Historical Census Data (100% Sample)

Access to the complete count U.S. Census decennial datasets (1790-1940) with names and addresses is provided to University of Michigan researchers through a Virtual Data Enclave. These data are made available through a license with the Minnesota Population Center. See the IPUMS website for descriptions of available data.

Researchers interested in applying to access these data at the University of Michigan should complete a Research Project Application and a Researcher Agreement. These documents should be emailed to Historical.Census@umich.edu

Any research produced using these data must cite:
Minnesota Population Center and Ancestry.com. IPUMS Restricted Complete Count Data: Version 1.0 [Machine-readable database]. Minneapolis: University of Minnesota, 2013.

Current Projects

Project Title (hover over to view abstract)	Primary Investigator	Graduate Advisor
We will compare our analysis of the determinants of self-employment and entrepreneurial activity among Cuyahoga County residents, residents of other industrial cities, and the country as a whole with our analysis of a dataset of approximately 3000 Cuyahoga County inventors and entrepreneurs.	Levenstein
This project uses census micro-data to examine the impact of natural disasters (as defined by the American Red cross) in the 1920s and 1930s. Special attention is paid to migration behavior as revealed from a panel sample linked from the 1940 sample to earlier censuses.	Rhode
The proposed project will link vital statistics birth and child mortality records to the 1940 census by name. To date, we have recovered over 40 million microdata records on births from 1900 to 1940. Birth records contain information on the infant’s full name, exact day and place of birth, sex and race and also the names of parents, parents’ day and place of birth, and typically parents’ occupations and race. Child death records contain the child’s full name, the exact day of the death, sex, age, race, and birthplace as well as parents’ names and sometimes parents’ birthplace and titles and address. We will link these data to the 1940 census to determine match rates, determinants of non-matches, and geographic coverage and representativeness of Vital Statistics microdata.	Bailey
This project will examine the causal outcomes of electoral participation for African Americans in the 1940s in the American south. Although the first half of the 20th century is widely regarded as a period of effective disenfranchisement for blacks in the south, there was variation in black voter registration rates over this period. However, registration was kept to a minimum because of violent and non-violent means of oppression and intimidation, a significant component of which was the application of economic pressure (e.g., by denying a bank loan to a black farmer who tried to register to vote). I propose to use the occupation mix of blacks at the county level to instrument for voter turnout, with the idea being that blacks who were employed by railroads were not dependent on the local economic power structure for income, and therefore faced a lower cost of voting. I will, in turn, then look for causal effects of voting on African-Americans’ health: did counties with a larger proportion of black voters because of the occupation mix in that county, experience on average lower mortality rates later on? And if so, what are the mechanisms for this –increased education funding? Better infrastructure?	Henderson	Bailey
While there has been much research examining the impact of compulsory schooling laws in place after 1915, there is little evidence regarding the efficacy of schooling laws in earlier years. In this project, we use the 1940 Census to examine the impact of schooling laws (including related laws that affected school attendance such as child labor laws) on schooling attainment and the economic returns to schooling.	Stephens
During the Second World War, the federal government assumed an unprecedented degree of control over the US economy. At the peak, the share of federal government expenditures in GNP soared to 44 percent, a level never attained before or since (the level has not even exceeded 25 percent in the post-WWII era). In addition to enrolling 16.4 million Americans (that is, about one-eighth of the 1940 population) in the armed forces, the federal government spent $196 billion between June 1940 and June 1945 on military supply contracts and $31 billion on investments in new facilities. In today’s purchasing power, this total would come to roughly two trillion dollars. This paper uses county-level economic and political data to investigate the determinants of the geographic allocation of World War II-era military spending, both for major war supply contracts and for new facility projects. We seek to compare the roles of pre-war armament capacity and political economy. We seek to use 1940 local micro-demographic data on detailed occupations to generate a more fine-grained mapping of military capacity and population mobility. We also seek to combine the census micro-data with political preference estimates from the Gallup polling data to produce a more fine-grained mapping of the location of swing and bitable voters.	Rhode
Names convey information regarding gender, race, and ethnicity, and there is some evidence that, in recent times, labor market outcomes respond to the degree to which names signal race or ethnicity. I propose to use the 1920 Census to explore the extent to which immigrants engaged in name-changing upon arrival to the United States in order to signal their intent to remain in the country, rather than migrate back to the home country after a short period (which was the case for up to 30% of immigrants). I will document this by creating a “native-ness index” using the names of natives to the US, and then assigning to each foreign-born individual the “native-ness” of his or her name relative to the average of the native-ness among all immigrants from that country of origin. This will reveal the extent to which the average immigrant from sending country changes his name to sound “American” relative to other immigrants from that country. I will then explore whether this can be interpreted as a signal of intent to remain in the United States, rather than migrate back to the home county, by exploring the relationship between the likelihood of name changing and home ownership among immigrants. The 1920 full-count Census is crucial for this project, as it contains birthplace, arrival date, and whether or not the respondent owned or rented a home.	Henderson	Bailey
Between 1907 and 1937, thirty-two U.S. states passed eugenic sterilization laws that authorized state institutions for the feebleminded and mentally ill to sterilize residents who were determined to be unfit for reproduction. Eugenic efforts to improve the reproductive fitness of the population were closely aligned with public health, and over 60,000 sterilizations were officially recorded nationwide. California led the nation in sterilization practice with approximately 20,000 sterilizations. Preliminary evidence suggests that women and racial/ethnic minorities were sterilized at higher rates than other groups. We have compiled individual-level data on 18,000 individuals recommended for sterilization in nine state homes and hospitals across California. We propose to link sterilization data to census microdata on institution populations (1920-1940) to formally compare rates of sterilization across demographic groups. Of particular interest is ethnic bias in sterilization practice. We will identify individuals with Spanish surnames to compare rates of sterilization between residents of Mexican origin and non‐ Hispanic residents. Access to census microdata will also allow us to compare rates of institutionalization across demographic groups.	Stern
A considerable body of work on the economic effects of enfranchisement has shown that enfranchisement affects public spending in ways consistent with models of distributive politics: the newly enfranchised group receives increased amounts of public spending concentrated in the areas that are of highest priority to them. However, neither the individual, nor long-term, effects are well understood. I propose to fill this gap by examining the long-run effects of enfranchisement on the accumulation of individual social capital using a unique set of state-level laws that enfranchised, and then disenfranchised, non-citizen immigrants from 1848-1926. These laws existed in 21 states and territories, and have not, to my knowledge, been used before in the economics literature. Among other things, I will create a linked sample of the children of immigrants from the 1910 to 1940 Censuses. The "naturalization status" variable in the 1910 Census, in conjunction with state of residence, will allow me to construct an indicator for whether, as of 1910, a child's father was enfranchised (according to the alien voting laws in place at the time). I will then compare children's outcomes in 1940 between immigrant fathers who had been allowed to vote, and fathers who had not been allowed to vote, in order to assess whether parental enfranchisement increases the social assimilation of second-generation immigrants, as measured by the American-ness of any children that are born, as well as by marriage between ethnic groups.	Henderson	Bailey
In 1892, the United States federal government launched the Alaska Reindeer Service, introducing domesticated reindeer to the Seward Peninsula to alleviate the deprivation and starvation of Inupiat communities. Reindeer herding was intended to provide a “dependable source of cash income and employment” in rural Alaskan villages through the sale of hides, meat, and antler velvet. Administered by the Bureau of Education, the program relied on Saami herders to instruct native apprentices on proper herding techniques. In this project, we examine the impact of the introduction of domesticated reindeer on economic outcomes of Inupiat households tracking those households from the inception of the project to 1940. Using a newly-transcribed dataset on a cross section of Native Alaskan households in the late 1930s, preliminary results find no statistically significant difference between households in communities with reindeer and those in communities without reindeer. However, evidence suggests that, within communities, owning reindeer contributed to greater asset holdings and less debt. Refined estimates will rely on linking native households to U.S. Census decadal data from 1900 to 1940. The results of this program may provide insight into the efficacy of programs providing low-income households with livestock as a consistent source of food and income.	Massey
The role of land reform in ameliorating racial inequality has long been debated. However, empirical analyses have been rare. This paper identifies an exogenous variation in post-Civil War policy to examine the effect of land distribution on racial inequality. The Cherokee Nation, located in what is now the northeastern corner of Oklahoma, permitted the ownership of slaves and joined the Confederacy in 1861. During postwar negotiations, the United States insisted that the Cherokee Nation provide free land for its former slaves. Using a newly constructed data set that follows former slaves in the Cherokee Nation from 1880 to 1900, I examine the impact of land access on racial inequality. I find evidence that blacks in the Cherokee Nation prospered in the decades following emancipation and had higher levels of wealth and income than blacks in the southern United States. Racial inequality also was strikingly lower in the Cherokee Nation in both 1880 and 1900. Land and the associated increase in incomes may have facilitated investment in both physical and human capital. In 1880, blacks in the Cherokee Nation reported higher numbers of fruit trees on their farms. By 1900, black adults in the Cherokee Nation had higher literacy rates than those in the South. Their children also had higher literacy rates and were more likely to attend school. These results suggest that land reform could have played an important role in reducing inequality.	Miller
This research project will examine social and demographic life among American Indians in the United States. The time period of our project extends from the beginning of census enumeration of Native Americans through the 1940 census. The project will examine several aspects of American Indian life including: school attendance and educational attainment, literacy (ability to read and write), ability to speak English, naming patterns, intermarriage, mixed ancestry, marital status, childbearing, and occupation. The project will document these aspects of Indian life at the beginning of our study and then document subsequent social change through 1940. We will examine how these levels and trends vary across different Indian groups and across different regions and states of the country. The documentation of such levels and trends will permit us to see how social changes among Indians are related to events and trends in the larger American society. The project will also construct and estimate statistical models of the influence of such factors as the Indian tribe, region, and state on the social changes observed. Also of great interest is the extent to which certain attributes of Indian life are influenced by other attributes of Indian life. For example, we will examine how marriage, childbearing, and school attendance are influenced by individual and familial attributes such as occupation, naming patterns, intermarriage, and mixed ancestry with Euro-Americans. Our investigations will proceed through the construction of univariate, bivariate, and multivariate tables and through the construction and estimation of multivariate regression equations.	Thornton
Two migration episodes, involving millions of blacks from the South and whites from the Great Plains, dramatically affected U.S. labor markets in the twentieth century. We use detailed geographic data on over one million migrants and a novel non-parametric estimation strategy to assess the importance of social interactions in long-run location decisions. We find very strong social interactions among blacks and much smaller interactions among whites. Black interactions mirror vertical migration patterns and were stronger in destinations with more manufacturing employment, suggesting the importance of access to information, low mobility costs, and demand conditions in facilitating social influence.	Stuart
In 1900 the Pennsylvania Railroad, which was the largest private employer in the U.S., imposed mandatory retirement on its employees at age 70. Workers could retire as early as age 65 if they were disabled or had the approval of supervisors. The railroad employed a very diverse workforce, and even managers and executives were covered by the retirement program. We examine two questions. First, what factors affected the decision to retire early? Early retirement responded to economic incentives in both expected and unexpected ways. Second, how was mortality after retirement affected by pre-retirement earnings and occupations? We consider the effects of income and occupational characteristics on longevity.	Alter
In the early 20th century, Chicago philanthropist Julius Rosenwald established the Rosenwald Fund, a school construction program aimed at improving educational opportunities of black children in the rural South. Between 1913 and 1932, over 5,000 schools were built across 15 Southern states, making it the largest educational initiative of its kind at the time. This project will assess the impact of Rosenwald schools on long-run labor market outcomes of blacks and whites, including measures of intergenerational mobility, using a new longitudinal dataset of birth records linked to Census records. I will explore this question in the context of North Carolina, the largest recipient of Rosenwald schools among all Southern states. This project will also use a new dataset, which I am currently involved in developing at the Population Studies Center as part of our postdoctoral fellowships: The Longitudinal, Intergenerational Family Electronic Micro-Database Project (LIFE-M). LIFE-M links vital records (birth, marriage and death certificates) and Census records to reconstruct families and link individuals over time using machine learning methods. This novel dataset will allow us to study individuals who were of school-going age during the Rosenwald era and which I can observe in the 1940 Census as (young) adults. Thanks to father-son linkages via birth certificate information (which contain parent names), I will be able to compare outcomes of sons in 1940 to that of their fathers in 1900, 1910, 1920 or 1940 and study intergenerational mobility.	Mohammed
The broad premise of access to university education as an engine of mobility rather than a birthright goes back to Jefferson’s founding vision for the University of Virginia (even as the early implementation fell notably short in the dimensions of race and gender). As Jefferson wrote to John Adams in 1813, his plan for public education in Virginia permitted for the University to attract students such that “Worth and genius would thus have been sought out from every condition of life, and completely prepared by education for defeating the competition of wealth and birth for public trusts.” But, within the legacy of the 19th century institution, there is little evidence to record the extent to which young men from different economic circumstances and different parts of the state were attracted to the University. Digitization of records from the University of Virginia (UVA), release of full Census enumerations (with names) from the 19th century, and advances in data processing provide a new opportunity to connect family background, university outcomes, and later life outcomes. This research aims to use 50 years (1825-1875) of individual-level UVa records in combination with Census data from the Minnesota Population Center, which has digitized complete-count databases of Census microdata. We will employ the methods of Abramitzky, Mill, and Perez (2019) to merge the UVA records with Census data.	Turner
Recent literature has considered the evolution of absolute and relative intergenerational mobility in past decades. While relative mobility has been found to be fairly stable over the past decades, absolute mobility has been shown to have declined in the past century. Likewise, marriage and fertility patterns have also evolved over time. In particular, since the 1960s, positive assortative matching (PAM) in the marriage market has increased while the average number of children born per couple has declined. In this paper, we estimate the contribution of marriage and fertility choice to intergenerational mobility. To do so, we construct a model of dynamic marriage matching and endogenize fertility choice. The model can be identified nonparametrically and is able to generate a relationship between marriage, fertility choice and both intra and inter-generational income inequality. We then use historical data from the US from 1850 to 1940 to check if the predictions of the model are valid. Specifically, we examine if the model is able to replicate the empirical trends intergenerational mobility over this time period. In addition, we conduct counterfactual exercises to measure the role of fertility and marriage patterns in inter and intra generational inequality.	Tan
Between 1907 and 1937, thirty-two U.S. states passed eugenic sterilization laws that authorized state institutions for the feebleminded and mentally ill to sterilize residents who were determined to be unfit for reproduction. Eugenic efforts to improve the reproductive fitness of the population were closely aligned with public health, and over 60,000 sterilizations were officially recorded nationwide. Our previous research in California (which led the nation in sterilization practice with approximately 20,000 sterilizations) found that women and Latinos were sterilized at higher rates than other groups. We are now using individual-level data on over 30,000 sterilizations that were recorded in 4 U.S. states: California, Michigan, Iowa and North Carolina. We propose to analyze sterilization data in conjunction with census microdata on institution populations (1920-1940) to formally compare rates of sterilization across demographic groups. Of particular interest is racial, ethnic and gender bias in sterilization practice. We will identify individuals with Spanish surnames to compare rates of sterilization between residents of Mexican origin and non-Hispanic residents. Access to census microdata will also allow us to compare rates of institutionalization across demographic groups.	Stern
This research project empirically investigates the long-term impacts of reparations provided by the United States government on the economic outcomes of Eastern Cherokees. In the 1830s, Eastern Cherokees were forced to migrate from their traditional lands in the American Southeast to a newly designated reserve in the West. Relocated Cherokees suffered a great deal as a result of this policy; thousands of Cherokees perished along the way. In 1905, the U.S. Court of Claims ruled in favor of the Eastern Cherokee Tribe’s claim against the government, resulting in an appropriation of over $1 million to the Tribe’s eligible individuals and families. Particularly, anyone who could prove him- or herself as a member or descendant of a person who had been included in the forced removal were entitled to share equally in the funds. Over 40,000 applications were filed, representing over 90,000 individuals. The court approved more than 30,000 individuals to share in the funds, each of whom received $133 in 1910 (approximately $2,700 in 2019 dollars). We hypothesize that these reparations could have made meaningful impacts on the lives of recipients and their families. Primary outcomes of interest include the number of children, human capital investment in children, employment, and earnings. Should we find these positive effects among eligible Cherokees and their family members, this would suggest that reparations or cash transfers could be an effective policy tool at improving the wellbeing of historically disenfranchised groups. This evidence would also help to inform the current active policy debate around reparations for the descendants of slaves.	Adhvaryu
This project seeks to shed light on how the type and availability of educational institutions impacts individual outcomes. Historically, there has been much heterogeneity in the amount of public education provided across the United States, both varying across states as well as within states. This project will investigate the impact of the resulting differences in educational attainment on skills acquired as measured through labor market outcomes and test scores.	Stephens

For further information, please contact Historical.Census@umich.edu. Access to the full count sample historical U.S. Census data is supported by funding from the MiCDA enclave, the Michigan Node of the NSF Census Research Network, and the Michigan Research Data Center.