GO Centers" that provide information about college, influence students' beliefs about the benefits of college, expedite the application process, and facilitate peer support for college. We estimate the program's impact on college application, acceptance, enrollment, and persistence rates using a differences-in-differences methodology where a propensity score matching procedure is used to identify schools that were in fact untreated but which closely matched the program's criteria for treatment. The findings indicate that the program had a large positive impact on all outcome measures, especially amongst Hispanic and low-income students. Chapter Three, also joint with Jesse Cunha, explores issues surrounding the measurement of the value-added by individual colleges and offers preliminary value-added estimates for all public colleges in Texas. Our problem differs from that of the primary and secondary educational system as college students specialize their instruction by choosing both school and major. Because wages are a measure of productivity, we argue for the use of labor market returns as a measure of value-added. Using administrative data from Texas, we estimate the labor market return to attending each of the 33 individual public colleges in the state. We present unconditioned estimates and find that labor differences are large. We then control for a rich array of observables that might be correlated with college choice, including demographics, SAT scores, parental income and education, local labor market conditions, high school GPA and course-taking patterns, and college major, and the vectors of college application decisions and subsequent acceptances. Upon conditioning, labor market returns across colleges tend to converge, yet significant differences remain. In Chapter 4, I examine the validity of distance to college as an instrument for educational attainment in earnings models. Numerous studies have used distance to college to instrument for educational attainment in models of the economic return to higher education. The assumption is that, conditional upon the included covariates, distance impacts educational attainment, but not earnings. One key problem with this assumption is the possibility that college location may be nonrandom. In this study, we use a large administrative database from the state of Texas to investigate the appropriateness of the use of distance to instrument for educational attainment and college choice in models of the labor market return to higher education. I show that colleges tend to locate in urban areas with robust local labor markets, and that IV models are highly sensitive to the exclusion of such information. I also show that distance instruments are highly sensitive to functional form in the first stage model.

This dissertation examines several policies in secondary and tertiary education and improves on existing methods of inference in settings that are salient for policy evaluation. In the first chapter, I propose a hypothesis test for clustered samples. This test is exact in samples with few clusters, few ever-treated clusters, cluster size outliers, or treatment intensity outliers; these features cause previous tests to over- or under-reject true hypotheses. I derive my test by inverting the distribution of the test statistic under a standard assumption about the errors, so that critical values can be selected from a distribution that matches the test statistic. I use Monte Carlo simulations to demonstrate where this adjustment is most impactful in achieving exact tests compared to previous hypothesis tests, and I apply my test to an empirical setting from the literature. The second chapter, previously published in Education Finance and Policy and co-authored with Lauren Schudde, examines the impact of a developmental education (dev-ed) reform for community colleges. Dev-ed aims to help students acquire knowledge and skills necessary to succeed in college-level coursework. The traditional prerequisite approach to postsecondary dev-ed—where students take remedial courses that do not count toward a credential—appears to stymie progress toward a degree. At community colleges across the country, most students require remediation in math, creating a barrier to college-level credits under the traditional approach. Corequisite coursework is a structural reform that places students directly into a college-level course in the same term they receive dev-ed support. Using administrative data from Texas community colleges and a regression discontinuity design, we examine whether corequisite math improves student success compared with traditional prerequisite dev-ed. We find that corequisite math quickly improves student completion of math requirements without any obvious drawbacks, but students in corequisite math were not substantially closer to degree completion than their peers in traditional dev-ed after 3 years. The third chapter, coauthored with Anjali Priya Verma, examines students who were removed from their regular instructional environments for disciplinary reasons and sent to disciplinary schools. We study the long-run effects of disruptive peers on educational and labor market outcomes of students placed at these institutions. The existing literature documents that students who are removed from their regular instructional setting and placed at disciplinary schools tend to have significantly worse future outcomes. We provide evidence that the composition of peers at these institutions plays an important role in explaining this link. We use rich administrative data of high school students in Texas which provides a detailed record of each student’s disciplinary placements, including their exact date of placement and assignment duration. This allows us to identify the relevant peers for each student based on their overlap at the institution. We leverage within school-year variation in peer composition at each institution to ask whether a student who overlaps with particularly disruptive peers has worse subsequent outcomes. We show that exposure to peers in the highest quintile of disruptiveness relative to lowest quintile when placed at a disciplinary school increases students’ subsequent removals (5-8% per year); reduces their educational attainment--lower high-school graduation (6%), college enrollment (7%), and college graduation (17%); and worsens labor market outcomes--lower employment (2.5%) and earnings (6.5%). Moreover, these effects are stronger when students have a similar peer group in terms of the reason for removal, or when the distribution of disruptiveness among peers is more concentrated than dispersed around the mean. Our paper draws attention to an unintended consequence of student removal to disciplinary schools, and highlights how brief exposure to disruptive peers can affect an individual’s long-run trajectory

This dissertation consists of three studies that shed light on the ongoing transformation of higher education's role in producing inequality and transmitting advantages across generations in the United States. The first chapter examines the most educated Americans: graduate and professional degree holders. The subsequent two chapters, by contrast, shift focus to young adults' transition into higher education, examining how schools and local labor markets shape racial inequality in the transition from high school to college.The first empirical chapter examines horizontal stratification among graduate and professional degree programs and their connection to the new economic elite. Compared to the baccalaureate level, there has been relatively little empirical research on distinctions among graduate and professional degrees and how they relate to labor market inequality. I add to this emerging literature with 30 years of linked survey data containing an unprecedented level of detail on the lives of the most educated Americans. I track recent historical changes in who attains top-ranked MBAs, JDs, MDs, and PhDs, finding a marked increase in the influence of parental education on elite degree attainment. This novel evidence suggests the solidifying of an intergenerational class of highly educated professionals in the United States. Second, I explore the earnings returns to program rank across different degree types, and by gender and parental education, with a particular focus on the top percentile of the earnings distribution. Unlike at the baccalaureate level, the earnings returns to prestige vary significantly across fields, such that they are much higher in MBA and JD programs than research doctorate or medical programs. I also find that the earnings returns to prestige are higher for children from less-educated families, suggesting a potential equalizing effect of elite postbaccalaureate programs. The second empirical chapter examines how local labor markets shape college attendance behavior differently by race and gender. A long-standing sociological literature has established that white students are substantially less likely to attend four-year colleges than are Black students with similar socioeconomic resources and academic performance. Drawing on accounts of racial labor market segregation among workers without bachelor's degrees, I hypothesize that racialized and gendered access to good sub-baccalaureate jobs-for instance, jobs in the trades-may account for racial differences in college attendance. I test this hypothesis empirically using administrative data on students attending high school in Wisconsin, examining net racial differences in college attendance across labor markets with varying degrees of racial occupational segregation. I do not find clear support for my hypothesis. However, I do find that white boys are more likely than Black boys to attend two-year colleges in places with more racially segregated labor markets. This finding suggests that a net-White advantage in vocational education pathways parallels the net-Black advantage in four-year college attendance, and provides some support for the hypothesized labor market mechanism. The third empirical chapter, co-authored with Christian Michael Smith, examines how high school course enrollment policies and school officials' decision-making affect racial inequality in high school tracking on the path to college. Prior work in sociology has produced conflicting evidence on whether and to what extent school officials' decision-making contributes to these patterns. We advance this literature by examining the effects of schools' enrollment policies for Advanced Placement (AP) courses. Using a unique combination of school survey data and administrative data from Wisconsin, we examine what happens to racial inequality in AP participation when school officials enforce performance-based selection criteria, which we call "course gatekeeping." We find that course gatekeeping has racially disproportionate effects. Although racialized differences in prior achievement partially explain the especially large negative effects among students of color, course gatekeeping produces Black-white and Hispanic-white disparities in participation even among students with similar, relatively low prior achievement. We further find that course gatekeeping has longer-run effects, particularly discouraging Black and Asian or Pacific Islander students from attending highly selective four-year colleges.

An introduction to the principles of microeconomics and macroeconomics that establishes strong links between theoretical principles and real-world experience, while incorporating clear and consistent international focus throughout the text.

My dissertation consists of three essays in empirical labor economics which are self-contained and can be read independently of the others. The first essay, coauthored with Professor Modestino, measures mismatch unemployment in US economy in the post-recession era and explores the heterogeneity among educational groupings. The second essay estimates the changing effects of cognitive ability on wage determination of college bound and non-college bound young adults between 1980s and 2000s. The third essay, coauthored with Professor Dickens, examines the impact of measurement error in survey data on identifying the extent of downward nominal wage rigidity in US economy. Essay I: No Longer Qualified? Changes in the Supply and Demand for Skills within Occupations-- In this study, we extend the framework developed by Sahin et al. (2014) to measure mismatch unemployment since the end of the Great Recession and explore the heterogeneity among educational groupings. Our findings indicate that mismatch across two-digit industries and two- digit occupations explain around 17- 20 percent of the recent recovery in the US unemployment rate since 2010. We also capture movements in employer education requirements over time using a novel database of 87 million online job posting aggregated by Burning Glass Technologies and further show that mismatch is not only greater in magnitude for high-skill occupations but also is more persistent over the course of the recent labor market recovery, possible accounting for the shift rightward that has been observed in the aggregate Beveridge Curve by other researchers. Furthermore, we shed light on at least one of the potential causes of mismatch on the demand side, providing evidence that labor demand shifts among high-skilled occupation groups exhibit a permanent increase in the share of employers requiring a Bachelor's degree as well as other baseline, specialized, and software skills listed on job postings, suggesting a role for structural shifts associated with changes in technology or capital investment. Our results demonstrate that equilibrium models where unemployed workers accumulate specific human capital and, in equilibrium, make explicit mobility decisions across distinct labor markets, can mean that workers are chasing a moving target-at least among high-skilled occupations. Furthermore, our findings inform debates focused on workforce development strategies and related educational policies where decision making could benefit from the use of real-time labor market information on employer demands to provide guidance for both job placement as well as program development. Essay II: The Changing Impacts of Cognitive Ability on Determining Earnings of College Bound and Non-College Bound Young Adults-- Using data on young adults from the 1979 and 1997 National Longitudinal Survey of Youth, I investigate the changing impact of cognitive ability, as captured by performance on AFQT tests, on wage determination of college bound and non-college bound young adults. My findings indicate that cognitive ability plays a substantially diminished role for the most recent cohort and its impact on wage determination has undergone a drastic change between 1980s and 2000s. My results tend to corroborate the findings of previous studies which emphasize the lifecycle path of technological development from adoption to maturation and trace back the labor market outcomes observed over these periods to pre- and post-2000 patterns in technology investment and its consequent boom-and-bust cycles in the demand for cognitive skills. Essay III: Measurement Error in Survey Data and its Impact on Identifying the Extent of Downward Nominal Wage Rigidity-- In this study, we employ data drawn from the 1996, 2001, 2004 and 2008 panels of the SIPP, which cover the years 1996-2013, to assess the effectiveness of dependent interviewing at reducing bias in the estimates of the extent of downward nominal wage rigidity in the US economy. In the 2004 and 2008 panels of the SIPP, dependent interviewing was used much more extensively than in the past. This questioning method by focusing on changes rather than levels of wages and using responses from prior interviews to query apparent inconsistencies over time reduces the incidence of reporting and measurement errors. Our change-in-wage distributions derived from SIPP 2004 and 2008 panels exhibit remarkably larger zero-spikes and asymmetries vis-℗♭℗ -vis those derived from 1996 and 2001 panels before dependent interviewing was used. These results are consistent with the findings of previous studies that used payroll data or statistical techniques to correct for reporting error. We apply one such technique to the SIPP panels before and after the introduction of dependent interviewing. In the pre-2004 panels the correction is large and results in a distribution that closely resembles the uncorrected distributions of the 2004 panel. When the correction is applied to the 2004 panel no evidence of errors is found.