U.S. housing voucher holders pay their landlord a fraction of household income and the government pays the rest, up to a rent ceiling. We study how two types of changes to the rent ceiling aect landlords and tenants. A policy that makes vouchers more generous across a metro area benefits landlords through increased rents, with minimal impact on neighborhood and unit quality. A second policy that indexes rent ceilings to neighborhood rents leads voucher holders to move in higher-quality neighborhoods with lower crime, poverty and unemployment.

We estimate returns to school resources in the Jim Crow era, as measured by young males' 1940 wage earnings, occupational status, and cognitive aptitude scores. Results point to a 16 cent annual return on each $1 invested in public schools. To the question of whether some school inputs mattered more than others, we find comparable 25-32 cent returns per dollar invested in extended school years, teacher salaries, and smaller classes. School spending and inputs had much more bearing on labor market outcomes than aptitude scores. We document diminishing returns to school expenditures, which, in combination with segregated schools, resulted in higher returns to expenditures in black schools relative to white.

Background: The authors examined the factors associated with sex differences in earnings for 3 professional occupations. Methods: The authors used a multivariate Blinder-Oaxaca method to decompose the differences in mean earnings across sex. Results: Although mean differences in earnings between men and women narrowed over time, there remained large, unaccountable earnings differences between men and women among all professions after multivariate adjustments. For dentists, the unexplained difference in earnings for women was approximately constant at 62% to 66%. For physicians, the unexplained difference in earnings for women ranged from 52% to 57%. For lawyers, the unexplained difference in earnings for women was the smallest of the 3 professions but also exhibited the most growth, increasing from 34% in 1990 to 45% in 2010. Conclusions: The reduction in the earnings gap is driven largely by a general convergence between men and women in some, but not all, observable characteristics over time. Nevertheless, large unexplained gender gaps in earnings remain for all 3 professions. Practical Implications: Policy makers must use care in efforts to alleviate earnings differences for men and women because measures could make matters worse without a clear understanding of the nature of the factors driving the differences.

Homeownership continues to be the foundation of the pursuit of the American Dream. It is a critical driver of economic mobility, delivering financial and social advantages to families and entire communities. In this era of economic, technological, and societal change, understanding the state of homeownership today and what trends will influence homeownership in the future can help inform the discussions necessary to preserve homeownership opportunities for the next generation. There are many characteristics that influence the likelihood someone is a homeowner. When analyzing the homeownership rate, it’s important to understand why these characteristics matter to the tenure choice decision. In this case study, we develop a model that allows us to look specifically at the importance of education to the attainment of homeownership and the American Dream.

Research on health and time use has primarily focused on the total time individuals spend in particular activities such as sleeping. However, as research on sleep and night workers demonstrates when such activities occur and how they are incorporated into individuals’ lives may be related to health in important ways. In this analysis we apply an innovative method to time diary data that allows us to consider this alternative perspective. We draw on the 2014 American Time Use Survey to investigate the relationship between daily temporal pathways and indicators of health. We begin by using multinomial logit latent class analysis to identify daily temporal pathways that characterize how individuals organize their time. Preliminary results show that the temporal pathways are dominated by work and passive leisure. Next steps will include fine tuning our measures of physical activity and investigating how temporal pathways are associated with health.

Regulatory restrictions on housing supply lead to high prices and discourage workers from locating in productive cities. I study why regulation emerges and how it affects allocation of labor across space, wages, housing prices and aggregate productivity. I document that the dispersion of wages and housing prices across U.S. metro areas has increased and the sorting of college graduates into highly productive and expensive places has become stronger since 1980. I argue that these rising regional disparities have been amplified by the choices of residents of the most demanded metro areas to tighten regulation. To quantify this amplification effect, I build a general equilibrium model with multiple locations and heterogeneous workers. Local housing prices depend on regulation, which is decided by voting: renters want less regulation and owners want more. Faster productivity growth in some locations attracts workers and raises housing prices there. High-skilled workers, being less sensitive to rising prices, sort into productive and expensive areas. The growing prices in turn make homeowners vote for stricter regulation, which raises prices even more. This amplification effect of endogenous regulation choices is sizable. I calibrate the model to the U.S. and find that the rise of regulation accounts for 23% of the increase in wage differences across metro areas and the entire jump in house price dispersion from 1980 to 2007. Absent the rise of regulation, aggregate productivity would be 2% higher. Policy interventions that limit regulatory powers of local governments and reduce their incentives to regulate could lower housing prices and wage inequality, and lead to productivity gains.

Based on detailed occupation titles and making use of measures that do not require pair-wise comparisons, this paper shows that the occupational segregation of African American women declined dramatically in 1940-80, decreased slightly in 1980-2000, and remained stagnant in 2000-10. This paper quantifies the well-being losses that African American women derive from their occupational sorting. The reduction of segregation was indeed accompanied by well-being improvements, especially in the 1960s and 1970s. Regarding the role that education has played, this study highlights that it was only from 1990 onward that African American women with either some college or university degrees had lower segregation (as compared with their peers) than those with lower education. Nevertheless, the well-being loss that African American women with university degrees derived in 2010 for being segregated from their peers in education was not too different from that of African American women with lower education.

In recent decades, knowledge spillovers have taken the helm as the driving force of growth in cities. The ease of communicating ideas and the sheer density of large urban areas have made this a plausible explanation for continued growth of employment and population in cities. However, there is little consensus on the nature of the optimal conditions for stimulating knowledge spillovers. This paper identifies these optimal conditions by exploring the relative importance of industry specialization, diversity and competition across all industries and all metropolitan areas from 1970 to 2011 in the U.S. Long-term employment growth in cities is found to be driven by industry diversity combined with a high level of competition. This combination fosters the greatest amount of cross-industry fertilization of ideas and knowledge spillovers.

There has been significant interest in the development of anonymization schemes for publishing graph data. However, privacy is a major concern in dealing with graph data. In this paper, an integrated framework for ensuring privacy in the presence of an authorization mechanism is proposed. Access control mechanisms provide additional safeguard against data breaches and ensure that only authorized information is available to end-users based on their assigned roles. The integrated framework highlights a tradeoff between privacy and authorized privileges. To attain a pre-specified privacy level, access privileges might need to be relaxed. For the proposed framework, we formulate the k-anonymous Bi-objective Graph Partitioning (k-BGP) problem and provide its hardness results. Heuristics solutions are developed to solve the constraint problem. The framework provides an anonymous view based on the target class of role-based workloads for graph data. The proposed heuristics are empirically evaluated and a detailed security analysis of the framework in terms of risk associated with re-identification attack is conducted.

This dissertation is focused on private health insurance coverage, health reform and labor market outcomes. Using novel and rigorous empirical strategies, the first two essays estimate the impact of health insurance tax credits adopted during Massachusetts’s 2006 health reform and as a part of the Affordable Care Act (ACA) in 2014 on non-group private health insurance coverage. In Massachusetts, I find a large response on the margin for the tax credits. For the ACA, I document robust, positive effects on private coverage at the lowest eligibility threshold and weak evidence of effects at higher thresholds. Separating these effects from other important ACA policies, such as Medicaid expansion or the individual mandate, is vital to future efforts to modify and sustain the progress made by the ACA. The third essay addresses a significant gap in the literature, examining how employersponsored health insurance (ESI) affects the earnings distribution. I examine the role of sample selection and selection bias as an explanation for the inconsistent findings in the literature. Using quantile regression, I show that that cost-shifting due to compensating wage differentials occurs and that cost-shifting can be offset for higher earnings due to higher marginal tax rates, producing net-positive effects. Together, my dissertation indicates that reducing reliance on ESI may have beneficial effects on earnings for low- and middle-income individuals and that health insurance tax credits provide an appealing, alternative coverage option

This case examines the complex issues in caring for a patient in the U.S. illegally who was involved in a motor vehicle accident. He was diagnosed as being in a persistent vegetative state, and proxy decision-making was complicated by his illegal immigration and marital status, and by his estranged relationship with his mother in Mexico.

Competing explanations for the longstanding gap between black and white earnings attribute different weight to wage discrimination and human capital differences. Using new data on local school quality, we find that human capital played a predominant role in determining 1940 wage and occupational status gaps in the South despite entrenched racial discrimination in civic life and the lack of federal employment protections. The resulting wage gap coincides with the higher end of the range of estimates from the post-Civil Rights era. We estimate that truly separate but equal'' schools would have reduced wage inequality by 40 - 51 percent.

While previous work has analyzed skill sorting across cities, our interest lies in how less educated households who remain in cities seeing growing demand for college graduates are faring, focusing in particular on their housing and neighborhood outcomes. A growing literature shows that pecuniary and non-pecuniary moving frictions mean that a substantial number of inframarginal (stayer) households may bear the incidence of these demand shocks (Bartik 2016, Zabek 2017). We focus specifically on less educated renters, as they are more vulnerable to rising housing costs. In future drafts we will explore outcomes for all less educated individuals. We define less educated individuals as those completing a high school degree or less and college graduates those completing a bachelors degree or more.

One potential consequence of suburbanization is weaker social connectedness. Based on data from the 2003 to 2013 American Time Use Surveys, this research uses difference of means t-tests, propensity score matching, and Tobit regression to assess whether suburban living is associated with less socializing than city living in mid-to-large American metropolitan areas. After controlling for personal characteristics, we find no meaningful difference in suburbanites and city dwellers time spent socializing across a wide range of social activities. Further, suburbanites and city dwellers spend a very similar amount of time traveling, and more time spent traveling is associated with more socializing, not less.

I exploit historical natural experiments to study how establishing a new college affects local invention. Throughout the nineteenth to the mid twentieth century, many new colleges were established in the U.S. I use data on the site selection decisions for a subset of these colleges to identify “losing finalist” locations that were strongly considered to become the site of a new college but were ultimately not chosen for plausibly exogenous reasons. The losing finalists are very similar to the winning college counties along observable dimensions. Using the losing finalists as counterfactuals, I find that the establishment of a new college caused 33% more patents per year in college counties relative to the losing finalists. To determine the channels by which colleges increase patenting, I use a novel dataset of college yearbooks and individual level census data to learn who the additional patents in college counties come from. A college’s alumni account for about 10% of the additional patents, while faculty account for less than 1%. Knowledge spillovers to individuals unaffiliated with the college or living in the college county prior to the establishment of the new college also account for less than 1% of the additional patents. Migration is the primary channel by which colleges affect local invention, as controlling for county population accounts for 40-65% of the increase in patenting in college counties relative to the losing finalists. In spite of this, the presence of geographic spillovers suggests that colleges cause an overall net increase in patenting, although I find no evidence that colleges are better at promoting invention than other policies that lead to similar levels of urbanization.

The purpose of this study was to examine the relationship and predictability of selected financial, institutional, and demographic factors on the graduation rate of senior college students at selected Historically Black Colleges and Universities (HBCUs) in the South. Specifically, this study was concerned with the predictive power of the variables financial aid, student loans, scholarships, housing, attendance status, generational statistics, gender, ethnicity and age on the graduation rate of (HBCUs) college students. The population for this study consisted of senior college students who were graduating at the target institutions of higher learning during the 2015-2016 academic school years. There were three institutions of higher learning selected to participate in the study. Ten percent of the graduating senior students were selected from each institution. This empirical study employed two parametric statistical techniques to test three hypotheses. They were the multiple correlations and standard multiple regression procedures. Analysis of data revealed that all three hypotheses were found to be significant at the .001 level of the three demographic factors (gender, age, and ethnicity), ethnicity was found to be negatively related to the graduation rate of senior college students. Furthermore, the variable “receiving a loan” was found to be the only financial factor correlated with the graduation rate of senior college students. Receiving a loan was found to be positively related to graduation rate. Finally, two of the three institutional factors were found to be related to the graduation rate of senior college students. The variables “teacher-student ratio” and “generational status” were found to be positively related to graduation rate. This study concludes that demographic factors such as gender, age and ethnicity should be included in a regression model employed to predict the graduation rate among senior college students. Any attempt to predict the graduation rate of senior HBCUs college students should take into account financial factors such as receiving financial aid, receiving a student loan, and receiving a scholarship. In addition, student-faculty ratio was the strongest predictor of the graduation rate among senior HBCU college students. Furthermore, the institutional factors of housing and generational status were good predictors of the graduation rate of senior HBCU college students.

Employer-to-employer (E-E) transitions have declined in the United States during the last 20 years from a monthly rate of 2.7 percent in 1996 to 1.7 percent in 2016. I study the factors behind this decline. I document that most of the decrease in E-E transitions is accounted for by declines in matches with less than 12 months of job tenure. I attribute this decline to an increase in information about the quality of job opportunities. I then develop a search model with heterogenous matches and on-the-job search with learning about match quality. I show that the information channel can be identified from the change in the wage growth of job switchers. I estimate my model and find that workers in recent years have substantially more information about matches before they are formed, turning jobs into inspection goods rather than experience goods. I find that this increase in information explains 50 to 60 percent of the decline in job mobility over the last two decades. ⇤

For various applications, the relations between the dependent and independent variables are highly nonlinear. Consequently, for large scale complex problems, neural networks and regression trees are commonly preferred over linear models such as Lasso. This work proposes learning the feature nonlinearities by binning feature values and finding the best fit in each quantile using non-convex regularized linear regression. The algorithm first captures the dependence between neighboring quantiles by enforcing smoothness via piecewise-constant/linear approximation and then selects a sparse subset of good features. We prove that the proposed algorithm is statistically and computationally efficient. In particular, it achieves linear rate of convergence while requiring near-minimal number of samples. Evaluations on synthetic and real datasets demonstrate that algorithm is competitive with current state-of-the-art and accurately learns feature nonlinearities. Finally, we explore an interesting connection between the binning stage of our algorithm and sparse Johnson-Lindenstrauss matrices.

This study investigates how veteran status influences earnings for working-age American women. Recent increases in women’s participation in the U.S. military mean that the proportion of female veterans is rising and is forecast to increase over the next 30 years. Yet we still know relatively little about the relationship between women’s military experience and later labor-market outcomes. Drawing on American Community Survey data from 2008 to 2010 and employing a new set of occupational categories better suited to veterans, we investigate how occupation and race/ethnicity influence the effect of veteran status on women’s earnings. Findings corroborate previous support for the “bridging hypothesis” in two ways. First, veterans are overrepresented in higher paying occupations and underrepresented in the lowest paying ones, partially accounting for their higher earnings. Second, military experience particularly enhances the earnings of disadvantaged race/ethnic minority women.