The purpose of this dissertation is to understand the causal relationships which link public policies to risky behaviors, housing, and health. Geographic variation in policy exposure, housing resources, or healthcare access generates socially meaningful and policy-relevant impacts on mental and physical health. The three essays which comprise this dissertation consider how state- and local-level public policies and healthcare infrastructure impact these health outcomes. Chapter One investigates the impact of nuisance ordinances (a housing-related policy) on substance-use related mortality rates in Ohio. Chapter Two explores how changes in access to psychiatric- and substance-use-related treatment affect subsequent eviction rates. Finally, Chapter Three examines the heterogeneous effect of cannabis- and opioid-related laws on licit- and illicit-opioid-related mortality rates. The goal of these findings is to generate policy recommendations and encourage debates that account for the potential unintended consequences of policies on mental health and risky behaviors.

Among US adults in 2013–18, we found high-deductible health plan enrollment to be the lowest among heterosexual and gay/lesbian adults in families with incomes below 200 percent of the federal poverty level and the highest among bisexual adults in families with incomes at or above 400 percent of poverty. Gay/lesbian and bisexual adults in these plans experienced greater financial barriers to health care than heterosexual adults.

The intersection of age with ethnicity is understudied, particularly for labor force outcomes. We explore the labor market for Hispanic high school graduates in the United States by age using information from the US Census, American Community Survey, Current Population Survey, and three laboratory experiments with different populations. We find that the differences in outcomes for Hispanic and non-Hispanic high school graduates do not change across the lifecycle. Moving to a laboratory setting, we provided participants with randomized resumes for a clerical position that are, on average, equivalent except for name and age. In all experiments, applicants with Hispanic and non-Hispanic names were treated the same across the lifecycle. These findings are in stark contrast to the differences and patterns across the lifecycle for corresponding Black workers and job applicants. We argue that these null results may explain the much smaller literature on labor market discrimination against less-educated Hispanic workers.

The spatial distribution of urban agriculture in cities is not well studied, and scholars have tended to overlook localized effects and physical access issues when contemplating urban agriculture (UA) futures. To address this gap, we ask: what is the current spatial distribution of community gardens in New York City (NYC), and what land-use policies will enable more accessible garden futures? We adopt the concept of the 15-minute city to map the future of community gardens in NYC. We analyze garden distribution in NYC using remote sensing and spatial regression and design an optimization-based spatial planning approach to evaluate the feasibility of turning NYC into a 15-minute garden city. Our results indicate that more than half of the city residents have access to a garden within 15-minute walking distance, and that neighborhoods with lower income, lower rates of white residents, lower rates of owner occupancy, and higher rates of educational attainment have higher rates of access. The most cost- effective increases in household access to gardens arise from developing new gardens on vacant parcels, which outperform modeled gardens sited on all other land uses, though a strategy of siting gardens on a range of land uses is required to maximize household access. By mapping gardens, analyzing their distribution, and modeling how to scale-up UA, this paper presents a novel spatial planning approach to expand urban amenities and ecosystem service benefits for a more just, sustainable, and resilient city. This spatial planning approach enables participatory planning processes for UA futures in a variety of urban contexts

I explore the impact of the Production Credit Associations (PCAs), an arm of the early Farm Credit System, on agricultural yield and input use following the farm crisis of the 1920s. Like many low- and middle-income countries today, farmers in the early 20th century United States found it difficult to access credit. The PCAs were established in 1933 and significantly increased the supply of short-term credit available to farmers. Using distance from the serving PCA as a proxy for credit access, I find that counties within 30 km of a PCA had 7% to 14% higher crop revenue per acre and 9% higher corn yields than counties more than 60 km from a PCA. These areas also had a small but statistically significant increase in the use of tractors (1% to 2%). These results provide crucial evidence of the impact of government-sponsored enterprises on the early US agricultural economy and its use as a cost-effective tool to address market frictions.

This study evaluated the contributions of land cover and land use change (LCLUC) and land management to landscape carbon production through a complex cause-effect path analysis of socioecological latent variables. Socioecological contributions to landscape carbon production are essential in landscape analysis, as their processes are both independent and interactive. We quantify the coherencies of social, economic, and environmental variables and their impact on net primary production (NPP) in an agroecosystem landscape. We ask whether LCLUC contributed to increased NPP and if land management and LCLUC play a more significant role than abiotic stressors on NPP. We applied a socio-environmental system framework to evaluate anthropogenic and environmental processes in the Kalamazoo River Watershed in southwest Michigan, USA from 1987 to 2017. Structural composition and functional contribution to NPP were evaluated by land cover type. We synthesized remote sensing, gridded climate, social and biophysical data in a principal component analysis (PCA) to inform a partial least squares structural equation model (PLS-SEM). Land cover type contributed to anthropogenic processes. Cropland contributed to Land Management, forest and water contributed to Land Cover Change, and urban to the Regional Development construct. Anthropogenic activities contributed more to NPP than abiotic processes. Attitudes of environmental stewardship strongly related to land use change likelihood. We disentangled anthropogenic and climatic changes’ contributions to terrestrial carbon production and the societal ties to potential carbon sequestration. No single landscape metric is suitable for all study areas; however, this framework is useful for a landscape-scale analysis of socio-environmental processes.

We examine disparities in Child Tax Credit (CTC) eligibility and anti-poverty effects since 1998 by family type. Initially, single mothers were least likely to be eligible and were underrepresented among those lifted from poverty by the CTC, because the credit was virtually nonrefundable. By 2017, disparities by family type mostly disappear, as eligibility and anti-poverty effectiveness of the CTC among single mothers increases dramatically, because of reforms increasing CTC refundability. When the credit doubles in 2018, disparities revert toward initial levels, as eligibility and the anti-poverty effectiveness of single mothers rises least, because of a phaseout threshold expansion and partial refundability.

We study inference on dynamic average treatment effect parameters for staggered interventions when parallel trends are only valid conditional on unobserved interactive fixed effects. Our identification strategy allows for any first stage system of moments that controls the column space of the unobservable trends including principal components, common correlated effects, quasi-differencing, and more. This result applies to data sets with either many or few pretreatment time periods. We also demonstrate the robustness of two-way fixed effects to certain parallel trends violations and describe how to test for its consistency. We investigate the effect of Walmart openings on local economic conditions and demonstrate that our methods ameliorate pre-trend violations commonly found in the literature.

With high and growing rates of maternal mortality and most maternal deaths occurring following delivery, increasing attention is being paid to health insurance coverage and health care during the postpartum period (Hoyert 2023; Petersen et al. 2019; Haley et al. 2021a; Johnston et al. 2021; Rodin et al. 2019; Burroughs et al. 2021; Haley et al. 2022). Medicaid may have a particularly important role, given that in 2021, Medicaid financed more than 4 in 10 births in the United States.1 Medicaid eligibility is more generous for pregnant people than for other nonelderly adults, but this coverage historically expired just 60 days after the end of pregnancy (Haley et al. 2021a). However, Congress enacted the Families First Coronavirus Response Act (FFCRA) in March 2020, which established a continuous coverage requirement mandating states not disenroll people from Medicaid, including those with pregnancy-related coverage, during the public health emergency (PHE).2 The policy resulted in historic growth in Medicaid enrollment, reaching 93.9 million by March 2023 with the pregnancy-related coverage pathway experiencing the highest enrollment growth.

Background and Objectives Older adult multimorbidity trajectories are helpful for understanding the current and future health patterns of aging populations. The construction of multimorbidity trajectories from comorbidity index scores will help inform public health and clinical interventions targeting those individuals that are on unhealthy trajectories. Investigators have used many different techniques when creating multimorbidity trajectories in prior literature, and no standard way has emerged. This study compares and contrasts multimorbidity trajectories constructed from various methods. Research Design and Methods We describe the difference between aging trajectories constructed with the Charlson Comorbidity Index (CCI) and Elixhauser Comorbidity Index (ECI). We also explore the differences between acute (single-year) and chronic (cumulative) derivations of CCI and ECI scores. Social determinants of health can affect disease burden over time; thus, our models include income, race/ethnicity, and sex differences. Results We use group-based trajectory modeling (GBTM) to estimate multimorbidity trajectories for 86,909 individuals aged 66 to 75 in 1992 using Medicare claims data collected over the following 21 years. We identify low-chronic disease and high-chronic disease trajectories in all eight generated trajectory models. Additionally, all eight models satisfied prior established statistical diagnostic criteria for well-performing GBTM models. Discussion and Implications Clinicians may use these trajectories to identify patients on an unhealthy path and prompt a possible intervention that may shift the patient to a healthier trajectory.

To understand rising income inequality in the United States, we use data from the March Current Population Survey (CPS) to decompose income inequality into components attributable to five personal traits: sex, race, education, occupation, and industry of work. By quantifying how income differences across these traits contribute to total inequality, and how those contributions have changed over time, we vet competing hypotheses for the rising gap. In performing this analysis, we correct for data censorship (“Top-coding”) within the CPS by fitting the upper tail of the income distribution and imputing the hidden observations; this represents an extension to previous studies that instead truncate the top several percentiles of income data. Our findings suggest that changes in the returns to education played an important role in driving the observed rise in inequality.

We examine how migrant workers impact firm performance using administrative data from the United States. Exploiting an unexpected change in firms’ likelihood of securing low-wage workers through the H-2B visa program, we find limited crowd-out of other forms of employment and no impact on average pay at the firm. Yet, access to H-2B workers raises firms’ annual revenues and survival likelihood. Our results are consistent with the notion that guest worker programs can help address labor shortages without inflicting large losses on incumbent workers.

My research examines the labor market performance of prime-age (20-54) male immigrant veterans in the all-volunteer era, using the American Community Survey (2010-2019) and a linear regression model. I compare them with two reference groups: U.S.-born male veterans and non-veteran male immigrants. Immigrant veterans show higher labor force participation and employment rates compared to U.S.-born veterans. Conversely, when compared to non-veteran immigrants, the trend is reversed. Overall, no statistically significant disparities are found in labor income and wage analysis. However, heterogeneous effects emerge based on education level, race, and birthplace. Immigrant veterans earn less income compared to U.S.-born veterans in the non-Hispanic White group, while no significant differences are observed in other races. Furthermore, compared to non-veteran immigrants, Hispanic/Latino and non-Hispanic Black immigrant veterans have higher labor incomes or wages, especially for those with a high school education as their highest level of attainment. In contrast, non-Hispanic White immigrant veterans earn less, particularly among individuals with a college education. Additionally, immigrant veterans from developed countries have lower incomes and wages compared to their non-veteran counterparts, whereas no such disadvantage is observed for immigrants from developing countries. These findings suggest potential advantages associated with military service in minority groups.

I formulated a fertility model to quantify preference for a specific gender and gender variety, providing a framework to explore the counterfactual impact of the preva- lence of sex-selection technology on sex ratios and fertility levels. To identify the parameters, I utilized the revealed preferences for having additional children, contingent on random vari- ation in the sex composition of the pre-existing offspring. Leveraging individual data from the American Community Survey (ACS)1 in the past decade, I find a strong preference for mixed-gender offspring and a slight preference for boys in the United States. Heterogeneity among different subgroups, such as race and parental education, is significant. I next signify that the prevalence of sex selection technology is unlikely to result in a skewed sex ratio, primarily because the strong preference for gender variety dominates the preference for a specific gender. I predict the average number of children in each household will increase by approximately 1% with the proliferation of such technology. However, the net effect may vary slightly depending on the cost of sex selection and the distribution of its application across different birth parities.

Objective: This study estimates associations of regional change in opioid prescribing with total suicide deaths and suicide overdose deaths involving opioids. Methods: A panel analysis was performed with 2009-2017 U.S. national IQVIA Longitudinal Prescription Database data and National Center for Health Statistics mortality data aggregated into commuting zones (N=886), which together span the United States. Opioid prescription exposures included opioid prescriptions per capita and percentages of patients with any opioid prescription, with high-dose prescriptions (>120 mg of morphine equivalents), with long-term prescriptions (≥60 consecutive days), and with prescriptions from three or more prescribers. Linear regression models were used with year and commuting zone fixed effects. Results: Suicide deaths were significantly positively associated with opioid prescriptions per capita (β=0.045), having any opioid prescription (β=0.069), having high-dose prescriptions (β=0.024), having long-term prescriptions (β=0.028), and having three or more opioid prescribers (β=0.046). Similar significant associations were observed between each of the five opioid prescription measures and suicide overdose deaths involving opioids (β range, 0.029-0.042). However, opioid prescriptions per capita, having any opioid prescription, and having three or more opioid prescribers were each negatively associated with unintentional opioid-related deaths in people in the 10- to 24-year and 25- to 44-year age groups. Conclusions: In this retrospective study of U.S. commuting zone-level opioid prescriptions and mortality, regional decreases in opioid prescriptions were consistently associated with declines in total suicide deaths, including suicide overdose deaths involving opioids. For some opioid prescribing measures, negative associations were observed with unintentional overdose deaths involving opioids among younger people. Individual-level inferences are limited by the ecological nature of the analysis.

This study builds on past research to examine whether the Medicaid expansion provision of the Affordable Care Act (ACA) affected cohabitation rates among adults ages 18–40. This provision of the ACA was designed to expand health insurance coverage for low-income individuals. The expansion has been demonstrated to have effects beyond increased access to health services, including effects on financial health, marriage, fertility, and child support payments. Nationally representative data from the 2007 to 2019 Current Population Surveys (N = 45,129–695,629) were used to estimate difference-in-difference models of the effect of the Medicaid expansion on cohabitation. Results indicate generally positive effects of the Medicaid expansion on cohabitation, but effects are primarily limited to states expanding in 2014 and are seen 2 years after expansion. Subgroup analyses indicate some heterogeneity by parenthood and race/ethnicity; results are robust to the exclusion of early and late expansion states and to a wider age range. Findings are consistent with behavior around eligibility notches and changes in non-economic factors post-expansion that may influence cohabitation decisions. As states continue to expand Medicaid coverage, such trends may be increasingly evident.

Municipal websites serve as central platforms for local governments to share information with the public. They offer authoritative, up-to-date, and free access for researchers to collect city-level data. However, until now, a comprehensive and accurate database of municipal web addresses did not exist. Here, we introduce a complete and manually verified dataset containing information on whether a municipality has an official website and, if so, what its web address is, of all 19,518 municipalities in the United States. With this dataset, researchers can easily conduct systematic searches on municipal official websites for self-defined keywords. The search results are well-suited for text-based analytics. This new data source benefits urban scholars who struggle to access high-quality local data for nationwide studies and contributes to narrowing the data gap.

Margaret Sanger established the first birth control clinic in New York in 1916. From the mid-1920s, “Sanger clinics” spread over the entire U.S. Combining newly digitized data on the roll-out of these clinics, full-count Census data, and administrative vital statistics, we find that birth control clinics accounted for 5.0–7.8% of the overall fertility decline until 1940. Moreover, birth control clinics had a significant and meaningful negative effect on the incidence of stillbirths and infant mortality. The effect of birth control clinics on puerperal deaths is consistently negative, yet insignificant. Further suggestive evidence points towards positive effects on female employment.

Prescription drug use has reached historic highs in the United States—a trend linked to increases in medicalization, institutional factors relating to the health care and pharmaceutical industries, and population aging and growing burdens of chronic disease. Despite the high and rising prevalence of use, no estimates exist of the total number of years Americans can expect to spend taking prescription drugs over their lifetimes. This study provides the first estimates of life course patterns of prescription drug use using data from the 1996–2019 Medical Expenditure Panel Surveys, the Human Mortality Database, and the National Center for Health Statistics. Newborns in 2019 could be expected to take prescription drugs for roughly half their lives: 47.54 years for women and 36.84 years for men. The number of years individuals can expect to take five or more drugs increased substantially. Americans also experienced particularly dramatic increases in years spent taking statins, antihypertensives, and antidepressants. There are also important differences in prescription drug use by race and ethnicity: non-Hispanic Whites take the most, Hispanics take the least, and non-Hispanic Blacks fall in between these extremes. Americans are taking drugs over a wide and expanding swathe of the life course, a testament to the centrality of prescription drugs in Americans' lives today.

In 2020, our everyday routines shifted dramatically, but over the past few years, it’s felt like things are getting back to where they were. Are we back? Or did our baselines move so low that any tick upwards seems like a lot? Based on data from the American Time Use Survey, we can see how our time spent has changed over the years. The charts below show the range of time for a hundred of the most common activities.