This is every expenditure transaction South Carolina state government recorded in fiscal year 2025, reorganized to answer a question the budget itself cannot answer: not which agency spent the money, but what the money was spent on.
Where the data comes from. The Comptroller General publishes a spending transparency file every month, covering all agencies that run through the state's central accounting system. All twelve FY2025 monthly files were downloaded and combined — 1,706,089 individual transactions totaling $38.24 billion across 97 agencies. This data is public and free. It requires no FOIA request and no special access. Twelve fiscal years are posted, back to FY2014.
What was done to it. Each transaction carries an expenditure object code — the state uses 960 of them, with labels like "Professional Services" or "Motor Vehicle Maintenance." All 960 codes were run through written, rule-based logic: an ordered list of terms mapped to categories, applied identically to every code, so the same input always produces the same output and any classification can be checked and disputed on the merits. Codes were first sorted into spending that is formula-driven and set in statute versus spending that is operational, then the operational portion was grouped into nineteen function areas. 685 codes matched a rule and carry 99.5% of all dollars; 275 low-value codes did not match and are reported as unclassified.
Why this view may be useful. The appropriations act is organized by agency, so a function that 90 agencies each buy separately appears as 90 unrelated line items and never as a single number. Reorganizing the same dollars by function makes duplication visible, shows where the state buys the same category of thing many times over, and separates spending that could be changed administratively from spending that would require changing the law.
Four in five dollars are entitlements and formula transfers set in statute. That leaves ~$6.6B of operational spend, examined in the nineteen categories below.
Every rectangle below is one of the nineteen operational function areas, sized by FY2025 dollars. Click any category to open it and see the individual expenditure object codes inside it, each sized the same way. This is the entire operational side of state spending on one screen — the largest categories are immediately visible, and the long tail of small ones is visible too, in proportion.
Toggle to switch what the bars measure. The first view ranks each function by total dollars spent. The second ranks the same nineteen functions by how many of the state's 97 agencies spend money in that function at all — a count of how widely each category is purchased across government, regardless of size.
This table covers all operational spending, not any single category — formula transfers and entitlement payments are excluded, and placeholder entries the state uses for bulk postings (payroll, employer contributions) have been filtered out. Every row is a single payee that received money from at least ten different state agencies during FY2025.
246 payees meet that threshold, accounting for roughly $964 million — about 14.6% of all operational spending. Sixty-nine of them were paid by twenty or more agencies. The twelve largest are shown below, with the function area that accounts for most of each one's spending.
Why it matters. Each of these relationships was generally established separately, by a different agency, with its own terms, renewal date and negotiated price. A payee serving 84 agencies through 84 arrangements represents 84 separately made decisions about the same category of thing — and there is no place in the current budget structure where that total appears as one number.
Rows marked STATE are state entities billing other agencies internally rather than outside vendors. They are worth reading separately: they show that shared-service arrangements already operate inside state government in some categories, alongside fully decentralized purchasing in others.
| Payee | Primary function | FY2025 operational spend | Agencies |
|---|---|---|---|
| Department of AdministrationSTATE | Information Technology | $98.4M | 69 |
| TAPFIN | Information Technology | $67.0M | 26 |
| Bank of America — P-Card | Vehicles, Fuel & Fleet | $61.3M | 84 |
| SFAA AdministrationSTATE | Insurance, Claims & Risk | $46.1M | 81 |
| USC Columbia CampusSTATE | Contracts with Govts & Nonprofits | $45.6M | 36 |
| Mansfield Oil Co | Vehicles, Fuel & Fleet | $34.8M | 57 |
| SHI International Corp | Information Technology | $34.0M | 50 |
| AT&T | Telecommunications & Connectivity | $30.2M | 41 |
| Dominion Energy South Carolina | Utilities | $28.7M | 34 |
| Clemson UniversitySTATE | Information Technology | $24.8M | 24 |
| KCI Technologies Inc | Buildings, Land & Capital Construction | $23.5M | 10 |
| Segra | Telecommunications & Connectivity | $17.7M | 79 |
Information technology illustrates the pattern. Eighty-eight of 97 agencies spend on IT, through 1,387 separate vendors, and no agency dominates — the five largest IT spenders account for 56% of the total. The median agency spends about $700,000: large enough to matter to that agency, small enough that it carries little leverage on price.
The vendor counts span three orders of magnitude. Travel and lodging flows to 12,706 distinct payees; personnel training and memberships to 7,570; insurance and claims — nearly universal at 89 agencies — to just 728, in part because the state already runs central insurance arrangements. The same government operates highly centralized and highly fragmented purchasing side by side, category by category. The data shows which is which; it does not say which is right.
All twelve FY2025 monthly spending files were pulled from the Comptroller General and combined into a single dataset of 1.7 million transactions, reconciled to $38.24 billion. All 960 expenditure object codes were then run through an ordered list of written rules mapping label terms to categories. Nothing depends on a model's judgment call at classification time — the rules are fixed text, applied identically to every code, and can be reviewed line by line.
The rules went through several audit passes, and the audits caught real errors — including roughly $1 billion of general contract spending that an early keyword had misfiled into information technology because "IT" appears at the end of the word "nonprofit." Each pass involved reading the largest codes in every category and the unclassified remainder, correcting the rules, and re-running. That loop is the method; the categories are only as good as the last audit.
Three patterns emerged. First, the addressable share is far smaller than the headline: 82% of state spending is formula-driven and changes only by changing statute. Second, dollars and breadth live in different places — roads and buildings hold the largest sums but sit in a handful of agencies (95% of highway spending is one agency), while IT, supplies, travel, facilities and training are purchased almost universally and independently. Third, a small set of payees quietly serves most of state government, with no single budget line reflecting it.
How this differs from the usual view. Budget documents present spending agency by agency — the right structure for accountability, and the wrong one for spotting duplication. A function that 88 agencies each buy appears as 88 unrelated items. Reorganizing the same dollars by what they purchased makes the pattern visible without changing a single number.
From object codes to programs. The state's chart of accounts and cost-center structure would connect expenditure codes to the programs agencies actually run. Object codes describe what was bought; programs describe what it was bought for. Without that link, this analysis can show that 88 agencies buy IT, but not which functions of government that IT supports.
Connecting funding to purpose. A program-to-fund crosswalk would show which revenue sources support which programs. Much operational spending is paid from federal or restricted funds that carry their own conditions, so a dollar that looks reducible on paper may not be legally available for redirection.
Personnel below the agency line. Position and staffing detail by program would show where people are actually assigned — the largest recurring commitment in most agencies, and the one this dataset sees only indirectly.
Contract terms and renewal dates. The payee concentration above identifies where overlap exists but not when anything can change. Terms and renewal calendars would turn a list of overlaps into a sequence of dated decision points.
A settled taxonomy. The nineteen categories here were built bottom-up from the object codes in a few evenings. The state already maintains function classifications of its own — including the strategic objectives on its budget request forms and the functional groupings in its annual financial reporting. A fuller version of this work should adopt an established taxonomy rather than invent one.