Business Buyer School Research · Methods · Version 2026.1

Methodology: The Industry Durability Study

How 152 industries were measured and scored for durability, and what the study cannot show.

Business Buyer School calculation. Scores, ranks and rates are calculated by Business Buyer School from official Census Bureau data. These rates describe whole industries, not any one business. An establishment exit is a location that closes, not a business failure, and a sold business is not an exit. They are not the chance that a business you buy will close.

Version 2026.1 published: October 6, 2026
Analysis prepared: October 6, 2026
Data window: Census Bureau Business Dynamics Statistics, 2023 release: annual rates averaged over 2014 to 2023 (March 2013 to March 2023)
Scoring: Four components as percentile ranks among 152 industries; Model A weights 40, 25, 20 and 15

Short version · Full study · Methodology · Data · How to cite

Version 2026.1. Run date 2026-10-06. Calculation versions durability-metrics-v1, durability-score-v1. Raw data and the industry universe are shared with Report 1, the Industry Resilience Study.

1. Question

How often do establishments exit, firms shut down and jobs get destroyed in acquisition-relevant U.S. industries, and which industries are most and least durable by those measures? This is industry durability. It is not business failure, profitability, or the odds that a particular business will close.

2. Source

U.S. Census Bureau, Business Dynamics Statistics 2023 release, file bds2023_vcn4.csv (national, 4-digit NAICS 2017 vintage-consistent, 1978 to 2023). One source and one NAICS vintage, so no crosswalk or splice is needed. BDS values are noise-infused by Census. In the columns used here, no included industry has a suppressed cell (the 'D' (suppressed) flags in exit and firm-death columns belong to six excluded industries); the script asserts it.

3. Universe and confidence

Identical to Report 1: 288 four-digit codes, 167 included by the structural rules U1 to U6, High 118, Medium 34, Low 15. High and Medium are ranked (152); Low are scored but unranked. Flags carry over: BDS 2023 re-matched establishments for 4451 and 4541 (2015 to 2023), which can affect establishment exit rates, so those are capped at Medium.

4. Metrics (all are averages of ANNUAL rates over a window)

Establishment exit rate: BDS estabs_exit_rate (exits divided by the BDS denominator, the average of establishments in years t and t-1), average 2014 to 2023 (ten years; includes COVID). Also averaged 2019 to 2023, 2014 to 2019 and 2004 to 2023 for robustness.
Firm shutdown rate: firmdeath_firms divided by the average of firms in t and t-1, average 2014 to 2023.
Job destruction rate: BDS job_destruction_rate (jobs lost at exiting and shrinking establishments), average 2014 to 2023.
Exit-rate volatility: standard deviation of the annual exit rate 2004 to 2023, in percentage points.
Context (not scored): entry rate, employees per establishment, cumulative exits 2014 to 2023, employment in 2023.
What is not computed: a cohort survival rate. BDS 4-digit tables carry no establishment age split, so no rate here can be read as the survival odds of a new or acquired business.

5. Score

Mid-rank percentiles within the 152 ranked industries (100 = most durable; lower exit, shutdown, destruction and volatility score higher). Model A (pre-specified): exit 40, firm shutdown 25, job destruction 20, volatility 15.

6. Sensitivity

Models: B exit-heavy 55/20/15/10; C firm-and-job-heavy 25/35/30/10; D equal; E exit rates only (10-year 50, 5-year 35, volatility 15); F winsorised z-scores with Model A weights; plus 3,000 Dirichlet weight draws. Minimum pairwise Spearman across the six models 0.949. Stability classes: 30 stable most durable, 33 stable least durable, 2 method-sensitive, 87 moderate.

7. Red-team review

Is exit rate just turnover? Largely yes: exit and entry rates correlate 0.81. Entry-adjusted rescore (exit rate residualised on entry rate): Spearman 0.891 with Model A, 20 of 25 retained. Disclosed in the report; the score is a measure of low turnover as much as of survival.
Does COVID distort the window? Window variants: 2014 to 2019 (Spearman 0.997, 24 of 25), 2019 to 2023 (0.995, 24), 2004 to 2023 (0.992, 24). Not sensitive.
Is establishment exit firm failure? No. Firm shutdown is a separate component and the two correlate 0.87, not 1. A location can close while the firm continues.
Do large establishments drive the result? Durability correlates 0.43 with employees per establishment and -0.32 with log establishment count. Large-establishment industries look more durable; the report says so and does not claim causation.
Age mix. BDS 4-digit tables have no age split. Industries with many young establishments may show higher exit rates regardless of how mature businesses behave. Not correctable with this source; the report states that rates are not odds for an acquired business.
Weights. Leave-one-component-out top-25 overlap: exit10 21, firm 23, jobdest 24, vol 23.
Small and unstable industries. Low-confidence industries are unranked; 7 of the top 25 are Medium confidence and marked.
Are buyer conclusions stronger than the evidence? Buyer section limited to churn, not-odds, and independence of durability from recession resistance.
BDS ownership treatment. BDS follows establishments through ownership change; a sale does not create an exit.

8. Limitations

Industry-wide annual rates, not cohort survival; no age split; employer establishments only; national; 4-digit groups mix trades; noise-infused data; ten-year window includes COVID; durability measures neither profit nor price.

9. Versioning

Version YYYY.N; a new BDS release triggers a re-run and version increment; weight or window changes require a calculation_version increment.

How to cite

Cite as: Business Buyer School, Which Small Businesses Last the Longest? U.S. Establishment Exit and Industry Durability Study (methodology), https://businessbuyerschool.com/research/small-business-exit-rates/methodology/. Version 2026.1. Published October 6, 2026.

A data license has not been published; short quotations with attribution and a link are welcome.

Take this into your own search

Industry research is context, not a verdict on any one business. These free parts of the school show how to use it.

Write your Buy Box: set the type of business, size and hard exclusions before you search.
Cleaning, landscaping and HVAC compared: how the same principles show up in three trades.
See the curriculum or start free with Business Buyer School.