The new category of workers officially labeled “essential” in the early stages of the COVID-19 pandemic in the United States includes a large percentage of women working in care services. In many of these services, health risks are often considered part of the job and are uncompensated by hazard pay. Building on previous feminist research explaining the devaluation of care work, this paper uses the most recent available data from the US Current Population Survey to show that workers in essential care service jobs – especially women – earn less than other essential workers. This pattern cannot be explained by differences in unionization rates and points to other differences in bargaining power, including institutional factors influencing the earnings of doctors and nurses. Care penalties have significant implications for the future supply of care services as the pandemic persists, highlighting the need to develop broad coalitions to challenge the undervaluation of care work.

The COVID-19 pandemic has had a dramatic effect on women's labor market outcomes. We assess the effects of state-level policies that mandated the closure of child care centers or imposed class size restrictions using a triple-differences approach that exploits variation across states, across time, and across women who did and did not have young children who could have been affected. We find some evidence that both of these policies increase the unemployment rate of mothers of young children in the short term. In the long term, the effects of mandated closures on unemployment become even larger and persist even after states discontinue closures, consistent with a permanent child care supply side effect.

I argue that the value of a particular set of committee assignments for a legislator is dependent on that legislator’s policy interests. By this, I mean that “good” assignments will match committee policy jurisdictions with member policy priorities. I develop this concept of committee-agenda matching and present a measure of this match for legislators in 12 state lower chambers. After some brief measure validation, I present a substantive application, demonstrating that this match poses serious consequences for individual legislator’s ability to shepherd their bills through the legislative process.

I exploit discontinuous changes in the likelihood of military service induced by the transition between different registration regimes under the World War I draft to determine if wartime service affected economic outcomes. A new 1900–1930 linked sample is constructed to study the short-term effects, while the 1960 census is used for the long-term analysis. Implementing a fuzzy regression discontinuity design and a difference-in-discontinuities approach, I find little evidence of a causal relationship between wartime service and economic status. This may be due to America's relatively short involvement in the war, coupled with comparatively less-generous benefits thereafter.

Intergenerational social mobility in the United States has declined over the last century, sparking a national debate about how to improve equality of opportunity. By analyzing data spanning the 20th century, we demonstrate strong temporal patterns operating across regions. Some areas of the United States have witnessed significant declines in social mobility, while others have had persistent low levels all along. Thus, the contemporary national picture is shaped by both powerful forces of change that reduce intergenerational mobility in some regions and deeply entrenched long-term forces generating persistence in others. It follows that improving social mobility will be challenging, as policy would need to respond to both forces and do so according to their varying mixture across different regions. New evidence shows that intergenerational social mobility—the rate at which children born into poverty climb the income ladder—varies considerably across the United States. Is this current geography of opportunity something new or does it reflect a continuation of long-term trends? We answer this question by constructing data on the levels and determinants of social mobility across American regions over the 20th century. We find that the changing geography of opportunity-generating economic activity restructures the landscape of intergenerational mobility, but factors associated with specific regional structures of interpersonal and racial inequality that have “deep roots” generate persistence. This is evident in the sharp decline in social mobility in the Midwest as economic activity has shifted away from it and the consistently low levels of opportunity in the South even as economic activity has shifted toward it. We conclude that the long-term geography of social mobility can be understood through the deep roots and changing economic fortunes of places.

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 election.

The events of the past several weeks remind us of the critical link between good health, employment, and financial security. We spent the past few months analyzing data on prime-age (25-54) people with disabilities and health issues in the US to understand their labor force participation and their experience with poverty. Although our interest in this topic predated the current crisis, what we learned offers important insights for policymakers as they consider ways to help low-income people in the coming months and years.

Ecological inference (EI) is the process of learning about individual behavior from aggregate data. We relax assumptions by allowing for “linear contextual effects,” which previous works have regarded as plausible but avoided due to nonidentification, a problem we sidestep by deriving bounds instead of point estimates. In this way, we offer a conceptual framework to improve on the Duncan–Davis bound, derived more than 65 years ago. To study the effectiveness of our approach, we collect and analyze 8,430 $2\times 2$ EI datasets with known ground truth from several sources—thus bringing considerably more data to bear on the problem than the existing dozen or so datasets available in the literature for evaluating EI estimators. For the 88% of real data sets in our collection that fit a proposed rule, our approach reduces the width of the Duncan–Davis bound, on average, by about 44%, while still capturing the true district-level parameter about 99% of the time. The remaining 12% revert to the Duncan–Davis bound.

Why does the United States lack affordable child care? I examine the long-term trend of the child care market and document a sharp price increase since the late 1990s. I show that a massive expansion of federal and state means-tested child care subsidies, which were intended to stimulate the market, instead crowded out child care supply. The evidence suggests that the subsidies discouraged home-based child care suppliers who were also working mothers. A simple calibrated equilibrium model captures the rising price, which eventually caused the female employment rate to decline. An effective policy should capitalize on the home-based care business.

The COVID-19 pandemic has spawned devastation across the United States. All segments of the population have been impacted, but people of color have borne the brunt of infections from the coronavirus and deaths from COVID-19. Nationally Latinos and Blacks are contracting the virus at rates three times higher than Whites,1 and Blacks are dying at a rate 3.6 times and Latinos 2.5 times higher than Whites.2 Furthermore, Blacks and Latinos have sustained major setbacks to their economic sustainability. The pandemic led to the loss of approximately 25 million jobs between February and April and a recovery of about 9 million jobs between then and June.3 These losses represent a historically unprecedented level of unemployment and while as of June some areas have exhibited a slow recovery, the near-term prospects for those who have lost jobs is uncertain at best. While the pandemic has affected everyone’s work life, it has done so unequally. Indeed, the rising job loss has been particularly dire for Blacks and Latinos who have experienced exceptionally high levels of unemployment and slow rates of job recovery.4

The three decades following the Second World War saw a period of economic growth that was shared across the income distribution, but inequality in taxable income has increased substantially over the last four decades. This work seeks to quantify the scale of income gap created by rising inequality compared to a counterfactual in which growth was shared more broadly. We introduce a time-period agnostic and income-level agnostic measure of inequality that relates income growth to economic growth. This new metric can be applied over long stretches of time, applied to subgroups of interest, and easily calculated. We document the cumulative effect of four decades of income growth below the growth of per capita gross national income and estimate that aggregate income for the population below the 90th percentile over this time period would have been $2.5 trillion (67 percent) higher in 2018 had income growth since 1975 remained as equitable as it was in the first two post-War decades. From 1975 to 2018, the difference between the aggregate WD[DEOHincome for those below the 90th percentile and the equitable growth counterfactual totals $47 trillion. We further explore trends in inequality by applying this metric within and across business cycles from 1975 to 2018 and also by demographic group

Educational assortative mating patterns in the United States have changed since the 1960s, but we know little about the effects of these patterns on children, particularly on infant health. Rising educational homogamy may alter prenatal contexts through parental stress and resources, with implications for inequality. Using 1969–1994 NVSS birth data and aggregate cohort-state census measures of spousal similarity of education and labor force participation as instrumental variables (IV), this study estimates the effects of parental educational similarity on infant health. Controlling for both maternal and paternal education, results support family systems theory and suggest that parental educational homogamy is beneficial for infant health while hypergamy is detrimental. These effects are stronger in later cohorts and are generally limited to mothers with more education. Hypogamy estimates are stable by cohort, suggesting that rising female hypogamy may have limited effect on infant health. In contrast, rising educational homogamy could have increasing implications for infant health. Effects of parental homogamy on infant health could help explain racial inequality of infant health and may offer a potential mechanism through which inequality is transmitted between generations.

Deep learning is one of the advanced approaches of machine learning, and has attracted a growing attention in the recent years. It is used nowadays in different domains and applications such as pattern recognition, medical prediction, and speech recognition. Differently from traditional learning algorithms, deep learning can overcome the dependency on hand-designed features. Deep learning experience is particularly improved by leveraging powerful infrastructures such as clouds and adopting collaborative learning for model training. However, this comes at the expense of privacy, especially when sensitive data are processed during the training and the prediction phases, as well as when training model is shared. In this paper, we provide a review of the existing privacy-preserving deep learning techniques, and propose a novel multi-level taxonomy, which categorizes the current state-of-the-art privacy-preserving deep learning techniques on the basis of privacy-preserving tasks at the top level, and key technological concepts at the base level. This survey further summarizes evaluation results of the reviewed solutions with respect to defined performance metrics. In addition, it derives a set of learned lessons from each privacypreserving task. Finally, it highlights open research challenges and provides some recommendations as future research directions

This article analyzes the impact arts institutions have on human capital attraction and gentrification in Census tracts experiencing different levels of population change. It addresses the ongoing interest in causal models of arts-driven neighborhood change and contextualizes the impact of arts institutions on neighborhoods with varying change trends. The analysis looked at U.S. Census and County Business Patterns data in 24,819 U.S. Census tracts between 2000 and 2010. Controlling for past trends of gentrification and different levels of ongoing population growth, multivariate regression modeling of tract change among propensity-score-matched tract groups suggests that new arts institutions have a positive effect on the number of highly educated residents and high-income groups, even in declining tracts. Furthermore, modeling suggests that new arts institutions are not strongly associated with an immediate change in resident composition associated with displacement gentrification, but may lead to displacement in the longer term.

Almost one-fifth of tropical forest cover has been lost in the last two decades, leading to local temperature increases that can surpass warming from 21st century climate change projections. Although it is known that working outdoors in high temperatures reduces worker productivity and increases heat strain, the extent to which deforestation-driven temperature change affects people across the tropics is unknown. Using satellite data combined with worker health guidelines, we show that warming associated with deforestation is already impacting human health and well-being in low latitude countries. We estimate that more than 2.5 million people in recently deforested tropical biome now face heat exposure that reduces safe work hours by at least two hours per day. This warming has particularly large impacts on populations in Brazil, Belize, Cambodia, Vietnam, Malaysia, Myanmar, Nigeria, and Cameroon. We highlight these effects by examining the Brazilian states of Mato Grosso and Pará, which have experienced particularly large-scale deforestation, and showing that temperatures and human heat exposure are increasing disproportionately quickly in these locations. Future global climate change magnifies heat exposure in deforested areas across the tropics; an additional 2°C of global warming will increase exposure to multiple hours of unsafe working conditions for over 1.6 million people in recently deforested areas. Tropical deforestation is hastening the arrival of climate change impacts in the tropics, highlighting the need to shift local land use practices and to slow global greenhouse gas emissions so the most vulnerable populations are not forced to bear the brunt of warming impacts.

Whether parents have health insurance coverage affects not only their health care access but the financial stability and well-being of the entire family. The harmful consequences of parents' uninsurance may be magnified for young children, given early childhood's importance for long-term growth and development. Moreover, compared with a vast body of evidence on maternal and child health, research on fathers' insurance coverage and its implications for children and families is limited. In this brief, we assess uninsurance among parents living with young children from birth to age 5, using National Health Interview Survey (NHIS) data from 2017 and 2018, and report separate estimates for mothers and fathers. Our main findings are as follows: - More than 4 million parents living with young children-2.2 million mothers and 2.2 million fathers-were uninsured in 2017-18, representing a 40 percent decline in uninsurance since 2013, following implementation of the major coverage provisions of the Affordable Care Act (ACA). - In 2018, fathers living with young children were more likely to be uninsured than mothers living with young children (15.2 percent versus 13.0 percent). Both mothers and fathers who were living with young children and were Hispanic, noncitizens, living in the South, lacking a high school diploma, or in a family with low income faced uninsurance rates greater than 20 percent. - Of the estimated 4.4 million uninsured parents living with young children in 2017-18, most were under age 35, about half were Hispanic, and more than half lived in the South. Compared with uninsured fathers, uninsured mothers living with young children were considerably less likely to be working and more likely to have low incomes. - Both mothers and fathers living with young children most commonly reported being uninsured because of cost, and many reported becoming uninsured after losing employer or public coverage. Uninsured fathers living with young children were more likely than such mothers to have gone more than three years without coverage. - Many uninsured parents living with young children also reported problems affording needed health care, low levels of health care use, and concerns about affording other basic needs. Many uninsured parents experience unmet health needs, financial hardships, and worries about affording the family's medical care, which can have adverse effects on both their and their children's health in the short and long run. Reducing uninsurance among parents-such as through Medicaid expansion in the remaining states, increasing coverage affordability and accessibility for people who are eligible for Medicaid or Marketplace coverage but not enrolled, and eliminating eligibility restrictions based on immigration status-would likely improve parents' ability to access and afford needed care and reduce related financial concerns. Moreover, expanding coverage for parents could improve family financial stability and long-term well-being for approximately 4.0 million children under 6 who live in a family with at least one uninsured parent.

Though U.S. life expectancy has increased over the past 50 years, this benefit has not been geographically uniform and certain rural persons and communities face a mortality gap.1,2 Rural residents experience a shorter life expectancy than urban residents,³ with higher mortality rates from specific causes such as chronic obstructive pulmonary diseases,⁴ coronary heart disease,⁵ and lung cancer.⁶ Overall, there are higher mortality rates among rural residents for all five leading causes of death – heart disease, stroke, cancer, unintentional injury, and chronic lower respiratory disease – as compared to urban residents.⁷ In addition, residents of poor rural areas had a mortality rate in 2005-2009 that was 42% higher than their affluent urban counterparts, and this disparity has increased over time.⁸ Prior studies suggest that individual and family socioeconomic characteristics are partially responsible for elevated rural mortality.⁹ For example, rural residents have lower incomes, lower educational attainment, and higher rates of being uninsured compared with urban residents;¹⁰ these characteristics have also been associated with higher mortality rates. A 2011 analysis of deaths in the U.S. in 2000 estimated that poverty and low education increased the risk of death by 75% and 80%, respectively, among adults aged 25-64.11 The rural-urban mortality difference is even larger for certain demographic groups. Poor blacks living in rural areas had two to three times the mortality risk as more affluent blacks and whites living in urban areas.⁸ Lack of access to health care may also explain the rural-urban mortality gap. Compared to urban places, rural communities have lower availability of primary care, particularly of specialty care, posing challenges to obtaining needed services for some rural residents. For example, rural residents are diagnosed with cancer at later stages of disease than those living in urban areas, which may be due to their more limited access to preventive care services.12 Diagnosis of cancer at a later stage would account for why residents of rural areas have lower cancer incidence rates but higher death rates than residents of urban areas.13 In addition, rural trauma deaths often occur outside the hospital setting, in contrast to urban areas.14 Further, among adults admitted to hospitals for a heart attack, rural residence is associated with higher rates of death, which may be due to differences in patient risk characteristics, such as higher comorbidity among rural residents, or to health care delivery challenges in rural areas.15,16

Users’ sensitive data can be collected and analyzed under local differential privacy (LDP) without the need to trust the data collector. Most previous work on LDP can be applied when each user’s data is generated and collected from a single service or data source. In a more general and practical setting, sensitive data of each user needs to be collected under LDP from multiple services independently and can be joined on, e.g., user id. In this paper, we address two challenges in this setting: first, how to prevent the privacy guarantee from being weakened during the joint data collection; second, how to analyze perturbed data jointly from different services. We introduce the notation of user-level LDP to formalize and protect the privacy of a user when her joined data tuples are released. We propose mechanisms and estimation methods to process multidimensional analytical queries, each with sensitive attributes (in its aggregation and predicates) collected and perturbed independently in multiple services. We also introduce an online utility optimization technique for multi-dimensional range predicates, based on consistency in domain hierarchy. We conduct extensive evaluations to verify our theoretical results using synthetic and real datasets.

Objective: To examine the changes in health insurance coverage, access to care, and health services utilization among nonelderly sexual minority and heterosexual adults between pooled years 2013-2014 and 2017-2018. Data Sources: Data on 3223 sexual minorities (lesbians, gay men, bisexual individuals, and other nonheterosexual populations) and 86 181 heterosexuals aged 18-64 years were obtained from the 2013, 2014, 2017, and 2018 National Health Interview Surveys. Study Design: Unadjusted and regression-adjusted estimates compared changes in health insurance status, access to care, and health services utilization for nonelderly adults by sexual minority status. Regression-adjusted changes were obtained from logistic regression models controlling for demographic and socioeconomic characteristics. Principal Findings: Uninsurance declined for both sexual minority adults (5 percentage points, P <.05) and heterosexual adults (2.5 percentage points, P <.001) between 2013-2014 and 2017-2018. Reductions in uninsurance for sexual minority and heterosexual adults were associated with increases in Medicaid coverage. Sexual minority and heterosexual adults were also less likely to report unmet medical care in 2017-2018 compared with 2013-2014. Low-income adults (regardless of sexual minority status) experienced relatively large increases in Medicaid coverage and substantial improvements in access to care over the study period. The gains in coverage and access to care across the study period were generally similar for heterosexual and sexual minority adults. Conclusions: Sexual minority and heterosexual adults have experienced improvements in health insurance coverage and access to care in recent years. Ongoing health equity research and public health initiatives should continue to monitor health care access and the potential benefits of recent health insurance expansions by sexual orientation and sexual minority status when possible.

Urban bikeshare systems have become increasingly popular in the U.S. in recent years. In this paper, I examine the effects of bikeshare systems on patterns of commuting to and from work in U.S. cities. To study this, I link cities across 2008 through 2016 American Community Survey (ACS) data and estimate the effects of bikeshare systems using both individual-level ACS records and city panel data. Event-study estimates suggest that bikeshare systems lead to a rise in bicycle commuting to and from work, and the effects of bikeshare systems are statistically and economically significant. I also find evidence of modal shifts after the introduction of bikeshare systems: While bicycle commuting rates increase, there is a decline in automobile commuting to and from work.