This paper presents evidence that firms had labor market power during the early 1930s. Using plant-level data from the Census of Manufactures between 1929 and 1935, I construct a Herfindahl-Hirschman Index of local labor market concentration at the State-Economic-Area-by-industry-by-occupation level. I find that local labor market concentration has a negative relationship with wages which is consistent with labor market monopsony power. The results are robust to excluding local labor markets with one firm, excluding industries with local product markets, as well as the inclusion of industry characteristic, SEA, and occupational time trends. I find evidence that New Deal minimum wage policies weakened monopsony power. I also find suggestive evidence that high unemployment rates during 1930 reduced workers’ bargaining power.

Technological innovation, such as self-driving trucks, threatens occupations, such as truck drivers, with sudden obsolescence. Using a bare-bones overlapping generations model, we examine an occupation facing possible obsolescence. The occupation pays ‘obsolescence rents,’ with fewer and older workers remaining in the occupation. We study teamsters at the dawn of the motor truck, current occupations threatened by computerization, and truckers dreading robotic trucks. As predicted, wages in threatened occupations rise, employment falls, and the occupations become ‘grayer’. Older workers become more likely to enter and less likely to exit the occupation than young ones and sometimes even increase in number.

America’s worst drought spanned the 1930s, coinciding with the most extensive environmental migration in United States history. Nearly 100 years later, we know little about who moved and who stayed. This paper studies heterogeneity in migration from drought by relating migration decisions recorded in the 1940 census to county drought conditions. Drought increased migration primarily for individuals with a 12th-grade education or higher. Drought migrants, both women and men, left rural and urban locations and most often relocated to rural destinations. These findings highlight the importance of individual-level characteristics for adaptation to climate shocks, challenge the perception that rural-to-urban is the dominant environmental migrant channel, and document the central importance of drought for internal migration during the 1930s.

This dissertation consists of three studies on political economy and economic history. The first study provides quasi-experimental evidence of electoral rewards for politicians who deviate from meritocracy in the selection of public sector personnel. The second study explores the intergenerational consequences of introducing old-age pensions. The third study, in joint work with Davis Kedrosky and Chiara Motta, investigates the wage effects of spatial competition in a monopsonistic market for Indigenous labor. While each chapter of this dissertation answers a separate research question, all three studies are motivated by a shared interest in the political economy of historical development. Whether studying patronage politics in Progressive Era New York City (Chapter 1), the first old-age pension program of the U.S. federal government during the Second Industrial Revolution (Chapter 2), or competition between colonial companies in Canada’s early modern fur trade (Chapter 3), this dissertation aims to shed light on the role of the state and quasi-state organizations in shaping the historical development of North America. In Chapter 1, I study patronage in personnel selection under the paradigmatic political ma- chine in the U.S. history, Tammany Hall in New York City. Electoral motives are frequently blamed for patronage in public organizations. But quantitative evidence on the behavior of patronage employees and whether they deliver an electoral return remains scarce. Focusing on the New York Police Department (NYPD) during 1900-1916 allows me to overcome the empirical challenges of estimating the electoral return to patronage: I collect new archival data to identify patronage employees, connect them to individual-level electoral responses, 1 2 and leverage a difference-in-differences design that compares patronage employees to the con- trol group of applicants who did not receive patronage. Linking NYPD patrolmen to their civil service exam results reveals that 21% of patrolmen in the period entered into police ser- vice despite lacking the required test scores. Consistent with historical narratives, I show that these patronage employees were more likely to be connected to the leaders of Tammany Hall, the city’s incumbent Democratic Party organization. Estimating a difference-in-differences design around the start of employment for patronage recipients, I find that patronage deliv- ered a 10.3% increase in electoral support within a 50 meter radius around the employee’s address (measured in the number of voters registering as Democrats). This electoral re- sponse – and complementary results on promotions tied to electoral support – suggest that patronage employees are incentivized to mobilize the votes of their neighbors. The electoral logic of patronage jobs in exchange for votes has important implications for performance: Patronage employees performed 22.7% worse than their meritocratically selected peers. In Chapter 2, I study the intergenerational effects of the first federal old-age pension law in the United States. In theory, when the state takes over services like old-age support, this could replace informal care by children who previously provided these services in the parental home, thereby enabling the next generation to move to better economic opportunities. In the context of an industrializing and urbanizing economy this means that state-support, by relaxing location constraints, could contribute to economic modernisation. I test this hypothesis using the 1890 Dependent and Disability Pension Act as a natural experiment. The 1890 Act transformed the Union Army Civil War pension into a federal old-age support program for Union veterans. Using restricted-access full-count census data, I track the sons of Union veterans and match them to their census records in 1870, 1880, 1900, and 1910. I provide difference-in-differences estimates of the effects of pension eligibility, comparing sons of pension-eligible fathers to sons born to ineligible men of the same generation before vs. after the 1890 Act got passed. I find that the pension reform decreased cohabitation between sons and eligible fathers by 1.6%, increased the share of sons settling in urban areas by 8.4%, and shifted affected sons out of farming and into better paying occupations. In Chapter 3, together with Davis Kedrosky and Chiara Motta, we assemble novel data from the account books of the Hudson’s Bay Company (HBC) to study the wage effects of a new entrant into a monopsonistic market for Indigenous labor. We show that in the setting of Canada’s early modern fur trade, where Indigenous labor was free and mobile between firms, increased competition improved wages for Indigenous workers. This unique case study allows us to isolate market structure as the main channel of labor market power and to quantify the impact on wages. We find that a 100km decrease in distance to the nearest competitor location was associated with a 1.5% increase in wages.

The average white male born in 1900 earned 2.6 times more labor income over their lifetime than the average Black male. This gap is nearly twice as large as the more commonly studied cross-sectional Black–white earnings gap because 48% of Black males born in 1900 died before the age of 30 as compared to just 26% of white males. We calibrate a model of optimal consumption in a world with mortality risk to data describing the life-cycle earnings and survival probabilities of Black and white males born between 1900 and 1970. We find that convergence in Black and white mortality rates led to a 50% reduction in Black–white welfare gaps between the 1900 and 1920 birth cohorts, even as cross-sectional Black–white income gaps for those cohorts remained relatively constant. However, the Black–white welfare gap stagnated for the 1920 to 1970 birth cohorts as gaps in Black–white life expectancy and income remained stable and large.

The cultural assimilation of immigrants into the host society is often equated with prospects for economic success, with religion seen as a potential barrier. We investigate the role of ethnic enclaves and churches for the assimilation of Danish Americans using a difference-indifferences setting. Following the ordination of a divisive religious figure in 1883, this otherwise small and homogeneous group split into rival Lutheran revivalist camps-so-called "Happy" and "Holy" Danes. The former sought the preservation of Danish culture and tradition, while the latter encouraged assimilation. We use data from the US census and Danish American church and newspaper archives, and find that Danish Americans living in a county with a "Happy" church chose more Danish names for their children. Newspapers read by "Holy Danes" saw a more rapid Anglicization of the language used. Religious beliefs thus facilitated assimilation. Divergence in behaviour only emerged following the religious division.

While New Deal scholarship has well established that racial divisions profoundly shaped the character and size of the American welfare state, the implications of this research for Asian Americans, particularly Chinese and Japanese Americans, remain unclear. This study moves beyond the White-Black binary that dominates analyses of New Deal programs by examining Chinese and Japanese Americans’ participation in the Works Progress Administration (WPA). Using full-count data from the 1940 census, we employ linear probability models and Oaxaca-Blinder decompositions to assess factors contributing to White-Chinese and White-Japanese employment differentials. While our results highlight the importance of differences in observable characteristics, the differential valuation of these characteristics and the unexplained portion of each employment differential suggest that discrimination led to Chinese and Japanese Americans’ underrepresentation in the WPA.

Background: Higher levels of body mass index (BMI), particularly for those who have obesity defined as class II and III, are correlated with excess risk of all-cause mortality in the USA, and these risks disproportionately affects marginalized communities impacted by systemic racism. Redlining, a form of structural racism, is a practice by which federal agencies and banks disincentivized mortgage investments in predominantly racialized minority neighborhoods, contributing to residential segregation. The extent to which redlining contributes to current-day wealth and health inequities, including obesity, through wealth pathways or limited access to health-promoting resources, remains unclear. Our quasi-experimental study aimed to investigate the generational impacts of redlining on wealth and body mass index (BMI) outcomes. Methods: We leveraged the Panel Study of Income Dynamics (PSID) and Home Owners’ Loan Corporation (HOLC) maps to implement a geographical regression discontinuity design, where treatment assignment is randomly based on the boundary location of PSID grandparents in yellowlined vs. redlined areas and used outcome measures of wealth and mean BMI of grandchildren. To estimate our effects, we used a continuity-based approach and applied data-driven procedures to identify the most appropriate bandwidths for a valid estimation and inference. Results: In our fully adjusted model, grandchildren with grandparents living in redlined areas had lower average household wealth (β = − $35,419; 95% CIrbc − $37,423, − $7615) and a notably elevated mean BMI (β = 7.47; 95% CIrbc − 4.00, 16.60), when compared to grandchildren whose grandparents resided in yellowlined regions. Conclusion: Our research supports the idea that redlining, a historical policy rooted in structural racism, is a key factor contributing to disparities in wealth accumulation and, conceivably, body mass index across racial groups.

Why are some ethnic or immigrant groups vastly over-represented in certain oc- cupations? I posit that these ethnic-occupational niches are driven by immigrants’ pre- migration skills, as well as social networks of fellow countrymen in the niched occu- pation. Turning to the United States during the Age of Mass Migration, I show that between 1850 and 1940, 46% of all immigrants were in niches. The presence and inten- sity of over-representation varied across nationality, time, and occupation. Digitizing new data on immigrant pre-arrival occupation and merging it to the historical US cen- suses, I find both skill background and social networks have a significant, positive effect on the occupational choice of newly arrived immigrants. But, the latter domi- nates the former, and a novel instrument for social networks of one’s countrymen in a niched occupation confirms these results and the primacy of social networks in caus- ing niches. By providing a better understanding of what drives niches, my findings inform their consequences for immigration and labor markets today.

This paper estimates the effect of the OxyContin reformulation and the subsequent heroin crisis on grandparent caregiving in the United States—grandparents who are primarily responsible for their grandchildren. The empirical approach is a difference-in-differences and a series of event studies that exploit variation in the timing of the reformulation of OxyContin and geographic variation in pre-reformulation rates of nonmedical use of OxyContin across states. I find that a standard deviation increase in pre-reformulation nonmedical use of OxyContin leads to higher grandparent caregiving. Increases range from 2.5 to 8.5%, on average, relative to the baseline mean. This change is more pronounced among grandparents between 46 and 65 years of age. When I examine changes in heroin-related mortality as a possible mechanism, I find modest increases in grandparent caregiving when no parents are present or when no or only one parent is present of 1.80 and 1.50% of the baseline mean, on average.

Multiple instance regression (MIR) was introduced by Ray and Page [2001] as an analogue of multiple instance learning (MIL) in which we are given bags of feature-vectors (instances) and for each bag there is a bag-label which matches the label of one (unknown) primary instance from that bag. The goal is to compute a hypothesis regressor consistent with the underlying instance-labels. A natural approach is to find the best primary instance assignment and regressor optimizing the mse loss on the bags though no formal generalization guarantees were known. Our work is the first to prove generalization error bounds for MIR when the bags are drawn i.i.d. at random. Essentially, with high probability any MIR regressor with low error on sampled bags also has low error on the underlying instance-label distribution. We next study the complexity of linear regression on MIR bags, shown to be NP-hard in general by Ray and Page [2001], who however left open the possibility of arbitrarily good approximations. Significantly strengthening previous work, we prove a strong inapproximabil-ity bound: even if there exists zero bag-loss MIR linear regressor on a collection of 2-sized bags with labels in [−1, 1], it is NP-hard to find an MIR linear regressor with bag-loss < C for some absolute constant C > 0. Our work also proposes a model training method for MIR based on a novel weighted assignment loss, geared towards handling overlapping bags which have not received much attention previously. We conduct empirical evaluations on synthetic and real-world datasets showing that our method outperforms the baseline MIR methods.

We present new evidence on the long-run trend of occupational task content by race in the United States, 1900-2021. Black workers began the transition to better paid, cognitive-intensive modern jobs at least a generation after white workers; substantial convergence only occurred from 1960 onwards. Longitudinal data suggests that transitions to new task content were racially biased: Black men moved to jobs with lower rewarded task content than white men, conditional on initial task content, though gaps decreased after World War II. Routine-intensive Black workers were less likely to move up into non-routine analytic work compared to white workers in both historical and modern periods. The results suggest that task-displacement shocks, such as automating routine-manual work, widen Black-white inequality.

We examine aggregate economic gains in the United States as the railroad network expanded in the 19th century. Using data from the Census of Manufactures, we estimate relative increases in county aggregate productivity from relative increases in county market access. In general equilibrium, we nd that the railroads substantially increased national aggregate productivity. By accounting for input distortions, we estimate much larger aggregate economic gains from the railroads than previous estimates. Our estimates highlight how broadly-used infrastructure or technologies can have much larger economic impacts when there are ineciencies in the economy. For helpful comments and suggestions, we thank the editor, referees, and many colleagues and seminar participants at: Berkeley, Brown, U.S.

Labor unions have many well-documented effects on economic outcomes that are plausibly related to family formation. I study the impact of unionization on fertility using evidence from the largest expansion of unionism in American history: the enactment of the 1935 National Labor Relations Act (NLRA). I introduce new estimates of local union membership and exploit variation in exposure to the NLRA shock to estimate the place level effect of union growth on fertility outcomes. Unionization has positive effects on birth rates and completed fertility, and can account for approximately 20% of overall fertility increases during the Baby Boom. Effects are driven primarily by wage growth, protection against adverse labor market shocks, and impacts on female labor force participation.

The Great Fire of 1910 in the northwestern United States burnt more than 1.2 million hectares in just two days and stands as one of the largest wildfires ever recorded. While it is known for having led to the introduction of a rigorous fire suppression regime that lasted for much of the twentieth century, it also generated a considerable amount of smoke far beyond the burnt areas that is likely to have impacted the health of those exposed. This paper examines the short- and long-term impact of this fire-sourced smoke pollution on children, combining historical data with smoke emission and dispersion modelling. The econometric results indicate a 119% increase in excess mortality during the week of the fire and a decrease of 4–14% in later-life socioeconomic status scores 20 and 30 years after the event. This research offers novel insights into wildfire smoke repercussions on health and long-run human capital formation in a setting where avoidance behaviour was minimal.

This paper studies the consequences of displacement resulting from the construction of the Interstate Highway System in the United States. It starts by showing that the placement of urban highways was not random. Exploiting cross-sectional variation between neighborhoods in 1950, the last census before the 1956 Federal-Aid Highway Act, I show that tracts with a higher proportion of the city’s Black population were more prone to receive a highway, even after controlling for socio-demographic and geographic characteristics of the tract. Subsequently, the paper examines the impact on affected indi- viduals. By geocoding the full-count 1940 census and linking it to administrative mortality records, I track affected individuals before and after highway construction. Exploiting the quasi-random variation in close proximity to highway developments, I show that displaced individuals die at a younger age, exhibit a higher likelihood of relocating from their neighborhood, and lived in neighborhoods with lower socioeconomic characteristics at their time of death. Moreover, although more nuanced, individuals residing near a highway are also negatively impacted. The mortality results are explained by the relocation of individuals into neighborhoods with lower health outcomes. These effects show up in the aggregate. By running an event study design for neighborhoods between 1930 and 2020, I found that highway construction is associated with outmigration, explained by a relative decrease in the Black population. I also find that these effects spill over to adjacent neighborhoods, affecting the demographic dynamics of cities.

Black-White mortality crossover is a well-studied demographic paradox. White mortality is lower than Black mortality until advanced ages, when Black mortality becomes lower than White mortality.

The proportion of elderly people living apart from children increased dramatically during the twentieth century. Most observers attribute this change to increasing incomes, but Ruggles argues that attitudes became less favorable to coresidence with children. We use the introduction of the Pennsylvania Railroad Pension to examine the role of income in residential decisions of the elderly. In 1900 the Pennsylvania Railroad announced that all employees over the age of 70 would be retired from service with pensions. Pensions were substantial, about 40% of pre-retirement wages, and workers in poor health could retire starting at age 65. We linked retirees from pension records to censuses to examine the relationship between incomes and household patterns in 1900, 1910, and 1920. Although men living with their wives were not affected, we find direct relationships between pension incomes and early retirement and household structure among widowers. Widowers with higher incomes were more likely to head their own households and less likely to live with a married child. Men who retired early because of poor health were less likely to remain heads of household and more likely to live with a married child. These patterns are consistent with Laslett’s ‘nuclear hardship’ hypothesis, and they support the view that higher incomes later in the century would have reduced coresidence with children, especially married children.

Automated census linkage algorithms have become popular for generating longitudinal data on social mobility, especially for immigrants and their children. But what if these algorithms are particularly bad at tracking immigrants? This study utilizes a database on nineteenth-century Irish immigrants, generated from the most widely used algorithms, created by Abramitzky, Boustan, and Eriksson (ABE). Our objective is to assess the extent to which different individuals are erroneously linked together across census years and the consequences of these “false positives” for calculating social mobility. Our findings raise serious questions about the quality of the matches generated by the “first generation” of automated census linkage algorithms. False positives range from about one-third to one-half of all links. These bad links lead to sizeable estimation errors when measuring Irish immigrant social and geographic mobility.

The proportion of elderly people living apart from children increased dramatically during the twentieth century. Most observers attribute this change to increasing incomes, but Ruggles argues that attitudes became less favorable to coresidence with children. We use the introduction of the Pennsylvania Railroad Pension to examine the role of income in residential decisions of the elderly. In 1900 the Pennsylvania Railroad announced that all employees over the age of 70 would be retired from service with pensions. Pensions were substantial, about 40% of pre-retirement wages, and workers in poor health could retire starting at age 65. We linked retirees from pension records to censuses to examine the relationship between incomes and household patterns in 1900, 1910, and 1920. Although men living with their wives were not affected, we find direct relationships between pension incomes and early retirement and household structure among widowers. Widowers with higher incomes were more likely to head their own households and less likely to live with a married child. Men who retired early because of poor health were less likely to remain heads of household and more likely to live with a married child. These patterns are consistent with Laslett’s ‘nuclear hardship’ hypothesis, and they support the view that higher incomes later in the century would have reduced coresidence with children, especially married children.