County Distress Index
How the County Distress Index Works
The CDI scores household financial distress across 3,144 U.S. counties from 0 to 100. A county's score is the share of U.S. counties it is more distressed than. The current engine is family-v1: five equal-weighted domains, distress-oriented indicator percentiles, published county ranks, and public score labels.
Current Specification
- Coverage
- 3,144 counties and county equivalents
- Average county score
- 50.0 as of 2026-08-31; the scores average 50 by construction
- Engine
- family-v1
- Score
- Share of the other counties with a lower mean of five equal-weighted domain scores
- County labels
- National rank, state rank, and score label, not national ADI bands
- Latest build
- 2026-09-30
Scoring Rule
Each county indicator is oriented so that a higher value means more household financial distress. The engine then converts each indicator into a Hazen percentile across counties, averages member indicators inside each domain, and averages the five domain scores into the CDI composite.
An average of percentiles bunches toward the middle, so the published score is not the composite. It is the share of the other counties whose composite is lower, (number of counties − rank) ÷ (number of counties − 1) × 100, with tied counties sharing the lower count, rounded down to a whole number. Rounding down keeps "more distressed than X% of U.S. counties" true for every county; a county scores 100 only when it is more distressed than every other county. The order of counties is the composite's order. The composite itself stays in the data as composite_mean.
National rank and state rank sort counties by the composite. Counties share a rank only when their composites are identical, shown as "tied for 6th." County pages print the national rank only for the 100 most and 100 least distressed counties, because ranks in between move between updates; the state rank always prints. Score labels come from the 0-100 score.
Five Domains
| Domain | Weight | Member indicators |
|---|---|---|
| Debt Burden (housing basis) | 20.0% | Rent To Income Ratio, Renters 50pct Burdened |
| Default & Legal | 20.0% | Debt In Collections Pct, Filing Rate Per 100k |
| Delinquency | 20.0% | Auto Loan Delinq Pct, Cc Delinq Pct, Subprime Pct |
| Labor | 20.0% | Prime Age Not Working Pct, Unemployment Rate 12mo Avg |
| Safety Net & Buffer | 20.0% | Child Poverty Rate, Disability Rate, Median Household Income, Poverty Rate, Uninsured Rate |
Ranks and Score Labels
County pages publish national rank, state rank, and a plain-language score label. Because the score is a percentile, each score label covers about a tenth of counties. Rank fifths remain available inside the data for QA, but public county labels should use the score range names below.
What Changed
The retired county engine used fitted factor weights and fixed score buckets. The family-v1 engine uses equal domain weights, national and state ranks, and county score labels without using the national ADI band names.
Composition 4 took effect September 30, 2026. It changed the Labor domain. Labor used to rest on one month of county unemployment, a figure that is not seasonally adjusted, so farm, resort and school-year counties could look distressed in their off-season. Labor now has two inputs of equal weight, a tenth of the score each: the county unemployment rate averaged over the latest 12 months, and the share of adults aged 25 to 54 who are not working, from the U.S. Census Bureau's five-year American Community Survey. The current average covers the monthly rates from September 2025 to August 2026 (U.S. Bureau of Labor Statistics published no figures for October 2025). The second input counts people who have stopped looking for work, whom the unemployment rate leaves out; service members count as working. People living in institutions such as prisons and nursing homes are left out of that share, because they cannot take a job. The survey does not count them by county, so we use each county's 2020 Census count of institutional residents aged 25 to 54; where that 2020 count is larger than the county's adults not working in the current survey, which happens in a few counties with large prisons, the county takes its state's median. The score rule did not change.
Composition 3 took effect September 29, 2026. It kept the same inputs and changed what the published number is: before it, the score was the composite itself, an average of percentiles that bunched toward the middle, so a score of 90 sat far above 90% of counties. The score became the share of the other counties with a lower composite, as a whole number, so it reads directly as the share of counties a county is more distressed than. Ranks still sort the composite. Scores published before that date are on the earlier scale.
Composition 2, which took effect July 28, 2026, removed SNAP participation, EITC claim share, and transfer income share. Those measures describe program reach or the government-response share of income rather than direct household need; they remain available as county data, but they do not enter the CDI score.
Citation
American Default Research. (2026). County Distress Index: Methodology and Scoring. Retrieved from https://americandefault.org/methodology/cdi/