This thesis contains three chapters in empirical economics. In Chapter 1, I use administra- tive data from British Columbia, Canada, to study the short-term effects of a parental layoff on children’s academic performance in grades 4, 7, and 10. I find that households where a parent suffers a layoff, earn approximately $8,000 - $10,000 less in after- tax income in the year after the layoff. In spite of such a large loss in financial resources, I find no significant short-term effects on childrens test scores due to parental job loss. My estimates for grade 4 and grade 10 rule out negative treatment effects larger than 3.5% of a standard deviation at the 95% confidence level, and the estimates for grade 7 rule out negative treatment effects larger than 5.3% of a standard deviation. In Chapter 2, I exploit close city council elections in California from 1996 to 2017 to imple- ment a regression discontinuity design and study the causal effects of a nonwhite candi- date’s victory against a white candidate. I find that in cities where the nonwhite candidate won (treatment), compared to cities where the nonwhite candidate lost (control), more new white candidates run in the next election. Heterogeneity analysis shows that this effect is driven by cities that have gone through bigger demographic changes over the past few decades, which suggests that changes in the racial composition of the city and the associ- ated perception of threat to the dominant status of whites within the city are an important factor in driving the main result. Chapter 3 is based on my joint work with Louis Bélisle and Vivek Nandur. We compare two prominent methods used to estimate production functions in the literature. We estimate total factor productivity for firms in four manufacturing industries in India for the time period 2005-2012 using both estimation methods and using two different intermediate inputs as proxies. We find large discrepancies in the distribution of estimated total factor productivity for firms depending on the estimation method and the proxies used.

Objectives: Attention-Deficit Hyperactivity Disorder (ADHD) is one of the most common disorders diagnosed among children in the US. However, little knowledge is available about ADHD prevalence among children of immigrants, the fastest-growing population in the US. This study seeks to examine ADHD rates among children of immigrants in different generations compared to children of US-born parents and their association with family poverty. Design: The sample includes 83,362 children aged 0–17 from the National Health Interview Survey (NHIS) 2010–2018 data. Multivariate logistic regression model is used to estimate prevalence of ADHD among children of immigrants compared to children of US-born parents. We then compare ADHD rates among the children sample in different immigrant generations. For all analyses, we examine ADHD occurrence separately for children in families living below the poverty threshold and those at or above the poverty threshold. Results: The odds of having ADHD were significantly lower among children of immigrants compared to children of US-born parents. Both first-generation children and second-generation children of immigrants had significantly lower odds of having ADHD than children of US-born parents. Post hoc tests find that first-generation children had lower odds of having ADHD compared to second-generation children. Likewise, additional analyses showed that children of immigrants, first-generation children in particular, were less likely to have ADHD compared to children of US-born parents, in both lower- and higher-income families. Conclusion: Using a nationally representative sample of children, we find that the likelihood of having ADHD increases with higher generations, detecting differences in ADHD prevalence by immigration generation. Importantly, first-generation children had a significantly lower risk of having ADHD conditions compared to second-generation children and children of US-born parents, regardless of family socio-economic status. Public health policy and program development would gain from a clear comprehension of the shielding attributes of ADHD among immigrant families.

Background: Research shows general increases in child maltreatment reports in the U.S. However, less is understood about how reporting varies across states and changes over time, from a perspective of referral sources. While recent studies during COVID-19 reported a reduction in maltreatment referrals, predominantly school referrals, little research has examined changes in maltreatment referrals by referral sources before the pandemic and how different referral sources are associated with case outcomes, particularly out-of-home placement. Objectives: This study examined 1) variations across states and changes over time in maltreatment reporting by referral source and 2) the relationship between referral sources type and two case outcomes: substantiated maltreatment and out-of-home placement. Participants and setting: We used 2008–2018 data (N = 24,349,293) from the National Child Abuse and Neglect Data System. Methods: We used descriptive trend analysis and pooled, fixed effects binary logistic regression. Results: We found gradual increases in reporting during 2008–2018, with substantial variations across states and referral sources. States rely differently on certain reporter types, while we see the largest increase in education referrals and a small decrease in social services referrals. Regression results showed that education referrals were less likely to result in out-of-home placement; law enforcement referrals were most likely to be substantiated, while social service referrals were most likely to result in out-of-home placement. Conclusion: This study makes unique contributions to literature by expanding our knowledge of referral sources and examining the likelihood of substantiation and out-of-home placement by referral source type. We provide child welfare policy and practice implications.

A key facet in the development of the American middle class in the 20th century was the Servicemen Readjustment Act of 1944 (GI Bill), which provided numerous benefits to World War II Veterans, such as lower mortgages, loans, and most importantly, funding to attend universities or vocational school. The ability to access affordable higher education and vocational training is what provided most Americans the ability to move to the middle class and build generational wealth. Unfortunately, while the GI Bill was race-neutral in its language, the implementation of the GI Bill did not challenge existing institutions of discrimination and segregation, so Black Americans were mostly unable to take advantage of the benefits of the GI Bill, despite their accomplishments as American soldiers. This lack of access to build generational wealth has left a gap in racial wealth to this day. This research analyzed how the Total Personal Incomes of male Black and White World War II veterans changed after the passage of the GI Bill in 1944 and the differences between Black and White veteran incomes by using the Census Data from the IPUMS USA database to analyze the Total Personal Income of male Black and White World War II veterans from 1950 to 1990 using R. This research finds that White veterans have higher incomes than their Black counterparts at a statistically significant level when looking at an OLS regression of Total Personal Income on Year and Race. Additionally, the research studies the income differences between the 25th percentile, median, mean, and 75th percentiles of incomes of Black and White veterans, where Black veterans earned less than their White counterparts across all of these measures.

The literature on food insecurity in America is extensive, but there is still a lack of understanding of urban/ rural differences. Food insecure populations in urban communities undoubtedly experience different and unique challenges to food access than food insecure populations in rural areas. While urban households may face a lack of options for fresh produce, rural households may find that they live an inaccessible distance from the nearest grocer. Through analysis of the Census Bureau's Current Population Survey and Food Security Supplement, this project seeks to discover common, underlying variables in food-insecure communities in metropolitan and nonmetropolitan areas in America's East North Central Region (Wisconsin, Michigan, Illinois, Indiana, and Ohio) to better understand the policy environment surrounding food security in America.

The labor force participation rates (LFPRs) of the Fourth Federal Reserve District’s four states—Ohio, Pennsylvania, Kentucky, and West Virginia—are lower than the nation’s. In this Data Brief I assess how population characteristics affect these states’ LFPRs. Relative to the nation, Fourth District states’ older populations and higher disability rates are key reasons for lower participation rates. The LFPR is an estimate of what fraction of people are active in an economy’s workforce.1 The higher the rate, the more of a state’s population that is either working or available for work. This can lead to higher per capita income and output, as well as greater availability of workers. Based on 2021 data, the most recent available, Ohio’s LFPR was 63.8 percent and Pennsylvania’s was 63.7 percent, while Kentucky came in at 60.1 percent and West Virginia at 53.6 percent. The LFPR for the nation as a whole was 63.9 percent.2 For purposes of comparison, the District of Columbia ranked first, with a LFPR of 71.4 percent, while Ohio ranked 29th, Pennsylvania 30th, Kentucky 44th, and West Virginia last.

We examine transitions into and out of car ownership among low-income households. We use a novel online survey of U.S. residents to investigate why households lose access to a car, how long they are without a car, why they regain a car, and how these transitions affect their quality of life. We find that car ownership transitions are primarily motivated by economic security and insecurity. The median length of a car-less episode is 1.7 years, while black and Hispanic respondents experienced longer car-less episodes. Despite their precarious grasp on car ownership, respondents felt that owning a car was worth it.

Objective: This study examines the re-employment prospects and short-term career consequences for mothers and fathers who lost their jobs during the COVID-19 pandemic. Background: The pandemic recession has been dubbed a "shecession," but few studies have explored whether mothers paid a higher or lower price upon labor market re-entry than fathers. Method: This study draws on March 2020-December 2022 Current Population Survey data and focuses on partnered parents with children under age 13 in the household. Exploiting four-month panels, we use multi-level discrete-time event history models to predict re-employment and linear regression models to predict job-level wage upon re-employment, while controlling for a wide array of factors. Results: Partnered fathers were more likely than partnered mothers to find re-employment during the pandemic. The gender gap in re-employment was concentrated only among parents without a bachelor's degree and persisted when all controls were held constant. Moreover, upon re-employment, fathers had higher job-level wages than mothers, which was consistent across educational levels. Even with the same job-level wage before labor market exit, mothers were penalized on re-entry relative to fathers and this penalty was rooted in gendered job segregation. Conclusion: This study extends previous research by analyzing re-employment and a critical material outcome for parents (i.e., job-level wage upon re-employment) during the entire pandemic, including the "new normal" (late 2022). The results reveal the intersectional inequalities in family and work: Compared to fathers, mothers, particularly less-educated mothers, paid a higher price for their time out of work during the pandemic.

Ambient fine particulate matter (PM2.5) causes health problems and is monitored by U.S. federal and state governments. Due to gaps in public monitoring, people rely on non-regulatory air quality monitoring networks (NAQMNs), comprised of low-cost PM2.5 sensors, to obtain air quality information. While environmental justice studies have examined social disparities in PM2.5 and sensor distribution, none have examined disparities in university-purchased vs. privately-purchased sensor distributions. This is important because who purchases sensors (and why) may impact sensor distribution equity. Using multivariable generalized estimating equations, we investigate university-purchased (n=68) and privately-purchased (n=230) sensors in the Salt Lake County, Utah NAQMN with a focus on socioeconomic status (SES), race/ethnicity/immigrant status, and PM2.5 at the census tract-level (n=208). Tracts with higher SES and PM2.5 concentrations had increased sensors per resident. Findings related to SES attenuate when predicting university-purchased vs. privately-purchased sensors. Tracts with greater Hispanic immigrant populations and higher PM2.5 concentrations had increased access to university-purchased sensors. Tracts with higher Asian concentrations were underserved regardless of sensor purchaser. This approach illuminates how university-purchased sensors may enhance distributional environmental justice. Findings have implications for enhancing equitable sensor distribution in NAQMNs.

While immigrants make up an increasing proportion of the U.S. labor force, there is little consensus on how ethnic concentration affects their labor market outcomes, and even less about the effect of the interaction between ethnic concentration and education. Higher levels of education increase both earnings and the likelihood of employment, but according to the literature, the effects of ethnic concentration seem to vary by the country of origin. Further difficulty is presented by the different definitions of ethnic concentration and ethnic enclaves used in the literature. I use individual-level data from 2016 for working-age immigrants throughout the entire U.S. to investigate the relationship between the interaction of education and ethnic concentration on employment and income of immigrant workers, which serves as an update of existing literature. Additionally, this paper is the first to focus on the interaction of education and ethnic concentration, rather than these two variables individually. Like previous literature, this study finds unambiguous positive effects of education on labor market outcomes, and positive effects of the interaction between education and living in an ethnic enclave: having a college degree and living in an ethnic enclave will increase an immigrant’s likelihood of being employed by anywhere between 22% and 26%. No statistically significant effect of the interaction on earnings was found.

In 2014, California lifted its legal work status requirement for dozens of occupational licenses-a major obstacle for undocumented immigrants in the US to access professional jobs. This paper assesses this policy's effects on the employment outcome of undocumented immigrants in the state. Analyzing likely undocumented immigrants in the American Community Survey, I find that the policy increased their employment, particularly in lower-education and blue-collar licensed occupations and for older and Hispanic workers. The effects were not driven by job switchers but by the unemployed transiting into employment. I also find that the law did not crowd out documented or US-born workers.

This dissertation examines the impact of state educational programs on the economics of education and labor, focusing on the New York State (NYS) STEM Incentive Program and China's tuition-free policy for vocational secondary education (VSE). The first two Chapters analyze the effects of the NYS STEM Incentive Program, an initiative implemented in 2014 to combat the state's shortage of skilled STEM labor. It provides tuition awards to top-performing students pursuing a STEM degree on the condition that they work in the state for five years after graduation. Utilizing extensive administrative data from a large R1 public university in New York State and the American Community Survey data, these studies employ an event study research design and a synthetic control method, respectively. The findings reveal a significant increase in the probability of students selecting STEM majors and the likelihood of these students subsequently choosing STEM occupations within New York State. However, this program has a more substantial effect on men, potentially exacerbating the gender gap in STEM fields. The third study investigates the impact of China's tuition-free policy on VSE, a move intended to encourage educational equity and the appeal of VSE. This policy's effects on academic, labor market, and social outcomes are evaluated using China Family Panel Studies data. The results demonstrate that tuition exemption has increased enrollment in vocational secondary schools and fostered greater labor market participation, particularly among women and people from rural areas. In addition, it has contributed to a rise in personal income, thereby decreasing the rural-urban wage gaps to some extent. Collectively, these studies shed light on the crucial role of state educational incentives in shaping educational and labor market outcomes, offering essential insights for policymakers considering similar initiatives.

A variety of tasks have been proposed recently to facilitate exploration and analysis of charts such as chart QA and summarization. The dominant paradigm to solve these tasks has been to fine-tune a pretrained model on the task data. However, this approach is not only expensive but also not generalizable to unseen tasks. On the other hand, large language models (LLMs) have shown impressive generalization capabilities to unseen tasks with zero-or few-shot prompting. However, their application to chart-related tasks is not trivial as these tasks typically involve considering not only the underlying data but also the visual features in the chart image. We propose PROMPTCHART, a multimodal few-shot prompting framework with LLMs for chart-related applications. By analyzing the tasks carefully, we have come up with a set of prompting guidelines for each task to elicit the best few-shot performance from LLMs. We further propose a strategy to inject visual information into the prompts. Our experiments on three different chart-related information consumption tasks show that with properly designed prompts LLMs can excel on the benchmarks, achieving state-of-the-art.

Prior family profiles on family structure have focused on children's relationship to their parents, their parents' union type (Eickmeyer 2017a, 2017b, 2017c; Westrick-Payne and Wiborg 2021) and sex composition (Manning and Payne 2021). However, less is known about minor children co-residing specifically with a biological/adoptive mother or father. This family profile uses the 2022 ASEC Current Population Survey from IPUMS-CPS to fill this gap. Our analyses explore minor children's experiences living with a biological/adoptive mother (hereafter mother) and living with a biological/adoptive father (hereafter father) separately. We note that many children reside with both their mother and father. We first present the proportions of children living with their mother and with their father and then generate these estimates by children's age groups. Then, we compare the distribution of children living with their mothers or living with their fathers according to family structure (married two biological/adopted parent, cohabiting two biological/adopted parent, married stepparent, cohabiting stepparent, single parent). • In 2022, 92% of all minor children were living with their mother compared with 70.3% who were living with their father. In other words, more than nine children out of ten lived with their mother, in contrast to seven children out of ten who lived with their father. Figure 1. Proportion of children living with a biological/adoptive parent by parents' gender, 2022 Source: NCFMR analyses of ASEC Current Population Survey, 2022 from IPUMS CPS, University of Minnesota, www.ipums.org Children's Age Figure 2. Proportion of children living with a biological/adoptive parent by parent's gender and age groups, 2022 Source: NCFMR analyses of ASEC Current Population Survey, 2022 from IPUMS CPS, University of Minnesota, www.ipums.org • In general, the proportions of children living with their mother or with their father decreased with children's age. Among preschool age children 91.7% and among teenagers 86.0% lived with their mother. Three-quarters (76.3%) of preschool age children lived with their father and 65.1% of teenagers lived with their father. • The difference in the share of children living with a parent declined more precipitously across childhood for those living with their father (11.2%) than their mother (5.7%). 92% 70% 0% 25% 50% 75% 100% Biological/adoptive mother Biological/adoptive father 92% 76% 89% 70% 86% 65% 0% 25% 50% 75% 100% Biological/adoptive mother Biological/adoptive father Under 6 years 6-11 years 12-17 years

The measurement of the disparate impact of a particular de facto discriminatory policy on a minority or otherwise legally protected group has been of importance since passage of the Civil Rights Act of 1964. When the data available for the measurement of disparate impact, as embodied in the so-called “disparity ratio,” come from samples, a statistical approach naturally suggests itself. This article reviews both the law and statistics literature with regard to statistical inference applicable to the disparity ratio and related measures of disparate impact. From that review, three primary approaches are evaluated, the difference in so-called “rejection” rates for the protected and non-protected groups, their ratio (the disparity ratio), and the natural logarithm of the disparity ratio. For various reasons, the direct ratio estimator is recommended for use in all but small samples, where the log-ratio approach is to be preferred. The main points are illustrated with two fair housing examples, one being the possible discriminatory effect by race owing to a landlord’s refusal to accept Section 8 housing vouchers in lieu of cash rent, and the other being the effects of occupancy restrictions on families with children. Various methodological issues that arise in the application of these three estimation approaches are addressed in the context of the more complex sample designs that underlie the data utilized.

We study the role of denominational competition in the expansion of higher education in the nineteenth-century United States. We document that nearly all colleges established in this period were affiliated with a Christian denomination. Empirical analysis reveals a robust positive relationship between the denominational fragmentation of the county and the number of colleges established. We take several steps to rule out competing explanations and also highlight the causal channel by utilizing two historical case studies. We conclude by estimating a model of school choice and showing that students exhibited strong preferences to attend same-denominational colleges in terms of willingness-to-pay and willingness-to-travel. Therefore, we argue that religious diversity softened the extent of tuition competition between institutions and precipitated an “excess” entry of schools.

The relative incomes and education-levels of Black and white populations in the United States and Brazil are considered after Abolition, and framed by earlier disparities in their natural rates of increase. For the post-World War Two period, the effects of demography, education, and regional migration on the Black-white income gap are disentangled using census microdata and a single-equation form Blinder-Oaxaca decomposition. These variables explain progressively less of income inequality over time, meaning that discrimination or other unobserved factors have become more-substantial determinants of relative earnings. Education, measured by literacy or years of schooling, was the major reason behind reductions in income gaps during this period, followed by demography and migration. While both countries have made gains towards racial equality, their timing is entirely divergent (and sometimes counter to popular understandings): the best decade in these terms for the US was the 1960s, and the worst, the 2000s or 2010s; and, vice-versa for Brazil.

The growing affordable housing crisis in high-cost metropolitan areas may force households to seek lower cost housing in the outer reaches of metropolitan areas contributing to the recent increase in commute distance. To explore this assertion, we test the relationship between the availability of affordable housing relative to jobs and commute distance in two diverse metropolitan statistical areas in Southern California: Los AngelesOrange (higher cost, coastal, older, more urban) and RiversideSan Bernardino (lower cost, inland, newer, more suburban). A worse “fit” between the number of low-wage jobs and affordable housing rentals is associated with longer commute distances in LA-Orange, but is not statistically significant in Riverside-San Bernardino. This study’s findings highlight the differences in housing dynamics and commute distances between higher cost coastal regions and lower cost inland regions—and underscore the importance of protecting and expanding the supply of affordable housing in job-rich neighborhoods located in more expensive, coastal cities.

Introduction: In the U.S., low-income individuals generally smoke more than high-income individuals. However, detailed information about how smoking patterns differ by income, especially differences by birth cohort, is lacking. Methods: Using the National Health Interview Survey 1983–2018 data, individual family income was calculated as a ratio of the federal poverty level. Missing income data from 1983 to 1996 were imputed using sequential regression multivariate imputation. Age‒period‒cohort models with constrained natural splines were used to estimate annual probabilities of smoking initiation and cessation and smoking prevalence and intensity by gender and birth cohort (1900–2000) for 5 income groups: <100%, 100%–199%, 200%–299%, 300%–399%, and ≥400% of the federal poverty level. Analysis was conducted in 2020–2021. Results: Across all income groups, smoking prevalence and initiation probabilities are decreasing by birth cohort, whereas cessation probabilities are increasing. However, relative differences between low- and high-income groups are increasing markedly, such that there were greater declines in prevalence among those in high-income groups in more recent cohorts. Smoking initiation probabilities are lowest in the ≥400% federal poverty level group for males across birth cohorts, whereas for females, this income group has the highest initiation probabilities in older cohorts but the lowest in recent cohorts. People living below the federal poverty level have the lowest cessation probabilities across cohorts. Conclusions: Smoking prevalence has been decreasing in all income groups; however, disparities in smoking by family income are widening in recent birth cohorts. Future studies evaluating smoking disparities should account for cohort differences. Intervention strategies should focus on reducing initiation and improving quit success among low-income groups.

Routine screening and testing for HIV are recommended for all adolescents and adults. However, only one-third of the U.S. population has been tested for HIV. Women, sexual minorities, and people who use alcohol are more likely to be tested for HIV, but less is known about how alcohol use and sexual orientation impact the likelihood of HIV testing synergistically. Examining both alcohol use and sexual orientation is especially relevant, because sexual minorities are at increased risk for alcohol use, including heavy drinking. This study tested an alcohol x sexual orientation interaction effect on HIV testing through logistic regression modeling with a nationally representative sample. Results of the significant interaction identify demographic groups that are particularly at-risk for not being tested for HIV. These groups include lesbian women who currently use alcohol or previously used alcohol; bisexual men who have not used or previously used alcohol; and gay men who previously used alcohol. Although efforts to test all adolescents and adults are warranted, these findings highlight the importance of assessing alcohol and sexual orientation and augmenting testing efforts for highrisk groups.