Research · Analysis · Public ContractsThe Public Service Institute
1.3M+ records2004–2026 coverage11 publications
Data Note PSI-2026-05 October 2026 Methodology ↓

Where the Contracts Are: Value Distribution, 2004–2026

Nearly half of all federal awards are worth less than $25,000. Three-quarters are worth less than $100,000. The distribution of federal contracts by value is a long tail attached to a very thin head — and almost every public conversation about procurement studies the head.

The Public Service InstitutePSI-2026-05Compiled from open government data
Key findings
  1. Of 1,298,088 banded awards, 46.0% (596,768) are under $25,000; 76.3% are under $100,000; 95.3% are under $1 million.
  2. The median federal award falls in the $25K–$100K band.
  3. Awards over $10 million — 9,098 in total — represent 0.7% of the count; awards of $1 million or more are 4.7%.
  4. Band counts cover 1,298,088 of 1,313,281 records; the residual (15,193) carries null or non-positive values and is excluded.
  5. Cross-reading vendor concentration: the top 100 vendors hold 12.8% of awards but 62.6% of value — the value mass sits in the thin head of this distribution.

The shape of the record. The federal proactive publication captures every contract over $10,000, which makes it a rare thing: a complete distribution, not a sample of large deals. Banded by value, the corpus is overwhelmingly small. 596,768 awards — 46.0% of everything the federal government has contracted since 2004 — are worth less than $25,000. Another 30.4% fall between $25,000 and $100,000. By $1 million, 95.3% of the distribution is exhausted. What remains above that line is a rounding error in count terms — 60,544 awards — yet that error is where the money lives.

under $25K46.0%$25K-100K30.4%$100K-1M19.0%$1M-10M4.0%over $10M0.7%
Share of federal awards by value band, 2004–2026 (N = 1,298,088 banded records).
Federal awards by value band
Value bandContractsShare of awardsCumulative
under $25K596,76846.0%46.0%
$25K-100K394,16930.4%76.3%
$100K-1M246,60719.0%95.3%
$1M-10M51,4464.0%99.3%
over $10M9,0980.7%100.0%

A limit of the published data, stated plainly. The corpus reports contract counts per band, not value per band, so the dollar share of each band cannot be computed directly — a gap we flag rather than paper over. The cross-reading that survives: the 100 largest vendors win 12.8% of awards and 62.6% of value, which is only possible if value concentrates radically in the upper bands. The distribution's two ends are therefore the same fact measured twice: a market of 9,098 awards over $10M carrying most of $1.21T, resting on a base of 990,937 sub-$100K awards that carry almost none of it.

What the tail is for. For suppliers, the sub-$100K region is the actual federal market — the part reachable without program-scale positioning, where a firm can build a performance record one modest award at a time. The vendor data bears this out: outside the top 100 vendors, the average award is $396K, and hundreds of long-lived federal suppliers hold positions assembled entirely from such awards. For policymakers, the same numbers frame the small-business set-aside question — set-asides operate on a segment that is 76.3% of transactions but a small fraction of value, which is a defensible design, and one the distribution now quantifies.

What the head is for. The 0.7% of awards over $10 million — ships, enterprise IT, aircraft fleets, construction programs — is where the vendor-concentration findings of The Concentration Economy come from. Oversight effort is conventionally aimed here, and the distribution explains why: it is the only tier where a single award moves departmental numbers. But the head is also where competition is thinnest, because fields of qualified bidders for $100M commitments are short by nature. The distribution does not say the attention is misplaced; it says the two ends obey different economics and should not be regulated as one market.

Housekeeping that matters. The $10,000 publication threshold defines the corpus, so the true count of small federal transactions is larger than any figure here — awards below the threshold are simply invisible. Banding uses the recorded nominal value per award; multi-year commitments amended upward over time will migrate between bands as recorded. The 2026 records are partial-year. None of these caveats disturbs the finding; a distribution with 46.0% of its mass under $25K has room for a great deal of noise.

Suggested citation
Public Service Institute. (October 2026). Where the Contracts Are: Value Distribution, 2004–2026 (PSI-2026-05). https://publicserviceinstitute.org/reports/value-bands.