RURAL AGRICULTURAL WORKFORCE BY COUNTY, 1800 TO 1900
Description of Variables and Summary of Estimation Methods
used to construct estimates of the rural agricultural workforce
by sex and age-group, by county at census dates from 1800 to 1900.
The data are contained in a set of files, one for each census year.
For each year, the file name is “yearCoRrAg.”
In addition, the files contain the rural population by sex and age-group
and the urban population by county.
* * * * * * * * * * * *Lee A. Craig
North Carolina State University and
National Bureau of Economic Research
University of Kansas andNational Bureau of Economic Research
* * * * * * * * * * * *
The preparation of these data sets was accomplished with the financial support of the National Science Foundation (grants nos. 92-08956 and
94-08525). They are available for use by other researchers without charge. In order to evaluate the impact of our project, and ultimately of NSF support, we need to know who uses which data sets. In return for their use, we request the following:
- Any use of the data in published reports or working papers should appropriately acknowledge the source and cite the source in any tables. The suggested citation is Lee Craig and Thomas Weiss (1998), “Rural Agricultural Workforce by County, 1800 to 1900,” University of Kansas.
- Please inform us of any errors or ambiguities uncovered in the process of working with the data. Only if we receive such feedback can we purge the data files of errors and update them on timely basis.
- We would appreciate receiving copies of all published papers, research memorandums, term papers, working papers, and submitted manuscripts that use or refer to these data. This will enable us to track usage and to inform others of your work. It will also enable us to alert you to any errors detected or any changes made to the data. We can also inform you of the work of others on related topics.
- Please do not pass the data sets on to others without notifying us. We would, of course, be very happy to supply other users with the data directly, and free of charge.
We have made every effort to check the accuracy of our data entry and documentation, but we cannot ensure that our product is error-free. The files have been prepared in Excel format, and those should be the most reliable source. If you use an ascii version or a version in some other format errors may appear in the process of conversion.
This project was sponsored in part by the National Science Foundation and we gratefully acknowledge that support. We would also like to thank EH.Net, housed at Miami University, Oxford, Ohio, for technical assistance and for providing disk space in its server.
Comments and questions may be sent to:
|Lee CraigDepartment of Economics310D Hillsborough BldgNorth Carolina State U.
Raleigh, NC 27695-8110
Department of Economics
University of Kansas
Lawrence, KS 66045
The Census did not provide workforce statistics by county and so these estimates of the rural agricultural workforce were constructed using information generated in earlier work carried out at the state level. Even at the state level the Census provided only limited detail regarding the age and sex of workers in some years and none at all in others. In order to have consistent coverage by state for the entire time period 1800 to 1900, the estimates were confined to only five population groups. Given the lack of detail at the state level in some years, the estimates did not distinguish between free white and nonwhite workers, and the estimates for slaves were confined to those aged 10 and over.
The county level data sets show separate estimates for male and female slaves that were constructed by assuming the same participation of male and female slaves. The county files also contain other data used to generate the rural farm workforce estimates. These are the rural population and the rural agricultural participation rate for the specified age-sex groups, the rural share of the free and of the slave population, and the urban population.
The organization of the files containing the estimates of the rural agricultural workforce by county for years 1800 through 1860 differs from that for the years 1870 through 1900. For the first set of years the file for each year is organized as shown below in the first list of variables; for the latter set of years the files are organized according to the second list of variables.
The number of observations (i.e. counties) in each year is as follows
1800 419 1810 572 1820 758 1830 985 1840 1275 1850 1620 1860 2077 1870 2289 1880 2568 1890 2776 1900 2849
I. Order of Variables in Files for 1800 through 1860
Col Name Definition 1 STATE State in which county is located, identified by ICPSR code no. 2 CNTY County, identified by ICPSR code number 3 RrAgLFM15 Estimated number of Rural Agricultural Workers, Males aged 10-15 4 RrAgLFM16 Estimated Number of Rural Agricultural Workers, Males aged 16+ 5 RrAgLFF15 Estimated Number of Rural Agricultural Workers, Females aged 10-15 6 RrAgLFF16 Estimated Number of Rural Agricultural Workers , Females aged 16+ 7 RrAgLFMS10 Estimated Number of Rural Agricultural Workers, Male Slaves aged 10+ 8 RrAgLFFS10 Estimated Number of Rural Agricultural Workers, Female Slaves aged 10+ 9 RrAgPopM15 Estimate of the Rural Population of Males aged 10-15 10 RrAgPopM16 Estimate of the Rural Population of Males aged 16+ 11 RrAgPopF15 Estimate of the Rural Population of Females aged 10-15 12 RrAgPopF16 Estimate of the Rural Population of Females aged 16+ 13 RrAgPpMS10 Estimate of the Rural Population of Male Slaves aged10+ 14 RrAgPpFS10 Estimate of the Rural Population of Female Slaves aged10+ 15 RrAgPrM15 Agricultural Participation Rate for Rural Males aged 10-15 16 RrAgPrM16 Agricultural Participation Rate for Rural Males aged 16+ 17 RrAgPrF15 Agricultural Participation Rate for Rural Females aged 10-15 18 RrAgPrF16 Agricultural Participation Rate for Rural Females aged 16+ 19 RrAgPrS10 Agricultural Participation Rate for Rural Slaves aged 10+ 20 RrlShFree Rural Share of the Free Population 21 RrlShSlave Rural Share of the Slave Population 22 RrlFree Rural Free population of all ages 23 RrlSlaves Rural Slave Population of all ages 24 TotFree Total Free Population of all ages 25 TotSlave Total Slave Population of all ages 26 TotPop Total Population of all ages 27 UrbanPop Urban Population of all ages 28 RGN1 Region code
II. Order of Variables in Files for 1870, 1880, 1890 and 1900
Col Name Definition 1 STATE State in which county is located, identified by ICPSR code no. 2 CNTY County, identified by ICPSR code number 3 RrAgLFM15 Estimated number of Rural Agricultural Workers, Males aged 10-15 4 RrAgLFM16 Estimated Number of Rural Agricultural Workers, Males aged 16+ 5 RrAgLFF15 Estimated Number of Rural Agricultural Workers, Females aged 10-15 6 RrAgLFF16 Estimated Number of Rural Agricultural Workers , Females aged 16+ 7 RrAgPopM15 Estimate of the Rural Population of Males aged 10-15 8 RrAgPopM16 Estimate of the Rural Population of Males aged 16+ 9 RrAgPopF15 Estimate of the Rural Population of Females aged 10-15 10 RrAgPopF16 Estimate of the Rural Population of Females aged 16+ 11 RrAgPrM15 Agricultural Participation Rate for Rural Males aged 10-15 12 RrAgPrM16 Agricultural Participation Rate for Rural Males aged 16+ 13 RrAgPrF15 Agricultural Participation Rate for Rural Females aged 10-15 14 RrAgPrF16 Agricultural Participation Rate for Rural Females aged 16+ 15 RrlShPop Rural Share of the Population 16 TotPop Total Population of all ages 17 UrbanPop Urban Population of all ages 18 RGN1 Region code
III. Summary of the Methods of Estimation
Estimates of the number of farm workers by age-sex category by county are based on Weiss’s recent reworking of the census gainful worker data for the nineteenth century and our construction of the rural population by county. Weiss’s work was carried out at the state level; the county figures are an extension of those estimates. The state-level estimates were used to calculate a ratio of agricultural workers to population for each of the specified population groups. The county-level gainful worker figures were obtained by multiplying those ratios by the appropriate population count in each county. A summary of the procedures for constructing the state and county level estimates follows. Additional details of the estimation are described in Weiss (1992 and forthcoming) and Craig and Weiss (1996).
First, the total workforce was estimated following the procedures laid out by Lebergott (1966). For the years 1870 to 1900, the figures were taken directly from the census, although corrections were made to the statistics for 1870 and 1890 (See Weiss, 1985). For the years 1800 to 1860, the total is the sum of estimates for each of five demographic groups; free males 16 and over, free females 16 and over, free males 10 to 15, free females 10-15, and slaves aged 10 and over. The estimate for each group is the product of the population in that group times a group specific participation rate.
The procedures for estimating the agricultural workforce can be separated into those used for the postbellum years and those for the antebellum period. For the years 1870 through 1900, the agricultural gainful worker figures were taken from the census, although the census figures, which reported numbers of workers by occupation, had to be reorganized in order to obtain the industry’s workforce. It was a fairly straightforward task to assign most occupations to their respective industries, especially so for agriculture. There were, however, some occupations that were found in more than one industry so procedures were used to allocate workers in such occupations to the various industries in which they might have been employed. The only occupation of this sort that impinged on agriculture was that of “laborers, not otherwise specified” (or not elsewhere classified). The allocation between agriculture and nonagriculture was based on the relationship between urban population and nonfarm occupations, including nonfarm laborers.1 In each census year between 1850 and 1900, Weiss distributed the reported number of laborers between agriculture and nonagriculture according to the 1910 proportions adjusted for changes in urbanization.2
For 1800 to 1860, the agricultural workforce is the sum of the free and slave farm workforces, both covering those aged 10 and over. For 1850 and 1860 these figures were obtained by summing up estimates for each age-sex component. The census provided figures for the largest group of workers, males aged 16 and over, in both years, as well as for females aged 16 and over in 1850. The other categories of free workers were estimated using evidence from the postbellum census and samples of rural households in 1860. The slave figures were obtained by allocating an estimated percent of the rural slave population aged 10 and over to farming. The allocation percentages were obtained from regression equations fitted to the county data for 1820 and 1840.
For 1820 and 1840, the census figures, as revised by Weiss, provided the total farm workforce aged 10 and over. Those figures were then distributed across the various population groups using the distributional evidence from later years and the estimates of the number of slaves engaged in farming derived from the regression equations. The procedures used to estimate the farm workforce figures for 1800, 1810 and 1830 were more complex and roundabout, and differed substantially between the free and slave states. Nevertheless, the estimates were based largely on the evidence from later years (See Weiss, 1992).3
In each of the years from 1800 to 1860 the number of urban agricultural workers was estimated and deducted from the revised census totals in order to obtain a rural count.4 It was assumed that all urban agricultural workers were males aged 16 and over.5
The preceding work yielded, either implicitly or explicitly, estimated ratios of rural agricultural workers to rural population for each age-sex component of the population in each state. These figures were then used to estimate the number of rural agricultural gainful workers in each county, it being assumed that each state’s ratio held for the rural population in each county in the state. No doubt there was further variation in these participation rates within states, that is across counties. Unfortunately, the available data do not permit us to capture all that detail.
Rural Population by County
In order to estimate the rural farm workforce we had to compile a series on the rural population by county and estimate the age-sex distribution of that population. We were able to construct a series on the rural population by county by using data on the population of cities and towns available on U.S. Census worksheets. Those data had been used to compile the urban and rural population series shown in Historical Statistics.6 We were able to reconstruct that series with a high degree of accuracy.7 We also were able to identify the county in which each city and town was located so that by subtracting the urban count from the county’s total population we obtained the rural population as the residual.
Those worksheet figures did not provide any details about the age-sex composition of the urban population. For the years 1800 to 1860, much of the needed detail, but not all, was obtained from the published census. The extent to which the distribution of the free population and slave population had to be constructed varied from year to year. At the state level, no estimation was required in 1850 and 1860. The reported data for 1830 and 1840 provided figures for the slave population aged 10 and over, but the figures for free blacks had to be reorganized. The 1820 census data required reorganization of the age groupings for both free blacks and slaves. For 1800 and 1810 it was first necessary to estimate the sex breakdown, and then the age composition. For the postbellum period, the age-sex distribution of the rural population was assumed to be the same as that for each county’s total population.8
IV. Codes Used
The state and county codes used in these files are the ICPSR codes as modified by Haines (1997). He has given the District of Columbia a state code of 98. Because DC had virtually no rural population or agricultural workforce it has been omitted from the Craig-Weiss “CoRrAg” files in the years 1820 to 1900. In 1800 and 1810, the District is included with the state of Virginia with the county code of 9800.
The county codes can be obtained from Michael Haines (firstname.lastname@example.org). The state and regional codes are as follows.
Regional Codes 1 New England 2 Middle Atlantic 3 East North Central 4 West North Central 5 South Atlantic 6 East South Central 7 West South Central 8 Mountain 9 Pacific State Codes Alphabetical Order Ordered by Code 41 ALABAMA 1 CONNECTICUT 61 ARIZONA 2 MAINE 42 ARKANSAS 3 MASSACHUSETTS 71 CALIFORNIA 4 NEW HAMPSHIRE 62 COLORADO 5 RHODE ISLAND 1 CONNECTICUT 6 VERMONT 11 DELAWARE 11 DELAWARE 98 DIST COLUMBIA 12 NEW JERSEY 43 FLORIDA 13 NEW YORK 44 GEORGIA 14 PENNSYLVANIA 63 IDAHO 21 ILLINOIS 21 ILLINOIS 22 INDIANA 22 INDIANA 23 MICHIGAN 31 IOWA 24 OHIO 32 KANSAS 25 WISCONSIN 51 KENTUCKY 31 IOWA 45 LOUISIANA 32 KANSAS 2 MAINE 33 MINNESOTA 52 MARYLAND 34 MISSOURI 3 MASSACHUSETTS 35 NEBRASKA 23 MICHIGAN 36 NORTH DAKOTA 33 MINNESOTA 37 SOUTH DAKOTA 46 MISSISSIPPI 40 VIRGINIA 34 MISSOURI 41 ALABAMA 64 MONTANA 42 ARKANSAS 35 NEBRASKA 43 FLORIDA 65 NEVADA 44 GEORGIA 4 NEW HAMPSHIRE 45 LOUISIANA 12 NEW JERSEY 46 MISSISSIPPI 66 NEW MEXICO 47 NORTH CAROLINA 13 NEW YORK 48 SOUTH CAROLINA 47 NORTH CAROLINA 49 TEXAS 36 NORTH DAKOTA 51 KENTUCKY 24 OHIO 52 MARYLAND 53 OKLAHOMA AND IND TERR. 53 OKLAHOMA AND IND TERR. 72 OREGON 54 TENNESSEE 14 PENNSYLVANIA 56 WEST VIRGINIA 5 RHODE ISLAND 61 ARIZONA 48 SOUTH CAROLINA 62 COLORADO 37 SOUTH DAKOTA 63 IDAHO 54 TENNESSEE 64 MONTANA 49 TEXAS 65 NEVADA 67 UTAH 66 NEW MEXICO 6 VERMONT 67 UTAH 40 VIRGINIA 68 WYOMING 73 WASHINGTON 71 CALIFORNIA 56 WEST VIRGINIA 72 OREGON 25 WISCONSIN 73 WASHINGTON 68 WYOMING 98 DIST COLUMBIA
Craig, Lee and Weiss, Thomas. 1996. “The Nineteenth Century Farm Labor Force and Rural Population: County-Level Estimates and Implications.” (mimeo) presented at the Social Science History Association annual meeting.
Geib-Gunderson, Lisa, and Zahrt, Elizabeth. 1996. “A New Look at U.S. Agricultural Productivity Growth, 1800–1910,” Journal of Economic History, 56 (Sept.) 679-86.
Lebergott, Stanley. 1966. “Labor Force and Employment, 1800-1960,” in Studies in Income and Wealth, 24, 117-204
U.S. Bureau of the Census, 1910. Census of Population vol. IV, Occupations.
U.S. Department of Commerce, 1975. Historical Statistics,
Weiss, Thomas,1985. “Adjustments to the Census Counts of Population and Labor Force, 1870, 1890 and 1910″ (mimeo) University of Kansas.
Weiss, Thomas. 1986. “Assessment and Revision of the Antebellum Census Labor Force Statistics: 1850 and 1860″ (mimeo) University of Kansas.
Weiss, Thomas. 1987. “Demographic Aspects of the Urban Population, 1800 to 1840″ in Quantity and Quidity: Essays in Honor of Stanley Lebergott. Middletown, Conn.: Wesleyan University Press.
Weiss, Thomas. 1987. “Assessment and Revision of the Antebellum Census Labor Force Statistics: 1840″ (mimeo) University of Kansas.
Weiss, Thomas. 1988. “Assessment and Revision of the Antebellum Census Labor Force Statistics: 1820″ (mimeo) University of Kansas.
Weiss, Thomas. 1989. “Estimation of the Farm/Nonfarm Distribution of Laborers, not otherwise specified; by State, 1850 to 1900″ (mimeo) University of Kansas.
Weiss, Thomas. 1992. “U.S. Labor Force Estimates and Economic Growth, 1800 to 1860,” in John Wallis and Robert Gallman, eds., The Standard of Living in the United States Before 1860, (Chicago, IL).
Weiss, Thomas. 1996. “Refined Estimates of the Agricultural Labor Force, 1850 and 1860,” (mimeo) University of Kansas.
Weiss, Thomas. 1998. “Estimates of White and Nonwhite Gainful workers in the United States by Age Group, Race and Sex; at Census Dates from 1800 to 1900.” (mimeo)University of Kansas.
1 The Census of 1910 reported the number of unspecified laborers according to their industry of employment, as well as by their residence in cities above 25,000. U.S. Bureau of the Census, 1910, various tables.
2 Earlier researchers, such as Edwards, Lebergott, and Miller and Brainerd, had also dealt with this problem of allocating laborers among industries, albeit only at the national level. See Weiss (1989) for a summary. More recently, Geib-Gunderson and Zahrt (1996) have estimated the farm-nonfarm distribution for 1880 and 1900 using logit regression techniques. Their distributions differ noticeably from those of Weiss, although their work confirms the importance of urbanization as a determinant of the number of laborers working in nonfarm industries. Unfortunately, their estimates are not available at the state level, and are confined to only two census years, so we could not gauge the impact of using their estimates rather than Weiss’s.
3 The figures shown in the county files differ slightly from the earlier estimates described in Weiss (1992). Those earlier figures were refined based on additional evidence available on the participation of females and youths in agriculture in the years 1870 through 1900. The refinements do not change much the total number of workers aged 10 to 15 in agriculture, but do apportion them differently between male and female.
4 The number of urban agricultural workers was estimated as the product of the urban population of all ages times an estimated ratio of workers to population of all ages. The estimating ratios could be calculated directly from the census data for 1820, 1840 and 1870, and the ratios for other years (1830, 1850, and 1860) were obtained by interpolation or by extrapolation (1800 and 1810).
5 This assumption in turn means that the estimated numbers of adult female agricultural workers and those aged 10 to 15 were all living in rural areas.
6 The “urban population comprised all persons living in incorporated places of 2,500 or more and areas (usually minor civil divisions) classified as urban under special rules relating to population size and density”. See U.S. Department of Commerce, 1975, p. 2 and series A:57-72.
7 The data we assembled for the 19th century have been combined with series assembled by Michael Haines, and a combined set of data covering 1800 to 1990 are available from Haines.
8 For details see Weiss 1987 and forthcoming.