Stay Connected

Follow ERS on Twitter
Subscribe to RSS feeds
Subscribe to ERS e-Newsletters.aspx
Listen to ERS podcasts
Read ERS blogs at USDA

Documentation

Nonmetro-Based Thresholds

Although ERS coded the typologies for all U.S. counties, the thresholds for determining the economic dependence types were set using nonmetro counties only. Most thresholds were roughly set at the nonmetro mean plus one standard deviation. ERS used counties that met the 2013 definition of nonmetro (micropolitan and noncore combined) in analyzing the means. The codes are primarily meant to be useful in the analysis of rural conditions, trends, and program needs. ERS coded metro counties to facilitate comparisons across the country. If researchers are solely interested in metro conditions, they should carefully test whether these classifications are meaningful in that context.

Methods for Determining the Economic Dependence Types

Labor and proprietors' earnings and employment by place of work are the basis for the economic dependence categories. Each industry's earnings and employment were calculated separately as a percent of total labor and proprietors' earnings or total employment in the county in 2010, 2011, and 2012. These percentages were summed, and divided by 3 to obtain annual average percentages. This averaging was done to minimize the effects of any one-year anomaly in an industry's earnings or employment.

County-level estimates of earnings and employment by place of work used to measure economic dependence come from the Bureau of Economic Analysis' (BEA) Regional Local Area Personal Income & Employment data. The BEA income and employment data used were released in November 2014. The BEA industry data use the North American Industry Classification System (NAICS). Publicly available data (with some industry suppression at the county level) were used for Florida, Massachusetts, Mississippi, New Hampshire, and Wyoming. For all other States, unsuppressed data, not publicly available, were used to develop the economic classifications.

Farming dependence was based on two thresholds—farm earnings accounting for an annual average of 25 percent or more of total county earnings or farm employment accounting for 16 percent or more of total employment during 2010-12.

Mining dependence was based on the mining industry accounting for an annual average of 13 percent or more of total county earnings or 8 percent or more of total county employment during 2010-12. Mining includes establishments that extract naturally occurring mineral solids, such as coal and ores; liquid minerals, such as crude petroleum; and gases, such as natural gas. The term mining is used in the broad sense to include quarrying, well operations, and other preparation customarily performed at the mine site, or as a part of mining activity. Also included are mining support activities.

Manufacturing dependence was based on the manufacturing industry accounting for an annual average of 23 percent or more of total earnings or 16 percent or more of total employment during 2010-12.

Federal/State government dependence was also based on Federal and State governments accounting for an annual average of 14 percent or more of total county earnings or 9 percent or more of total employment during 2010-12.

Recreation counties were computed using three data sources:

  1. Percentage of wage and salary employment in entertainment and recreation, accommodations, eating and drinking places, and real estate as a percentage of all employment reported by the Bureau of Economic Analysis;
  2. Percentage of total personal income reported for these same categories by the Bureau of Economic Analysis; and
  3. Percentage of vacant housing units intended for seasonal or occasional use reported in the 2010 Census

The three variables measuring employment, earnings, and seasonal housing were converted to z-scores and combined into a weighted index (weights of 0.3 were assigned to income and employment and 0.4 to seasonal housing) to reflect recreational activity. Counties with index scores of 0.67 or higher were regarded as recreation counties. Seasonal housing was given a higher weight because in some areas employment and income may not reflect recreational activity because of the seasonality. This method is a variation of the original developed by Ken Johnson and Calvin Beale in their nonmetro recreation classification completed in 2002. See Rural America for more information.

Nonspecialized counties were counties that did not qualify for either the farming, mining, manufacturing, Federal/State government, or recreation county types.

If a county qualified for more than one economic type, it was classified in the industry which accounted for the largest percentage of total earnings. For research purposes we have included in the typology codes both the non-overlapping county type variable and the overlapping dummy variables for each of the economic types (available in the download file only).

Methods for Determining the Policy Types

The low-education (2008-12), low-employment (2008-12), persistent poverty (1980, 1990, 2000, 2007-11), persistent child poverty (1980, 1990, 2000, 2007-11), population loss (1990, 2000, and 2010), and retirement destination (2000 and 2010) classifications are all based on census data from the years in parentheses after their names. Previous versions of the policy types were based on Decennial Census data. In the 2015 update, 2010 Decennial Census data were not available for all of the county classifications. The 5-year American Community Survey (ACS) was the only data source available for the low-education and low-employment classifications, and the 2008-12 were the most recent data available at the time this update was done.

For the persistent poverty and persistent child poverty classifications, data from the 2007-11 ACS American Community Survey were used for consistency with previous Censuses. The mid-point year of the 2007-11 ACS is 2009, exactly 10 years from the income reporting years of 1979, 1989, and 1999 in the 1980, 1990, and 2000 Censuses.  

The retirement destination classification was created with data downloaded from the Applied Population Laboratory at the University of Wisconsin-Madison. See the Applied Population Laboratory website for more information. Net migration for the population 60 years old and older in 2010 was calculated using the residual method. Using this method, net migration equals the change in population between 2000 and 2010 after removing the change in population due to deaths. If the net migration of persons 60 years and older added 15 percent or more to the number of persons in that age group in 2010 it was classified as retirement destination county.

County Coverage

All 3,143 counties, boroughs, independent cities, parishes, and other county-equivalents that were enumerated in the 2010 Census of Population and Housing were classified. Data for other U.S. territories subdivisions were not available. Details about the independent cities and counties that were analyzed as combined units are available in the data file.

Last updated: Friday, June 03, 2016

For more information contact: Timothy Parker