This appendix discusses the general approach and procedures used to develop the 1989 revised ERS typology. The ERS research team considered different models for revising the ERS typology ranging from doing a complete overhaul to updating the original 1979 typology while making only modest definitional changes. The team close a middle-ground approach. Several initial decisions were made to guide the analysis:
Eleven types were identified for inclusion in the revised typology. Four types were retained from the original typology: farming-dependent, manufacturing-dependent, mining-dependent, and government-dependent. A fifth type, services-dependent counties, was added in response to the growing proliferation of services jobs in the national economy. The sixth type, nonspecialized counties, includes those counties whose economies did not meet the criteria for any of the major economic specializations.
Five additional policy types were identified. Three of the policy types--retirement-destination, Federal lands, and persistent poverty--were revised from the original typology. A commuting type was added to reflect county economies influenced by workers commuting to jobs in other counties, and a transfers-dependent type was added to identify county economies heavily reliant on income from government transfer payments.
Data Sources
Several data sources were used to implement an analysis following the criteria laid out above. Three-year annualized averages of earnings by industry obtained from the Local Area Personal Income Series produced by the Bureau of Economic Analysis (BEA) were used to define the six economic types and one of the policy types (transfers-dependent counties). BEA furnished ERS with unsuppressed estimates for all but 5 of 50 States--Alaska, Maine, Michigan, Ohio, and Wyoming. Public use estimates (with suppression) for these five States were utilized to cover as many of their counties as possible.
County-level data from the Bureau of the Census's dicennial Censuses of Population were employed to define three of the policy types--retirement-destination, commuting, and persistent poverty. The Federal lands type was defined using data from the 1987 Natural Resources Inventory, produced by the Soil Conservation Service, U. S. Department of Agriculture.1
Data to perform the descriptive analysis of the various types were drawn from the data sources noted above and, additionally, from BEA's county employment estimates reported by industry and county unemployment estimates from the Bureau of Labor Statistics (BLS).
Procedures To Define the Types
The analysis to develop definitions for the 11 types and subsequently to classify 2,276 counties designated as nonmetro in 1993 according to these types involved several steps.
Establishing Cutting Points
The first step was to develop measures and establish cutting points for defining the types (excluding the persistent poverty category) from data drawn from the sources specified above (appendix table 1). The exact measures and the years covered are indicated in the definitions provided in the introduction section. Distributions of these measures were evaluated along with means and standard deviations to establish threshhold levels or cutting points. As with earlier typologies, the rule of mean plus one standard deviation was used as a rough guideline (along with the researchers' judgment of the nature and relative importance of the various types) to establish the cutting points shown in appendix table 1. A cutting point for the persistent poverty type was selected by evaluating the use of two measures: poverty rates in excess of 20 percent or 25 percent in each of 4 selected years.
Adjusting for Economic Overlaps
The second step was to classify the counties according to the designated cutting points and implement procedures to establish the nonoverlapping economic types. Counties overlapping on one or more economic types were handled in the following manner. Counties that qualified as farming, mining, or manufacturing and also qualified as government and/or services counties were assigned to the farming, mining, or manufacturing types, respectively. Other overlaps such as farming-mining, mining-manufacturing, or government-services were assigned to the type with the largest percentage point difference above the cutting point divided by the standard deviation.
Following these procedures, we were able to classify 2,259 of the 2,276 nonmetro counties into one of the six economic types (appendix table 2.) As noted above, data were suppressed in some instances for some counties in five States. Data were available to classify most of these suppressed data counties into an economic type. However, 17 such counties that did meet the definitional criteria for an economic type were omitted from the analysis.2
The number of nonmetro counties qualifying for one or more policy types was 1,197. The number of two-way overlaps of the policy types with the economic types and with other policy types appears in appendix table 3.
Appendix tables 4-18 report the detailed economic and socioeconomic statistics used to construct the descriptive profiles. These statistics are reported as either numbers, unweighted county averages, or aggregated values as indicated in the table footnotes. In a few instances, the profiles may contain statistics that do not appear in the tables.
Endnotes
Go to: