Communities of color and poor neighborhoods are disproportionately exposed to more air pollution—a pattern known as environmental injustices. Environmental injustices increase susceptibility to negative health outcomes among residents in affected communities. The structural mechanisms distributing environmental injustices in the USA are understudied. Bridging the literatures on the social determinants of health and environmental justice highlights the importance of the environmental conditions for health inequalities and sheds light on the institutional mechanisms driving environmental health inequalities. Employing a critical quantitative methods approach, we use data from an innovative state racism index to argue that systematic racialized inequalities in areas from housing to employment increase outdoor airborne environmental health risks in neighborhoods. Results of a multilevel analysis in over 65,000 census tracts demonstrate that tracts in states with higher levels of state-level Black–white gaps report greater environmental health risk exposure to outdoor air pollution. The state racism index explains four-to-ten percent of county- and state-level variation in carcinogenic risk and noncarcinogenic respiratory system risks from outdoor air toxics. The findings suggest that the disproportional exposure across communities is tied to systematic inequalities in environmental regulation and other structural elements such as housing and incarceration. Structural racism is an environmental justice issue.

We analyze differences in mode of transportation to work by sexual orientation, using the American Community Survey 2008-2019. Individuals in same-sex couples are significantly less likely to drive to work than men and women in different-sex couples. This gap is particularly stark among men: on average, almost 12 percentage point (or 13%) lower likelihood of driving to work for men in same-sex couples. Individuals in same-sex couples are also more likely to use public transport, walk, or bike to work: on average, men and women are 7 and 3 percentage points more likely, respectively, to take public transportation to work than those in different-sex couples. These differences persist after controlling for demographic characteristics, partner's characteristics, location, fertility, and marital status. Additional evidence from the General Social Survey 2008-2018 suggests that these disparities by sexual orientation may be due to lesbian, gay, and bisexual individuals caring more for the environment than straight individuals.

The subject of our essay is the impact of educational and labor market discrimination on diff erences in income and wealth across races and ethnicities. The disparities across the racial and ethnic divides in income are striking (see table 1). Even more remarkable are the disparities in the wealth-to-income ratios (see table 2), and therefore in the disparities of wealth. According to the Federal Reserveʼs “Survey of Consumer Finances 2019” (Board of Governors of the Federal Reserve System 2019b), the top decile of households by income accounts for 46 percent of total income, but that same top decile by net worth holds 76 percent of the countryʼs total net worth. Studies of the impact of racial discrimination on households normally focus on its impact on income, probably because the eff ects of discrimination in education and in labor market opportunities aff ect income before they aff ect wealth. In our essay, however, we fi rst address the issue of the greater disparities that coexist with wealth. We fi nd that these disparities are, at least in part, the result of infl uences that are not necessarily the direct result of racial discrimination. In part 2 of the essay, we address the impact of educational and labor market discrimination on income.

We examine the initial post-college geographic location decisions of students from hometowns in the Appalachian region that often lack substantial high-skilled job opportunities, focusing on the role of non-pecuniary considerations. Novel survey questions allow us to measure the full non-pecuniary benefits of each relevant geographic location, in dollar equivalents. A new specification test is designed and implemented to provide evidence about the quality of these non-pecuniary measures. Supplementing perceived location choice probabilities and expectations about pecuniary factors with our new non-pecuniary measures allows a new approach for obtaining a comprehensive understanding of the importance of pecuniary and non-pecuniary factors for location decisions. We compare this approach to alternative expectations-based approaches. We also combine the non-pecuniary measures with realized location and earnings outcomes to characterize inequality in overall welfare.

We test whether low-income mothers are more likely to skip breakfast, lunch and/or dinner and thereby increase their risk of overweight and obesity. Low-income mothers are significantly more likely to be overweight or obese relative to low-income women not raising children and all men, but the mechanisms generating these disparities remain unclear. Using 2006–2008 and 2014–2016 American Time Use Surveys restricted to 18–55-year-old respondents, we predict whether respondents eat breakfast, lunch, and/or dinner as a meal (i.e., eat as a primary activity during specified times) or as a snack (i.e., eat as a secondary activity during the same specified times). We then predict respondents’ risk of overweight and obesity (corrected for bias in self-reports). All models examine conditional relationships between sex, presence of children in the home, and income category. While eating specific meals varies by pairwise combinations of sex, presence of children, and income category, low-income mothers are not significantly less likely to eat lunch or dinner meals, but they are significantly less likely to eat within 2 h of waking relative to all other individuals. Yet including mealtime measures does not notably attenuate the significantly higher risk of overweight/obesity for low-income mothers. Results highlight the complex interplay between gender, parenting, and income for food consumption behaviors, but disparities in overweight and obesity remain difficult to explain.

Intersectional inequality-the notion that disparities run along the lines of combinations of salient social groups such as gender or ethnicity-has become an increasingly prominent concept in the social sciences. Yet, because of the number of possible combinations of groups, embedding intersectionality into the measurement of inequality has proven to be analytically complex. Hence there is little empirical research applying an intersectional framework to measure inequality. We incorporate intersectionality into the measurement of between-group inequalities in educational attainment using DHS data from 39 low-and middle-income countries and the United States. Using schooling ratios between lower and higher educated groups as an inequality measure, we show that the intersectional perspective unveils a lot of inequality that remains masked if gender and ethnicity are considered in isolation. Generally, intersectional inequality in education is driven more by ethnic inequality than gender inequality. Further, we develop a novel metric to evaluate the relevance of intersectionality compared to standard approaches of between-group inequality measurement. The new metric, Surplus Intersectionality, reveals substantial heterogeneity between countries in terms of how much intersectionality is present. While gender, ethnic and intersecting inequality all seem to be driven mainly by a country's average education level, this is not the case for Surplus Intersectionality.

I examine the impacts of the Chinese Exclusion Act of 1882, America's first ever immigration restriction that prohibited entry to Chinese laborers, on native labor market outcomes. To identify the causal effect, I utilize variation in pre-1882 Chinese settlement and match on individual-and labor market-level characteristics. Using individual-level linked Census data from 1880-1900 of over two million US workers, I find the Chinese exclusion significantly slowed the long-term occupational mobility of native workers, with the effects strongest for low-skilled and unemployed workers. I find evidence in support of what I term a "honeypot" effect: low-skilled natives likely benefited from the labor shortage through higher wages in the short-run, but in the long-run the shortage disincentivized upskilling and slowed occupational upgrading. Moreover, I show Chinese laborers were almost entirely substituted by immigrants from other countries in the long-run, likely negating positive wage effects.

Federal policies that perpetuate inequities and drive Black, Native, Latinx, and immigrant families into poverty also have led to disproportionate involvement in child welfare for families of color. To address the existing inequities and advance anti-racist child welfare prevention strategies, we must implement solutions that support families in meeting their concrete needs. Th is paper outlines existing research and identifi es anti-racist policy solutions to prevent children and families of color from experiencing child welfare involvement.

American men experience higher premature mortality than women, while women are poorer. These gender inequalities are substantially different across racial groups. Based on these facts, I explore in this paper two complementary questions. First, what is the most disadvantaged gender group when combining poverty and mortality data? Second, are there racial disparities in the pattern of gender inequalities in total deprivation? This study uses the generated deprivation index, a novel indicator that aggregates poverty and mortality as components of total deprivation, to answer those questions. Two main conclusions emerge from the analysis. First, since the 1990s, men and women have been experiencing very similar total deprivation rates, whereas, before then, men were more deprived than women. The reduction of the gender gap in mortality combined with the lack of significant progress in the gender inequality in income poverty resulted in a steeper decline in total deprivation among men. Second , this near gender equality in total deprivation hides sizable disparities across races. The gender gap against women is higher for Hispanics and Blacks compared to Whites Non-Hispanics. This finding suggests that women in Minorities face more severe racial penalties than men.

This paper estimates the contribution of changes in major risk factors to mortality trends in the United States during the period 1997-2015. The risk factors investigated include cigarette smoking, obesity, alcohol consumption, educational attainment, health insurance coverage, and mental distress. It uses National Health Interview Surveys followed into death records to investigate the relationship between mortality and risk factors and to identify changes in the prevalence of the risk factors over the period of observation. All models control for age, sex, and race/ethnicity. It concludes that increases in educational attainment and reductions in smoking prevalence are the most important contributors to mortality change over the period of study.

We study the multidimensional sorting of males and females in the U.S. marriage market over the past decade using a model of targeted search. We find strong vertical sorting on income and education, and horizontal sorting on race. We find that women put significant effort into targeting men at the top of the desirability scale, while men put less effort and target women with similar characteristics. We find no improvement in quality of matching and no noticeable changes in sorting patterns or individual search behavior, despite rapid improvement in search technology. Finally, we find that targeted search substantially reduces income inequality across married couples, even when compared with random matching, by producing a large number of matches between low income and high income individuals.

There is evidence that smoking prevalence rates are related to acculturation, education, and gender among Asian Americans. However, no studies have examined how smoking rates among Asian Americans vary based on acculturation, education, and gender together. This study used National Health Interview Survey (NHIS) data (2010–2018) to examine cigarette smoking prevalence among Asian American men and women aged 18 and older (N = 14,680). Multivariate logistic regression models were used to estimate associations between educational attainment (i.e., college graduate or higher vs some college or lower), years spent in the United States (U.S.) as a proxy for acculturation (i.e., less than 10 years (less acculturated) vs 10 years or more (more acculturated) vs U.S.-born), and cigarette smoking prevalence across gender controlling for age, marital status, poverty (at/above vs below poverty threshold), country of origin (Chinese vs Filipino vs Asian Indian vs Other Asian), and the survey year. Current smoking prevalence was 9.0 % among all Asian Americans − 5.0 % among women and 13.5 % among men. Among respective gender-specific subgroups, U.S.-born Asian women without a college degree and more acculturated Asian immigrant men without a college degree had the highest odds of smoking (OR: 4.096 [95 % CI: 2.638, 6.360] and 1.462 [95 % CI: 1.197, 1.774], respectively). Findings indicated that less educated U.S.-born Asian women and less educated Asian immigrant men are at greatest risk for smoking. Smoking prevalence among Asian Americans is highly related to acculturation, education, and gender. Findings may inform development of policies and programs that are targeted toward smoking cessation among Asian Americans.

In the age of mass migration, the US became economically crucial with the development of Atlantic trade and attracted labour flows from the rest of the world. Meanwhile, the Ottoman Empire was suffering severe economic and political problems. The Ottoman citizens of various ethnic origins, such as Turks, Jews, Armenians and Arabs, moved to different countries, including the US, for economic and political reasons. This paper examines the effect of institutional reforms in the Ottoman Empire during the Second Constitutional Era on immigration decisions of different ethnic groups to the US. Data come from the US census data of IPUMS (the Integrated Public Use Microdata Series) for 1910. The empirical results show that the institutional reforms could reduce emigration from the Ottoman Empire through granting additional civil liberties. However, this effect works heterogeneously for different ethnic groups. This result points out the importance of institutional improvements in home countries on reducing migration outflows.

Joinpoint regression can model trends in time-specific estimates from aggregated data. These methods have been developed mainly for nonsurvey data such as cancer registry data assuming that the time-specific estimates are uncorrelated from time point to time point. This independence assumption can be violated for trends in time-specific estimates from complex survey samples due to using the same primary sampling units across time and, therefore, the full variance–covariance matrix of the time-specific estimates should be incorporated into the regression model fitting. This article extends these joinpoint methods for analyzing complex survey data within the National Cancer Institute’s Joinpoint software and empirically compares the extended method to existing methods for analyses of time trends in three surveys.

We use the Current Population Survey's Food Security Supplement data to investigate food insecurity disparities among Black and White households across different age groups and socioeconomic characteristics. We find that disparities in the probability of food insecurity between Black and White households vary considerably across specific socioeconomic strata, in particular education, poverty status, and home ownership. Black households are systematically more food insecure than White, even when conditioning on other attributes, and even once household heterogeneity is eliminated. Thus, factors beyond socioeconomic characteristics may be more important in determining food insecurity disparities across Black and White households.

The 2020 Decennial Census in the United States was released with a new disclosure avoidance system in place, putting differential privacy in the spotlight for a wide range of data users. We consider several key applications of census data in redistricting, developing tools and demonstrations for practitioners who are concerned about the impacts of this new noising algorithm called TopDown. Based on a close look at nine localities in Texas and Arizona, we find reassuring evidence that TopDown did not threaten the ability to balance districts, describe their demographic composition accurately, or detect signals of racial polarization.

We examine whether employment discrimination increased after the 2016 presidential election in the United States. We submitted fictitious applications to publicly advertised positions using resumes that are manipulated on perceived race and ethnicity (Somali American, African American, and white American). Prior to the 2016 election, employers contacted Somali American applicants slightly less than white applicants but more than African American applicants. After the election, the difference between white and Somali American applicants increased by 8 percentage points. The increased discrimination predominantly occurred in occupations involving interaction with customers. We continued data collection from July 2017 to March 2018 to test for seasonality in discrimination; there was no substantial increase in discrimination after the 2017 local election.

This paper proposes a new empirical strategy to estimate the causal e˙ects of 1930s federal “redlining” – the mapping and grading of US neighborhoods by the Home Owners’ Loan Corporation (HOLC). Our analysis exploits an exogenous city size cuto˙: only cities above 40,000 residents were mapped. We employ a di˙erence-in-di˙erences design, comparing areas that received a particular grade with neighborhoods that would have received the same grade if their city had been mapped. The control neighborhoods are defined using a machine learning algorithm trained to draw HOLC-like maps using newly geocoded full-count census records. For the year 1940, we find a substantial reduction in property values and homeownership rates in areas with the lowest grade along with an increase in the share of African American residents. We also find sizable house value reductions in the second-to-lowest grade areas. Such negative e˙ects on property values persisted until the early 1980s. Our results illustrate how institutional practices can coordinate individual discriminatory choices and amplify their consequences.

Automobiles are central to participation in economic, social, and cultural activities in the United States. The ability to drive as one ages is fundamental to the quality of life among older adults. Driving rates decline significantly with age. Researchers using cross-sectional data have studied the reasons former drivers have stopped driving, but few have followed individuals over time to examine changes in relationships among driving cessation, socio-demographics, and health conditions. We used longitudinal data from a national sample of 20,000 observations from the University of Michigan Health and Retirement Study (HRS) to examine relationships among demographic variables, health conditions, and driving reduction and driving cessation. Longitudinal data allow analysis of generational differences in behavior, a major advantage over cross-sectional data which only allow comparisons of different people at one point in time. We found, like many other studies, that personal decisions to limit and eventually stop driving vary with sex, age, and health conditions. In addition, unlike most previous studies, we also found that those relationships differ by birth cohort with younger cohorts less likely to stop and limit their driving than their older counterparts. The findings indicate an evolution in the association between driving cessation and its causes.

This dissertation studies the role of occupation-specific human capital in explaining the long-run decline in labor market dynamics observed in the United States for the past four decades. Chapter 1 presents empirical facts on labor market outcomes by required occupation-specific training. This is to provide evidence that (i) required length of occupation-specific training is a proxy for the specificity of human capital to perform the occupation and that (ii) increasing occupation specificity has led to the decline in labor market dynamics. First, I find from the Dictionary of Occupational Titles and O*NET that for the past four decades, within occupations, there has been an increase the amount of time needed to become trained in the occupation. I then find from the Survey of Income and Program Participation that the average wage loss experienced by occupation switchers after unemployment increases when their occupation held before unemployment has faced over time an increase in occupation-specific training. I take this as evidence that the observed increase in occupation-specific training over time has made human capital less transferable across occupations. I then proceed to use the Monthly Current Population Survey, combined with the required length of occupation-specific training by occupation from the Dictionary of Occupational Titles and O*NET, to do a shift-share decomposition of the decline in labor market outcomes. The decline in the aggregate job separation rate and the increase in unemployment duration is accounted for mostly by the increase in specific training within occupations. Motivated by my empirical analysis, in Chapter 2, I then build a search-and-matching model to learn how the increase in specificity within occupations explains the decline in the aggregate job separation rate. The main ingredients are endogenous job separations and occupation-specific human capital that workers acquire during employment and lose when they switch occupations. My model has two occupation specificity parameters: (i) the average duration of occupation-specific training and (ii) the output gap by which nontrained workers are less productive because they have not yet acquired the occupation-specific capital. To ask my model how much of a decline it predicts in the aggregate job separation rate when occupations become more specific, the occupation specificity parameters in the model are increased to match the increase in occupation specificity in the data. The increase in the average duration of occupation-specific training matches the required length of occupation-specific training from the Dictionary of Occupational Titles and O*NET. The increase in the output gap is informed by the estimated increase in the wage penalty faced by occupation switchers (relative to non-occupation switchers) when their previously held occupation requires more occupation-specific training, obtained from the Survey of Income and Program Participation. The model predicts 60% of the decline in the aggregate job separation rate. Chapter 3 relaxes the assumption that occupation switching is exogenous in Chapter 2, endogenizing occupation switching in addition to job separations. The model predicts a greater increase in the average unemployment duration in line with the data. In the model, the longer unemployment spells are due to the unemployed trained workers, whose human capital has become more specific to their previous occupation, choosing not to switch occupations. If they switch occupations, they could quickly end their unemployment spell. This would however come at the cost of larger wage cuts because their human capital has become less transferable to a different occupation. Occupation switchers would also have to earn these lower wages for a longer period of time until they become trained in their new occupation. Hence, despite a low probability of getting reemployed in the same occupation as before, previously trained workers increasingly choose not to switch occupations, which increases the average unemployment duration.