The recent digitization of complete count census data is an extraordinary opportunity for social scientists to create large longitudinal datasets by linking individuals from one census to another or from other sources to the census. We evaluate different automated methods for record linkage, performing a series of comparisons across methods and against hand linking. We have three main findings that lead us to conclude that automated methods perform well. First, a number of automated methods generate very low (less than 5%) false positive rates. The automated methods trace out a frontier illustrating the tradeoff between the false positive rate and the (true) match rate. Relative to more conservative automated algorithms, humans tend to link more observations but at a cost of higher rates of false positives. Second, when human linkers and algorithms have the same amount of information, there is relatively little disagreement between them. Third, across a number of plausible analyses, coefficient estimates and parameters of interest are very similar when using linked samples based on each of the different automated methods. We provide code and Stata commands to implement the various automated methods.

We study private synthetic data generation for query release, where the goal is to construct a sanitized version of a sensitive dataset, subject to differential privacy, that approximately preserves the answers to a large collection of statistical queries. We first present an algorithmic framework that unifies a long line of iterative algorithms in the literature. Under this framework, we propose two new methods. The first method, private entropy projection (PEP), can be viewed as an advanced variant of MWEM that adaptively reuses past query measurements to boost accuracy. Our second method, generative networks with the exponential mechanism (GEM), circumvents computational bottlenecks in algorithms such as MWEM and PEP by optimizing over generative models parameterized by neural networks, which capture a rich family of distributions while enabling fast gradient-based optimization. We demonstrate that PEP and GEM empirically outperform existing algorithms. Furthermore, we show that GEM nicely incorporates prior information from public data while overcoming limitations of PMWPub, the existing state-of-the-art method that also leverages public data.

Between 1940 and 1970, eight million white southerners left the US South and settled in states outside that region bringing with them their distinctive political preferences, religious practices, and racial attitudes. I demonstrate that their settlement had a large effect on the Republican vote share in northern and western counties even in the 2020 election, fully 50 years later. The political impact of these southern migrants was largely due to their adherence to evangelical Protestantism. I also show that the migration caused a long-lasting increase in anti-Black hate crimes that is reflected in measures of implicit and explicit racial bias of the white population. I obtain causal estimates of these effects through the use of a shift-share instrument to isolate plausibly exogenous variation in the southern push-factors that drove the migration. Importantly, I show using survey data that these effects partly operated through a spillover on non-southerners. The impact of southern migrants on the beliefs, actions, and politics of northern and western communities was not simply the result of a compositional change of the population.

This study evaluates differences in parent–child leisure and parent affect across single- and two-parent families. The Ecology of Family Experiences framework suggests contextual factors such as family structure and the novelty of the activity environment contribute to heterogeneity in how parents experience leisure, partly because constraints may differ across family structures. Using a large, nationally representative data sample of parents from the American Time Use Survey (N = 78,353), this study shows single-parents experience leisure deficits compared to other parents, and leisure deficits are greatest in home-based leisure. Additionally, using a subsample (N = 16,214), we found that at-home leisure is more meaningful for single parents than other parents, suggesting avoidance behaviors do not drive differences. Instead, structural constraints like time and money likely curb leisure in single-parent homes. Findings have policy implications, because many leisure programs target away-from-home leisure which is less restorative to single parents.

Although the fairness community has recognized the importance of data, researchers in the area primarily rely on UCI Adult when it comes to tabular data. Derived from a 1994 US Census survey, this dataset has appeared in hundreds of research papers where it served as the basis for the development and comparison of many algorithmic fairness interventions. We reconstruct a superset of the UCI Adult data from available US Census sources and reveal idiosyncrasies of the UCI Adult dataset that limit its external validity. Our primary contribution is a suite of new datasets derived from US Census surveys that extend the existing data ecosystem for research on fair machine learning. We create prediction tasks relating to income, employment, health, transportation, and housing. The data span multiple years and all states of the United States, allowing researchers to study temporal shift and geographic variation. We highlight a broad initial sweep of new empirical insights relating to trade-offs between fairness criteria, performance of algorithmic interventions, and the role of distribution shift based on our new datasets. Our findings inform ongoing debates, challenge some existing narratives, and point to future research directions.

We analyze the 25-year trajectory of Temporary Assistance for Needy Families (TANF), the nation's cash safety net for families with children. Since its creation, TANF's story has been one of greatly diminished access, low benefits, and states diverting funding away from cash assistance, changes that have disproportionately affected the assistance available to Black families. We document how the weakened safety net under TANF contributed to a rise in deep poverty among children in single-mother families and how two defining features of TANF-work requirements and time limits-contributed to TANF's weakness. Finally, we make policy recommendations to address TANF's failures.

Tent encampments have become an especially common form of homelessness in West Coast cities like Oakland, California, where the number of people living in tent encampments increased by 130 percent between 2017 and 2019. Relative to other experiences of homelessness, tent encampments provide residents both potential benefits and risks that depend on their location, size, and stability. Using novel data collected from Google Street Views, I document the growth and spatial dynamics of tent encampments in West and Central Oakland over the last decade. The number and size of tent encampments rapidly increased between 2014 and 2019, varying widely in their stability. City interventions like the city’s outdoor transitional housing sites displaced several large tent encampments. Combined with overall tent encampment growth, these displacements dispersed tent encampments throughout both nearby neighborhoods and other parts of the city

The Historical Roots of Human Trafficking pp 9–22Cite as Capitalism, Colonialism, and Imperialism: Roots for Present-Day Trafficking Download book PDF Download book EPUB Capitalism, Colonialism, and Imperialism: Roots for Present-Day Trafficking Gonzalo Martínez de Vedia Chapter First Online: 23 May 2021 419 Accesses Abstract Modern definitions of human trafficking are the most encompassing in the history of official action against it. This rightfully identity-blind framework risks creating naively identity-blind plans of action. This chapter will argue that anti-trafficking practitioners must allow that both force and coercion can be used against any person in any setting and also understand that historically, and in the present, certain othered groups in certain sectors of the economy have been disproportionally affected by human trafficking. To understand and unravel the forces that maintain this trend in the US context in particular, practitioners must unpack how unfree labor evolved in the region through several major historical chapters: the launch and growth of European imperialist projects in the Americas, early US colonial history, and the rise and boom of US capitalism. By outlining the unique characteristics of unfree labor in these periods, the chapter offers practitioners a rehistoricized understanding of the scope and character of forced and coerced labor in the modern US context.

The unemployment rate has sharply increased as a result of the lockdown associated with the spread of COVID-19. The negative effect of the lockdown is more conspicuous among the less-educated workers than the highly-educated workers. Because Asian Americans are more likely to have a bachelor or higher degree than any other racial group, they are expected to be relatively immune to the drop in employment unless the detrimental impact of the lockdown is severer for Asian Americans. Exploiting the panel aspect of the Current Population Survey – Merged Outgoing Rotation Group, we examine the changes in At-work status before and after the lockdown and between the lockdown and months of the reopening. The empirical results uncover that Asian Americans are more negatively affected by the lockdown than any other racial group, net of education, immigration status, and other covariates. Surprisingly, the negative impact of the lockdown is entirely concentrated on less-educated Asian Americans. Regardless of gender, less-educated Asian Americans are substantially more likely to lose employment than equally educated Whites and are not more likely to regain employment during the reopening months. Other less-educated racial minorities do not experience more reduction in At-work status than Whites, net of covariates. Highly-educated Asian Americans’ employment is equally affected by the lockdown with equally educated Whites.

A manufacturing wage premium is the average percentage difference between the wage a worker earns in manufacturing and the wage similar workers earn in industries other than manufacturing. Using standard wage regressions, we find that, between 1979 and 2018, the manufacturing wage premium declined much more for production workers (such as machine operators) than for nonproduction workers (such as managers or administrative assistants). As a result, the production-workers’ wage premium was 4 percent during 2015 to 2018, while the nonproduction-workers’ wage premium was 14 percent. The decline in the production-workers’ wage premium is broad based, both demographically and geographically. We argue that this decline was most likely caused by the loss of manufacturing production jobs, which put downward pressure on wages for this work. The decline in manufacturing wage premiums has important implications for the difficulties associated with recruiting manufacturing workers and for public investment in training for manufacturing jobs.

This paper proposes a novel strategy for identifying the effects of import competition on economic outcomes that avoids standard concerns related to the endogeneity of trade policy and provides a consistent measure of exposure to trade over time. Conditioning on the level of import tariffs, our approach exploits cross-industry differences in the relative importance of specific rather than ad valorem tariffs. As they are expressed in per unit terms rather than as a share of value, the effective protection provided by a given specific tariff varies with price levels. Using digitized tariff line data between 1900 and 1940, we relate inflation-driven changes in trade protection to changes in imports and labor market outcomes in the full count U.S. census. We show that our measure predicts import growth at both the industry and county level. Using our measure as an instrument, we show that import competition reduces labor force participation in traded sectors during this period. Labor market effects are widespread but fall most heavily on those with little experience or fewer outside labor market options: the young, seniors, and those in rural areas.

Water quality trading (WQT) programs aim to efficiently reduce pollution through market-based incentives. However, WQT performance is uneven; while several programs have found frequent use, many experience operational barriers and low trading activity. What factors are associated with WQT existence, prevalence, and operational stage? In this paper, we present and analyze the most complete database of WQT programs in the United States (147 programs/policies), detailing market designs, trading mechanisms, traded pollutants, and segmented geographies in 355 distinct markets. We use hurdle models (joint binary and count regressions) to evaluate markets in concert with demographic, political, and environmental covariates. We find that only one half of markets become operational, new market establishment has declined since 2013, and market existence and prevalence has nuanced relationships with local political ideology, urban infrastructure, waterway and waterbody extents, regulated environmental impacts, and historic waterway impairment. Our findings suggest opportunities for better projecting program need and targeting program funding.

Background: Exposure to excessive heat is associated with a higher risk of death. Although the relative risk of death on extreme-heat days has decreased over the past several decades in the United States, the drivers of this decline have not been fully characterized. In particular, while extreme heat earlier in the warm season has been shown to confer greater risk of mortality than exposure later in the season, it is unknown whether this within-season variability in susceptibility has changed over time and whether it is modified by region, climatic changes, or social vulnerability. Methods: We used distributed-lag nonlinear models and meta-regression to estimate the association between ambient maximum daily temperature during the early, late, and overall warm seasons and the relative risk of mortality for two decades, 1973–1982 and 1997–2006, in 186 metropolitan areas in the United States. We assessed changes in relative risk nationally, regionally, and between places with differential changes in early-season relative extreme heat and indicators of social vulnerability. Results: Most of the reduction in heat-related mortality nationally between the two decades is driven by decreases in late-season mortality, while substantial early-season risk remains. This difference is most apparent in the Northeast, in cities with greater increases in early-season relative extreme heat, and in places that have become more socially vulnerable. Conclusions: Early-season heat mortality risks have persisted despite overall adaptations, particularly in places with greater warming and increasing social vulnerability. Interventions to reduce heat mortality may need to consider greater applicability to the early warm season.

Health care was found to be one of two industries (the other being education) that experienced the fastest growth in nonstandard work arrangements (i.e., work arrangements other than full-time, year-round) between 1995 and 2015.1 The US Bureau of Labor Statistics (BLS) definition of nonstandard work arrangements (referred to as nontraditional work arrangements by the US Government Accountability Office) includes overlapping concepts of: 1) contingent work (based on self-employment status, length of work, method of payment, and connection to employer), 2) alternative work (i.e., temporary agency worker, on-call worker, contract company worker, and independent contractor), and 3) electronically-mediated work (sometimes referred to as “gig work”).2,3 Nonstandard work arrangements as well as self-employment alone are not new in health care. Not well-documented is which health care occupations are frequently involved in these work arrangements, and the characteristics of these workers, a gap which this study aims to address. This study used data from the Contingent Worker Supplement (CWS) to the Current Population Survey (CPS) that was fielded in May 2017 by the US Census Bureau for the BLS. Using BLS definitions, we examined health care workers across three work arrangements: 1) self-employment, 2) contingent work, and 3) alternative work. We did not analyze health care workers involved in electronically-mediated work given BLS concerns about false positives in the survey responses.4

Adoption of health information and communication technologies ("HICT") has surged over the past two decades. We survey the medical and economic literature on HICT adoption and its impact on clinical outcomes, productivity and labor. We find that HICT improves clinical outcomes and lowers healthcare costs, but (i) the effects are modest so far, (ii) it takes time for these effects to materialize, and (iii) there is much variation in the impact. More evidence on the causal effects of HICT on productivity is needed to guide further adoption. There is little econometric work directly investigating the impact of HICT on labor, but what there is suggests no substantial negative effects on employment and earnings. Overall, while healthcare is "exceptional" in many ways, we are struck by the similarities to the wider findings on ICT and productivity stressing the importance of complementary factors (e.g. management and skills) in determining HICT impacts.

Stein and Yannelis (2020) (SY) study the short-term impact of the Freedman's Savings Bank on human capital, labor market, and wealth outcomes. This short note is a response to Célérier and Tak (2021) (CT), which offers comments on SY, claiming to "empirically reject the assumptions of the study's identification strategy" and arguing that "financial inclusion can be detrimental to minorities." We show their claims are driven by a serious data error and by omitting data, and that their empirical tests do not evaluate the identification assumption in SY. After using an alternative matching strategy which throws out four-fifths of matches, CT present estimates with very large standard errors. We show that these estimates cannot statistically reject large effects, including many of the point estimates in SY.

This paper disaggregates the GTAP U.S. labor data into 20 worker types, by gender, education, and broad occupation category, to understand how different workers in the U.S. are exposed to hypothetical changes in tariff rates. First, a methodology is provided to disaggregate the U.S. labor input in GTAP into twenty types. A database of wage bill shares and mean wages by GTAP sector are calculated using 2017 data from the Current Population Survey (CPS) Annual Social and Economic Supplements (ASEC). The wage bill share matrix can be used as a tool to split wage payments in GTAP to each labor type. Second, the paper provides illustrative simulations with both manufacturing and services trade, demonstrating differential impacts on worker types for each of the scenarios considered.

The COVID-19 pandemic has upended in-person public education across the United States, a critical infrastructure of care that parents—especially mothers—depend on to work. To understand the nature and magnitude of school closures across states, we collected detailed primary data—the Elementary School Operating Status database (ESOS)—to measure the percentage of school districts offering in-person, remote, and hybrid instruction models for elementary schools by state in September 2020. We link these data to the Current Population Survey to evaluate the association between school reopening and parents’ labor force participation rates, comparing 2020 labor force participation rates to those observed prepandemic in 2019. We find that, across states, the maternal labor force participation rate fell to a greater extent than that of fathers. In 2019, mothers’ rate of labor force participation was about 18 percentage points lower than fathers’. By 2020, this gap grew by 5 percentage points in states where schools offered primarily remote instruction. We show that schools are a vital source of care for young children, and that without in-person instruction, mothers have been sidelined from the labor force. The longer these conditions remain in place, the more difficult it may be for mothers to fully recover from prolonged spells of nonemployment, resulting in reduced occupational opportunities and lifetime earnings.

This thesis studies possible factors associated with the earnings differences between Finnish and Swedish immigrants at the beginning of the 20th century in the United States. Economics literature has provided multiple explanations for the economic assimilation of immigrants. Fundamentally immigrant assimilation in economic theory is seen as a process of country-specific human capital accumulation. Many factors, however, can affect this process. In this thesis, I will test whether the earnings differences between Finnish and Swedish immigrants were associated with their location choices, size of the immigrant communities, English skills and linguistic distance between immigrant’s mother tongue and English. In addition, I test whether even within industries, the earnings differences persist and whether in addition to Swedes, Finns also differ from other Nordic immigrants. To do this, I will use a longitudinal dataset of 28 621 European male immigrants observed in 1900, 1910 and 1920 and use pooled OLS to estimate the earnings differences of Finnish and Swedish immigrants at different stages of their migration period. First, I find that at the time of arrival or at most 5 years after it, Finnish immigrants earn more than otherwise comparable Swedish immigrants. After more than 30 years in the United States, however, Finns earn substantially less than their Swedish counterparts. Also, not only did the Finns diverge from the earnings of the Swedes with the time spent in the United States, but also from their initial level of earnings. I then find that controlling for location choices and the size of the immigrant communities slightly reduce the earnings gap between long-term Finnish and Swedish immigrants. Differences in English skills, on the other hand, are not found to be associated with the earnings differences. Instead, I find that long-term Finnish and Swedish immigrants have almost equal probabilities of knowing English and that Finnish immigrants accumulate language skills at a substantial rate after arrival. The negative earnings gap among long-term Finnish and Swedish immigrants persists even within industries. However, although remaining negative, controlling for industry fixed effects reduces the earnings gap by over 50 %, suggesting that a large fraction of the negative earnings gap was associated with Finnish immigrants being active in low-paying industries. Finally, I find that the similar pattern applies not only to Swedish immigrants but also to other Nordic immigrants. At the time of arrival, Finns seem to earn more than otherwise comparable Nordic immigrants but after more than 30 years in the United States, Finns earn substantially less than their Nordic counterparts.

This report seeks to document and explore North Carolina’s twin challenges of child hunger and poverty. They constitute, jointly, one of the state’s most wrenching and most embarrassing problems. They are, as well, massively inadequately attended to, constituting little of our public policy discourse, deliberation and legislative focus. Ignoring the pervasive dignity- and opportunity-denying difficulties of so many of our youngest and most vulnerable members is increasingly impossible to square with our foundational commitments and declarations, our constitutions and our creeds