The Distressed Communities Index (DCI) combines seven complementary economic indicators into a single holistic and comparative measure of community well-being. The index is constructed from the U.S. Census Bureau’s American Community Survey 5-Year Estimates and Business Patterns datasets.
The seven component metrics of the DCI are:
- No high school diploma: Percent of the 25+ population without a high school diploma or equivalent
- Housing vacancy rate: Percent of habitable housing that is unoccupied, excluding properties that are for seasonal, recreational, or occasional use
- Adults not working: Percent of the prime-age population (25-64) not currently in work
- Poverty rate: Percent of the population living under the poverty line
- Median income ratio: Median household income as a percent of the state’s median household income (to account for cost of living differences across states)
- Change in employment: Percent change in the number of jobs
- Change in establishments: Percent change in the number of business establishments
Each component is weighted equally in the index, which itself is calculated by ranking communities on each of the seven metrics, taking the average of those ranks, and then normalizing the average to be equivalent to a percentile. Distress scores range from approaching zero to 100.0, such that the zip code with the average rank of 12,500 out of 25,000 will register a distress score of 50.0. Communities are then grouped into quintiles, or fifths. The best-performing quintile is considered “prosperous,” the second-best “comfortable,” the third “mid-tier,” the fourth “at risk,” and the fifth, or worst-performing, “distressed.”
At its heart the DCI is a local measure, and the zip code-level analysis produces many of its most powerful insights. However, calculating the DCI at other geographic scales can help shed light on the many different dynamics operating at different spatial scales that influence how individual Americans experience the economy. Therefore, we also calculate the DCI separately at the city, county, and congressional district levels. In all, the DCI captures 99 percent of the U.S. population and covers all 25,800+ zip codes and 3,100+ counties with over 500 people, as well as the over 800 U.S. cities that contain at least 50,000 people.
This year’s DCI presents results for two time periods reflecting the composition and availability of the Census Bureau’s ACS 5-Year Estimates: 2007-2011 and 2012-2016. The 5-year estimates are constructed from five years’ worth of data collection, reflecting the length of time required to compile enough data to provide meaningful and accurate estimates at fine geographic scales (such as zip codes). The 2007-2011 5-year estimates represent the first installment available at the zip code level, and the 2012-2016 5-year estimates are the most recent. Business Patterns data are available annually, and we have naturally chosen to include and analyze the years that match the ends of each ACS window.
This year’s materials, report, and data interactive present the results from the DCI calculated for each time period, with the discussion defaulting to the more recent period where not specified. The two universes of zip codes are not completely identical, as just over 500 zip codes appear in the latter dataset but not the former and just over 700 zip codes the former but not the latter. These zip codes were included in the DCI calculated for their respective periods but excluded from any section of the analysis that directly compares zip codes over time. The discrepancies in coverage can be attributed to the rare formation of new zip codes and to zip codes falling below the 500-person threshold necessary to be included in either period (the threshold excludes institutionalized residents, students, and active duty personnel in order to prevent the results from being artificially biased by prisons, colleges and universities, and military bases).
In instances where employment estimates from Business Patterns were suppressed by the Census Bureau in order to preserve privacy, the DCI defaulted to the next-highest level geography to produce a growth estimate that could enter into the index. This affected nearly 4,000 zip codes in the 2012-2016 DCI and 36 counties. For example, the statewide job growth rate of 4.7 percent entered into the index as the job growth rate for the three North Dakota counties for which employment figures were suppressed.
Zip codes are characterized as rural, suburban, or urban based on their population density. Cutoffs were determined so that roughly one-third of the country’s population would reside in each category. As a result, the population density of rural zip codes ranges from 0-150 people per km-squared; suburban zip codes from 150-1,085 people per km-squared; and urban zip codes from there on up.
All of the auxiliary data presented in the DCI suite of materials come from the same sources as the components of the DCI themselves: the U.S. Census Bureau’s ACS 5-year estimates and Business Patterns datasets.
Zip codes as presented and discussed here should be considered approximations of geographies and communities. Zip codes represent postal routes defined by the U.S. Postal Service (USPS), not the U.S. Census Bureau, and their boundaries can and do change frequently. The U.S. Census Bureau builds its own proprietary approximations of zip codes called Zip Code Tabulation Areas (ZCTAs) from census blocks once after each Decennial Census. The DCI includes data tabulated by both zip code and ZCTA: Two of the underlying variables (those from Business Patterns) are defined by zip code and five (those from the American Community Survey) are defined by ZCTA. Since ZCTAs are static over each decade but zip codes may change, it is important to interpret the findings of the DCI as general trends for an approximate area rather than discrete developments within a clearly-defined set of lines. In addition, Business Patterns data are subject to errors that the Census Bureau does not go back to correct. Both boundary changes and these errors may affect change over time calculations.
Zip code-level estimates were used to create city- and congressional district-level estimates for change in employment and change in establishments. Zip code values were aggregated up to the city level using ZCTA-to-place relationship files provided by the Census Bureau and at the congressional district level using the congressional district mapping file for the 115th Congress provided by the U.S. Department of Housing and Urban Development (HUD). In instances where zip code boundaries cut across city or congressional district lines, zip code portions were attributed to cities and congressional districts according to the share of the associated ZCTA’s population falling within the boundaries of the larger geography using relationship files provided by the Missouri Census Data Center and the mapping file by HUD. Relationship and mapping files were also used to report the share of a city’s, county’s, state’s, or congressional district’s population residing in distressed and prosperous zip codes.