The blank column
Washington agencies filed 91,421 contracts in fiscal year 2025. On 47,780 of them, the space for what the contract cost is empty.
What the record shows
Washington publishes every state agency contract in an annual dataset. Each row carries the agency, the contractor, the purpose, the dates, and four money fields: state, federal, other, and total cost.
Across the twelve years from FY2014 to FY2025 — 642,567 contract records — 41.4% report no dollar figure at all. The money fields are blank, or zero, or a dash.
The rate is not stable. It sat between 25% and 34% for seven years, then stepped up sharply in FY2021 and has stayed there. FY2025, at 52.3%, is the worst year in the series.
Which contracts
The dataset records how each contract was procured. From FY2018 — the first year the field is populated — one category stands apart.
Client Service – Exempt contracts disclose no dollar figure 65.3% of the time, against 40.1% for competitively procured ones. The gap holds in every single year.
That category is not an obscure corner of state spending. It is the channel through which Washington pays for child care, home care, behavioral health, shelter, and reentry services — the contracts that reach people with the least capacity to ask questions about them.
The obvious objection
Some contracts genuinely cost nothing. Interagency agreements, data-sharing arrangements and memoranda of understanding move no money, and the dataset has a field for exactly this: a non-financial contract flag.
If most of the blanks were non-financial contracts, there would be no story. So we checked.
The flag exists from FY2021 onward. Across all five of those years it accounts for a small minority of the blanks — at most 15.5%, and in FY2022 almost none. Strip the non-financial contracts out entirely and between 36.9% and 48.7% of everything Washington files each year remains a financial contract with no amount disclosed.
| Fiscal year | Contracts filed | No cost stated | Flagged non-financial | Financial, unexplained | Share of all filings |
|---|---|---|---|---|---|
| FY2021 | 66,577 | 29,116 | 4,524 | 24,592 | 36.9% |
| FY2022 | 67,931 | 30,841 | 133 | 30,708 | 45.2% |
| FY2023 | 74,495 | 36,844 | 1,686 | 35,158 | 47.2% |
| FY2024 | 79,329 | 36,176 | 1,220 | 34,956 | 44.1% |
| FY2025 | 91,421 | 47,780 | 3,298 | 44,482 | 48.7% |
What this is not
We expected to find that human-services providers were disclosed less than other contractors. They are not.
Contracting entities linked to Washington’s 2,833 licensed child care sites report no cost on 40.9% of their rows — statistically level with the 41.4% statewide rate, not worse.
The opacity is a property of the procurement channel, not of the people paid through it. Any version of this finding that singles out a sector, a vendor or an agency would be reading something into the data that is not there.
Nor is a missing figure evidence of wrongdoing. Nothing here shows money misspent. It shows that for roughly half of the contracts Washington signs, the public record does not say what they cost — which means the question cannot be asked from the published data at all.
Why it matters
A contracts database exists so that a resident, a reporter or a legislator can see what the state bought and what it paid. When the amount is absent, every downstream use fails: you cannot total an agency’s spending, compare vendors, track a program across years, or notice that a provider’s public funding tripled.
The gap is closable without new legislation. The field already exists on the form. In FY2016, three-quarters of contracts were filed with it completed.
Method, sources and limitations
Source. Washington Office of Financial Management, Agency Contracts, one dataset per fiscal year, FY2014–FY2025, published on data.wa.gov. All twelve datasets were downloaded in full: 642,567 records.
Definition. A record counts as “no cost stated” when the total cost field is empty, zero, or a placeholder dash. Amounts are parsed from the published strings, which arrive in several formats ($15,000.00, $ - 0, 0). A parsed value of “nothing here” is kept distinct from a real zero.
Limitations.
- The procurement-type field is unpopulated in the FY2014–FY2017 datasets, so Exhibit 2 begins at FY2018. The category labels are also inconsistently spelled across years and were grouped by prefix.
- The non-financial flag is optional and self-reported by agencies, and only present from FY2021, so Exhibit 3 begins there. Because it is optional it is probably under-used, which would make the unexplained share smaller than stated — the check is run in the direction that weakens our own finding, and the finding survives it in all five years.
- Contract counts are records, not distinct contracts: amendments appear as separate rows, and a multi-year contract recurs across fiscal years. Rates are therefore shares of filings, not of unique agreements.
- This measures disclosure, not spending. A contract with no stated cost may still be fully documented in the agency’s own records and in OFM’s accounting systems.
Reproducing it. Every figure comes from the published datasets with no manual entry. The child care comparison joins the contract records to the DCYF licensed provider registry (data.wa.gov/d/was8-3ni8) by normalized name, and counts each contracting entity once rather than once per site.