Rationalizing Rural Area Classifications for the Economic Research Service: A Workshop Summary (2016)

Appendix A

ERS Goals for Workshop on Rural Classifications

1. The workshop is intended to help ERS make decisions regarding the generation of a county urban-rural scale for public use. This scale need not satisfy every purpose, but it should be generally useful and have face validity. If it could be adopted both for research and policy, that would be ideal, but perhaps that is too much to shoot for.
2. Our current scales classify all counties, with a basic distinction between metropolitan and nonmetropolitan areas. We probably want to keep that distinction (or not?), but this still leaves a number of questions.
3. What is a reasonable number of categories? Too many, and one can get lost; too few, and one is not capturing enough.
4. For metropolitan areas,
   1. Do we want to distinguish the most urban from the less urban counties? The original ERS version distinguished central from outlying counties, for metropolitan areas of 1 million or more residents (1M+). Over time, OMB widened the definition of central, leaving far fewer outlying counties. Now, one can use percent residing in urbanized areas to define central, or percent residing in rural areas to define outlying, or set a density threshold, or use other criteria.
   2. Do we want to use such a distinction only for 1M+ counties? Over ¾ of the U.S. population are in these counties.

5. 3. Do we want to distinguish metro counties that have more than 50% population in rural density settings (as per Isserman).
   4. Do we collapse all counties in metropolitan areas of less than 1M population into one category, or are there important distinctions at other thresholds?
6. For nonmetropolitan areas,
   1. Both of the current county codes classify nonmetro counties along two dimensions, proximity and size. For proximity:
      1. Should we shift from adjacency to estimated driving time to an urbanized area?
      2. Should we only consider proximity to 1M+ metro areas, or include smaller metros?
      3. How many distance/proximity categories are useful?
   2. For size:
      1. How should we incorporate urban into the nonmetro side of the scale?
      2. Should we incorporate micropolitan, as is done in the Urban Influence Codes?
      3. How many urban size categories are useful for nonmetro counties? The Beale Codes have 1 proximity measure and 3 urban size measures. Perhaps it should be 2 and 2, as proximity may have gained salience.
7. Are there methodologies for helping to make these decisions, such as ability to distinguish across a set of socioeconomic characteristics? Past decisions regarding population thresholds and related criteria were made with little or no supporting research, e.g., the 2,500 urban threshold, the 50,000 metro threshold, the 500 people per square mile rural-urban boundary, the 25 percent commuting metro-nonmetro boundary.
8. ERS most likely will continue to define one or more subcounty classifications (tracts, ZIP codes). Most of the same questions apply.
9. Are there ways to tie together the subcounty and county classifications, conceptually and empirically? For instance, the FAR codes introduce a grid-based approach to classifying urban-rural settlement. Does this approach offer a way to unify the different classifications, that is, downcast data to grids, carry out analyses at the grid level, then aggregate results to needed geographic units?