Geography · Data snapshot
Where DBE firms cluster — and why per-capita federal funding doesn't predict it
Published May 24, 2026 · ~5 minute read
DBE Source's database holds 132,698 certified disadvantaged, small, and women-owned businesses, sourced from every state DOT's public certification directory plus the SBA and major reciprocal programs. That's a useful vantage point for a question we get asked surprisingly often: where are DBE firms actually located, and what explains the differences?
The headline finding is the federal-enclave cluster. Maryland, Virginia, and the District of Columbia are home to 35,552 certified firms — about 27% of every DBE in the database, in 4.6% of the country's population. They're also the top three by per-capita density: 264 DBEs per 100,000 residents in DC, 231 in Maryland, 218 in Virginia. The next state down, Alaska, sits at 86. There is a real cliff between the federal-contracting region and everywhere else.
That part isn't a surprise. The District is a federal enclave; northern Virginia and the Baltimore-Washington corridor host the contracting infrastructure that's grown up around DC for half a century. Where it gets more interesting is when you try to explain the rest of the map.
The intuitive theory: follow the federal highway money
DBE goals on federally-funded transportation contracts are the mechanism that originally created most of these certification programs. So a reasonable hypothesis: states that receive more federal-aid highway dollars per resident should have more DBE-eligible contracting activity per resident, which should translate into more certified DBE firms per resident.
We tested it. For each of the 50 states plus DC, we looked at FY2024 federal-aid highway apportionment from the FHWA, divided by Census-estimated 2024 population. Then we plotted that against DBE density.
The relationship is essentially flat. The Pearson correlation is +0.06 across all 51 jurisdictions and +0.01 if you remove DC. Statistically indistinguishable from no relationship.
Why the intuition fails: floor protections
FHWA apportionment formulas have strong minimum-funding protections for small-population states. The top five recipients per capita are Alaska ($934), Wyoming ($601), Montana ($497), Vermont ($431), and North Dakota ($430). None of those are high-DBE-density states except Alaska. Vermont in particular receives more than three times the federal highway funding per resident that California does, while supporting roughly one tenth the DBE density.
Meanwhile the high-DBE-density cluster — DC, Maryland, Virginia — receives between $132 and $313 per resident, which is middle-of-the-pack. The per-capita normalization that lets you compare DBE density across states actively inverts the federal-funding signal: the states getting the most highway dollars per person are the ones with the least absolute contracting activity per person.
On an absolute basis — total DBE firms versus total FHWA dollars, ignoring population — there is a strong correlation (r = 0.58). But that's essentially a restatement of “big states have more of both”: total DBE count correlates equally strongly with population alone (r = 0.60). The federal-spending channel doesn't survive either normalization choice.
Top and bottom states by DBE density
For the curious, here are the rankings. The cliff between the federal-enclave region and the next tier is the most striking feature.
| Rank | State | DBEs | Population | Per 100k | FHWA $/cap |
|---|---|---|---|---|---|
| #1 | DC | 1,857 | 702K | 264.4 | $313 |
| #2 | MD | 14,477 | 6.3M | 231.1 | $132 |
| #3 | VA | 19,218 | 8.8M | 218.1 | $159 |
| #4 | AK | 639 | 740K | 86.3 | $934 |
| #5 | OR | 3,162 | 4.3M | 74.0 | $161 |
| #6 | GA | 7,224 | 11.2M | 64.6 | $159 |
| #7 | LA | 2,229 | 4.6M | 48.5 | $210 |
| #8 | AZ | 3,434 | 7.6M | 45.3 | $133 |
| #9 | MN | 2,633 | 5.8M | 45.5 | $155 |
| #10 | DE | 438 | 1.1M | 41.6 | $222 |
| Rank | State | DBEs | Population | Per 100k | FHWA $/cap |
|---|---|---|---|---|---|
| #51 | IA | 281 | 3.2M | 8.7 | $209 |
| #50 | KY | 639 | 4.6M | 13.9 | $200 |
| #49 | UT | 497 | 3.5M | 14.2 | $137 |
| #48 | VT | 100 | 648K | 15.4 | $431 |
| #47 | AR | 478 | 3.1M | 15.5 | $231 |
A few specific observations from the long tail. Oregon (#5) stands out as the most DBE-dense state outside the federal-enclave region. Louisiana (#7) and Georgia (#6) are notably higher than their neighbors, likely reflecting active state-DOT certification outreach. At the other end, Iowa is the lowest-density state at 8.7 firms per 100,000 residents. Vermont, despite collecting more federal highway funding per resident than nearly any other state, has only 100 certified DBE firms in our database.
What we think is actually going on
With FHWA per-capita funding ruled out, the leading remaining candidates for what drives DBE density are:
- Metropolitan concentration. DBE firms cluster in metro areas where contracting opportunities concentrate. States that are more metro-dominant have proportionally more firms.
- State-DOT program aggressiveness. Some state DOTs run very active outreach, certification assistance, and ongoing recruitment. Others process applications and not much else. Active programs build deeper registries over time.
- Historical compounding. The original DBE programs date to the early 1980s. States that adopted early and stayed engaged accumulated firm registries that persist even when current activity is modest.
- Reciprocal certification. A firm headquartered in Texas can hold a Maryland certification — and the firm shows up in our database under Texas (HQ) but the Maryland certification reflects Maryland program activity. Our HQ-state count understates this for states that aggressively accept out-of-state firms.
None of those are things we can pull from a single public API, which is the next research problem.
Methodology
All numbers reproducible from public sources. The full analysis script is open-source on GitHub: scripts/analysis/dbe-density-by-state. Clone, run, and tell us what we missed.
- DBE company counts
SELECT state, COUNT(*) FROM companies GROUP BY state. HQ state, normalized to two-letter postal at import. Filtered to the 50 states plus DC; 899 records with non-standard state codes (Puerto Rico, Guam, Northern Mariana Islands, American Samoa) excluded from this analysis since the comparison sets cover only those 51 jurisdictions.- Population
- US Census Bureau Population Estimates Program (PEP) vintage 2024, NST-EST2024-ALLDATA.csv. Column
POPESTIMATE2024. No API key required. - FHWA apportionment
- FHWA Highway Statistics, Table FA-4 (FY2024 Federal-Aid Highway Program Apportionments). fa4.xlsx,
APPORTIONED TOTALcolumn. Per-capita = total apportionment / 2024 population. - Statistical method
- Pearson correlation and ordinary least-squares regression via SciPy
stats.linregressand NumPylinalg.lstsq. DC reported separately as a sensitivity check given its outlier nature.
Caveats
A few things worth keeping in mind. n=51 is statistically thin; small effects shouldn't be over-interpreted. FY2024 FHWA apportionment is one fiscal year; actual obligations and outlays differ and lag. FHWA captures highway dollars only — FTA transit and FAA aviation DBE programs are not included. And our count is certified firms, not minority-owned firms in general; gaps may reflect certification-program participation rather than the underlying firm population.
Update: 3-year FHWA averaging didn't change the answer
One natural objection to the headline finding above is that FY2024 might be a noisy single-year snapshot — maybe year-to-year volatility in FHWA apportionment is what's breaking the correlation, and a multi-year average would reveal the spending channel. We re-ran the analysis with the three-year FY2022–FY2024 average per-capita apportionment instead of FY2024 alone.
The correlation didn't move. Pearson r = +0.062 (p = 0.668) across all 51 jurisdictions, identical to the single-year number to three decimal places. Excluding DC, r = +0.008. The reason is structural: FHWA apportionment is stable across years because the underlying formulas are IIJA-budgeted multi-year amounts, not annually-appropriated line items. Alaska's per-capita figure moved from $934 (FY24) to a $916 three-year average; Wyoming's from $601 to $589. The top-per-capita ranking didn't change. Whatever is breaking the federal-spending channel isn't single-year noise.
Update (May 26, 2026):we've since sourced the FTA and FAA data and completed the extension. That follow-up post is here. Short version: adding FTA + FAA produces a significant all-states correlation (combined r = +0.29, p = 0.04), but the relationship collapses to r = +0.05 (p = 0.75) once DC is excluded. The combined-spending channel is a DC-metro story, not a national one. FTA alone shows the strongest single-program correlation (r = +0.62 all-in, +0.13 ex-DC), and the transit-heavy northeast corridor accounts for the weak ex-DC signal.
Data snapshot: May 24, 2026. Census PEP vintage 2024, FHWA Table FA-4 FY2024, DBE counts current as of publication.