M-18-12 made IT modernization a loan to repay and a fund agencies lacked the authority to fill
OMB’s February 27, 2018 memo put the Modernizing Government Technology Act to work. Its central fund lent money to be repaid, and its agency funds needed transfer authority the act did not give, so budget offices became the gatekeepers.
Looking Back revisits past weeks one at a time, up to the most recent, and asks what it meant for the banks, health systems and public agencies we serve. The series was written in 2026, with the benefit of hindsight, and each article says which week it looks back on.
Two-minute summaryThe dek and the first sentence of each section, quoted from the article
OMB’s February 27, 2018 memo put the Modernizing Government Technology Act to work. Its central fund lent money to be repaid, and its agency funds needed transfer authority the act did not give, so budget offices became the gatekeepers.
- The week On Tuesday, February 27, 2018, the Office of Management and Budget issued memorandum M-18-12, Implementation of the Modernizing Government Technology Act.
- The proposal: two pages that had to answer a budget question The initial proposal was short by design.
- The transfer: money in increments, tied to milestones An approved project did not receive its money at once.
- The repayment: a loan with terms set by GSA and OMB The third part is where the fund departed from an ordinary appropriation.
- The working capital fund: an account with no way in The fourth part was the agency’s own vehicle.
- How this differed across agencies and public bodies The four parts weighed differently on different organizations.
- Objections, in order We take five objections, numbered in the order we address them.
- What a leader could check These checks follow from the four parts.
- With hindsight The fund was capitalized slowly at first and then all at once.
The week
On Tuesday, February 27, 2018, the Office of Management and Budget issued memorandum M-18-12, Implementation of the Modernizing Government Technology Act. It was addressed to the heads of executive departments and agencies. The memo set out the administration’s objectives for the act and the actions agencies should take to implement it.
The Modernizing Government Technology Act, the MGT Act, had been enacted on December 12, 2017, as sections 1076 through 1078 of the National Defense Authorization Act for Fiscal Year 2018, Public Law 115-91. The memo described its two main provisions. The first established a central Technology Modernization Fund in the Treasury, overseen by a Technology Modernization Board. The second authorized each agency covered by the Chief Financial Officers Act to set up its own information technology working capital fund.
For the central fund, the memo explained the model. Agencies would submit project proposals to the board, which would weigh financial, technical and operational criteria. Approved projects would receive money incrementally, tied to milestones, and the board would monitor them. Agencies had to reimburse the fund under a written agreement setting out the purpose and the repayment terms. Under the act, repayment terms could not exceed five years unless OMB approved. The act authorized up to $250 million in appropriations for each of fiscal years 2018 and 2019. The memo noted that fiscal 2018 appropriations were still pending.
The memo described the board. Under the act it would be established on March 12, 2018, with seven members. The Administrator of the Office of E-Government, the federal chief information officer, would chair it. The others would be a senior official from the General Services Administration with technical expertise, a member of the Department of Homeland Security’s National Protection and Programs Directorate, and four federal employees appointed by OMB. Those four would bring expertise in IT development, financial management, cybersecurity and privacy, and acquisition. The board would evaluate proposals, set criteria, recommend funding to the GSA Administrator and monitor approved projects.
The memo opened the proposal process the same day. Agencies were encouraged to send initial project proposals to an OMB mailbox from February 27, 2018, using a template in Appendix A. The template limited proposals to two pages in 10-point type. It asked five things: how the project served the mission, the funding request and cost estimate method, the metrics and milestones, why the project would succeed, and why the fund suited the project. Appendix B listed considerations, including impact on users and security, feasibility, ability to repay the fund, and common solutions.
For working capital funds, the memo restated the act’s permitted uses. An agency fund could only be used to improve, retire or replace existing systems, to move legacy systems to commercial cloud and other commercial platforms, and to address evolving security threats. It could also reimburse the central fund, and pay for activities Congress had not denied or restricted. By March 26, 2018, each CFO Act agency was to tell OMB whether it intended to set up a fund in fiscal 2018. An agency that set one up would report to OMB every quarter on planned transfers, IT Dashboard updates and a summary of obligations, expenditures and unused balances.
The memo ended its working capital fund section with a limit. It reminded agencies that all transfers and reprogramming into the funds were subject to applicable appropriations law. It said that the MGT Act “does not confer transfer authority.” Agencies could move money into a working capital fund only if some other authority allowed the transfer. It encouraged them to review their existing agency-specific authorities to find accounts with that authority.
A funding memo can look like plumbing. We read M-18-12 as the moment the federal government priced IT modernization. It set two prices. For the central fund, the price was a loan, repaid on a written schedule from the agency’s own later appropriations. For the agency working capital funds, the price was finding transfer authority that the act itself did not supply.
Both prices fell on the same people: agency chief information officers and chief financial officers, working with their budget offices and OMB. Both prices turned a technology question, whether a legacy system should be replaced, into a budget question, whether the agency could commit future money to pay for it. The memo was explicit about the second price. The first was in the structure of the fund.
We take the memo in four parts, following the money. Each part is a separate point where a modernization project could stall. Together they cover the path from an idea to a repaid investment:
- The proposal: what an agency had to show to get the board’s attention, and what that cost.
- The transfer: how money moved to an approved project, in increments tied to milestones.
- The repayment: what the agency committed to pay back, from what, and on what schedule.
- The working capital fund: the agency’s own vehicle, and the authority it lacked.
Our argument is that the memo made the agency’s budget office, not its technology office, the gatekeeper of modernization. A strong technical case could not get through the repayment or the working capital fund without a budget plan that committed future money. Agencies that read the memo as an invitation to pitch projects, without a repayment source and a transfer path, would find the money harder to use than it looked. After the four parts, we look at how this differed across agencies and other public bodies, at objections, and at what a leader could check.
A note on sources. We rely on the memo itself, the act as enacted, the fund’s own site and the Government Accountability Office’s July 2026 review. Appropriations law questions are specialized, and agencies should rely on their own counsel and budget offices for any particular transfer or reprogramming.
The proposal: two pages that had to answer a budget question
The initial proposal was short by design. Two pages, in 10-point type, with general information excepted. It asked for a CIO sponsor and a CFO sponsor by name. That second name was the first sign that the fund was a budget instrument as much as a technology one.
Five questions, one of them financial
The template’s five questions covered mission, money, milestones, the case for success and the case for the fund. The second question asked for the funding request and the cost estimation method. It added that if the request was less than the total project cost, the agency should state the total and explain how it intended to repay the fund. Repayment was in the first submission, before any board review.
For a technology team, the cost estimate was familiar ground. The repayment plan was not. It required the budget office to say where money would come from in future years, when the agency did not yet know its future appropriations. The template did not demand certainty. It did ask how the agency intended to repay, and it asked for a CFO sponsor by name.
Who writes the proposal
The template’s sponsor fields named a project sponsor, a CIO sponsor and a CFO sponsor. Three names meant three offices had to agree before submission. In an agency where the CIO and CFO had not worked closely on investment decisions, the proposal forced the conversation. The template asked the agency to avoid acronyms and terms unfamiliar outside the program, which suggests OMB expected readers from outside the agency’s technology community.
Our reading is that the proposal should be drafted jointly from the start. A technology team that wrote the case and sent it to the CFO for a repayment line would get a weak answer. A CFO asked early could shape the project so that its savings, if any, landed in an account from which repayment was possible. That design choice is easier at the proposal stage than after approval.
The considerations in Appendix B
Appendix B listed considerations agencies were encouraged, though not required, to address. They grouped under impact on the mission, feasibility, opportunity enablement and common solutions. The opportunity heading included expected cost savings, lasting financial impact and the ability to repay. It asked how the agency would save costs or improve service, and how business process and change management would be achieved, including repurposing of workforce as necessary.
That last phrase linked savings to people. A project that replaced a legacy system would save money only if the work and staff around the old system changed too. An agency proposing savings would need a credible account of that change. Savings estimates without it would be hard to defend to the board, and hard to deliver.
The cost of proposing
Writing two pages is cheap. Producing a defensible cost estimate, a milestone plan and a repayment source is not. An agency would spend staff time from its CIO, CFO and program offices before knowing whether the board would take an interest. The template said a proposal selected for further consideration might receive help developing an in-depth proposal and business case. The first round was the agency’s own cost.
Our reading is that agencies with existing capital planning disciplines had an advantage. An agency that already maintained business cases and portfolio reviews could adapt existing material. An agency without that discipline would be building it for the proposal. The memo rewarded preparation that predated it.
Projects that save nothing
Some modernization does not save money. Replacing an insecure system with a secure one may cost as much to run as before, or more. The memo’s considerations included improving security as a source of impact, alongside savings. A security project could be a strong candidate on impact and a weak one on repayment.
For such a project, repayment had to come from the agency’s future budget without any offsetting saving. That made the loan a straightforward advance of money the agency would have to find later. Whether that was worth it depended on how urgent the risk was and whether the agency could not otherwise fund the work in time. The proposal had to make that case openly, rather than dress a security project in savings it would not deliver.
Duplicating appropriations
The memo said the fund was not meant to duplicate funding provided through existing appropriations, including agency working capital funds. It also said agencies could not seek fund money for any project Congress had expressly denied or restricted. A proposal therefore had to show that the project was not already funded and not opposed. For a project that an agency had sought in its budget and not received, the second test needed care. Whether Congress had denied it, or simply not funded it, was a question for the budget office and counsel.
The transfer: money in increments, tied to milestones
An approved project did not receive its money at once. The act required that funds transferred for improving, retiring or replacing systems be released incrementally, tied to metric-based development milestones, absent compelling circumstances the GSA Administrator documented. The memo restated that approved projects would receive funds incrementally, tied to specific milestones, with regular board monitoring.
What incremental funding demands of a project
Incremental funding works well with iterative development. The act tied milestones to rapid, iterative development processes. A project built in brief releases, each delivering something measurable, could meet milestones and unlock the next tranche. A project designed as a single long build, with value only at the end, would struggle to show milestones the board could accept.
That shaped acquisition. A contract structured for a single delivery at the end does not fit funding released in tranches. Contracting officers would need vehicles that allowed incremental orders and exit points. The act also preferred commercial products and services, and required the written agreement to document that preference. Projects heavy in custom development would need to justify it.
The risk of a stalled tranche
Incremental funding protects the fund. It shifts risk to the agency. If a milestone slipped, the next tranche would wait, while contractors and staff remained on the project. An agency would need a plan for that gap: other funds to bridge it, or a contract structure that could pause without penalty. A stalled tranche could cost more than the delay itself if the team dispersed.
Our reading is that agencies should plan each tranche as if the next might be late. That means contracts with natural stopping points, teams that can shift to other work, and a clear account to the board of what each milestone delivers on its own. A milestone that delivers value by itself is a safer bet for both the agency and the fund.
Acquisition staff in the loop
The act tied the working capital funds to full and open competition among commercial items to the greatest extent practicable. It tied central fund transfers to a documented preference for commercial products and services. GAO was to report on whether agencies used full and open competition for custom development bought with fund money. Acquisition offices were therefore part of the project from the start, not an afterthought once funding arrived.
That had a staffing cost. Contracting officers familiar with agile delivery, modular contracts and commercial cloud were in demand across government. An agency proposing a fund project needed to know who would run the acquisition, on what vehicle, and how long it would take to award. A milestone plan that assumed a fast award without that check would slip at the first milestone.
When a project should stop
Incremental funding also creates a decision point at each milestone: continue, change or stop. A project that is not delivering can be halted before the full amount is spent. That protects the fund and, in principle, the agency. But stopping has its own costs. Money already transferred must still be repaid under the written agreement, and the agency may be left with a partly built system and a legacy one still running.
Our reading is that each milestone review should include an honest stop option, with its costs set out. An agency that had planned for the possibility of stopping, with contracts that could end cleanly and a fallback for the legacy system, would be able to make that call. One that had not would be under pressure to continue a failing project to justify the money already spent.
Monitoring and public reporting
The act required OMB to maintain, and publish on a public website, a list of funded projects updated at least quarterly. The list had to show project status, schedule delays and cost overruns, financial expenditure data, and the extent of commercial products and services use. An agency taking fund money accepted that its project’s progress would be public, in a standard format, every quarter.
That visibility is a cost and a discipline. A project in trouble would show it publicly. For a CIO, that argues for honest milestones, set conservatively, rather than ambitious ones that would show as delays. The act also required the Comptroller General to report every two years on the fund’s costs, the reliability of savings estimates and the use of competition. Agencies could expect their savings claims to be tested by GAO.
The repayment: a loan with terms set by GSA and OMB
The third part is where the fund departed from an ordinary appropriation. Money from the fund had to be repaid. The act said so, and the memo restated it.
The written agreement
Before any transfer, the GSA Administrator, in consultation with the OMB Director, and the agency head had to sign a written agreement. It documented the purpose of the funds and the terms of repayment, which could not exceed five years unless the Director approved. The agreement was recorded as an obligation. The act set the accounting rule: an obligation to make a payment under the agreement in a later fiscal year would be recorded in the fiscal year the payment was due.
That accounting rule mattered to budget offices. It meant the agency did not record the whole repayment as an obligation up front. Each year’s payment would appear in the year it fell due. The agency would still need appropriations in each of those years to make the payment.
Where repayment money comes from
The act allowed an agency to repay from any appropriation made available after enactment for information technology activities, consistent with reprogramming law and the appropriations committees’ guidelines. That gave flexibility on paper. It did not create new money. Repayment would come from the agency’s future IT budget, competing with operations, maintenance and other projects.
Savings were the natural source. A project that reduced operating costs could, in principle, pay the fund back from those savings. But savings arrive after the project is complete and the old system retired, and they arrive in the operating budget, which may be in a different account. Moving them to repay the fund would itself need reprogramming or transfer authority. The repayment plan in the proposal had to account for that path.
Who carries the repayment inside a department
In a large department, the borrower named in the written agreement is the agency, but the project may belong to one component. The repayment then has to be assigned inside the department: to the component that benefits, to a central IT account, or shared. If that assignment is not agreed before the loan, the repayment can become a dispute among components years later, when the people who agreed to the project have moved on.
An internal agreement, mirroring the written agreement with GSA, settles that early. It names the paying account for each year and what happens if the component’s budget falls short. It also records who receives the savings, so the party that repays is the party that benefits.
Prices fixed for solvency
The act told the GSA Administrator, in consultation with the Director, to set repayment amounts and terms at levels sufficient to ensure the fund’s solvency, including operating expenses. That meant repayment could include more than the principal. The memo pointed to further guidance on disbursement and repayment. As our hindsight section describes, GSA collected fees for a period and later stopped.
Repayment in the budget request
A repayment obligation recorded in the year it falls due has to be funded in that year. In federal budgeting, the agency’s request for that year is prepared well in advance. The budget office therefore had to carry the repayment into future requests, explain it to OMB’s budget examiners and defend it to appropriations committees. A repayment that was not requested would have to be found inside an enacted budget.
That created a coordination job between the agency’s IT budget and its overall request. Repayment lines would sit alongside requests for new work, and the committees would see both. An agency asking for new IT money while repaying the fund would need to explain how the two related. The fund’s promise of faster modernization came with a longer tail of budget explanations.
Enforcement against the agency
The act gave the fund a remedy if an agency did not pay. The remedy applied if payment was not made within 90 days after a repayment period in the written agreement expired. The Administrator could then obtain reimbursement by issuing transfer and counterwarrants or other lawful transfer documents, supported by itemized bills. By the statute’s design, the fund could take its money back from the agency’s accounts.
For an agency CFO, that changed the nature of the commitment. A missed repayment would not sit as a debt to settle later. It could be collected. The repayment schedule had to be treated as a fixed charge on future budgets, with the risk that a shortfall elsewhere could not be covered by delaying it.
The working capital fund: an account with no way in
The fourth part was the agency’s own vehicle. The act authorized each CFO Act agency to establish an IT system modernization and working capital fund. The memo gave the guidance, and named the limit.
What could go in
The act named two sources. One was reprogramming and transfer of funds made available in later appropriations acts, including funds for operating and maintaining legacy systems. That route had to comply with reprogramming law and committee guidelines, or rest on transfer authority in appropriations law. The other was discretionary appropriations made to the fund directly. Both depended on Congress, either through an appropriation or through transfer authority.
The memo said it directly: the act does not confer transfer authority. An agency that wanted to move savings from operations into its working capital fund needed authority from somewhere else. Agencies were encouraged to review their existing authorities and identify accounts that had transfer authority. Some would find it. Others would not.
What could come out
The act limited uses. A working capital fund could pay for improving, retiring or replacing existing systems, and for moving to commercial cloud and other commercial platforms. It could also pay for addressing security threats, reimbursing the central fund, and activities Congress had not denied or restricted. It could not supplant funds already appropriated for operating and maintaining a system when the fund was set up. The agency head had to prioritize cost-saving activities approved by the CIO, and could deposit savings from those activities into the fund, consistent with reprogramming rules.
The three-year clock
Money deposited in a working capital fund was available for obligation for three years from the end of the fiscal year of deposit. Unobligated amounts after that went to the general fund of the Treasury. That was longer than one-year money and shorter than no-year money. An agency that put savings into the fund had to obligate them within the window, or lose them.
For a multi-year modernization, the clock shaped planning. Deposits had to be matched to project phases so that money would be obligated in time. A project that slipped could see deposits expire before they were needed. The fund rewarded agencies with steady pipelines of modernization work, ready to absorb deposits.
Savings first
The act told agency heads to prioritize working capital fund money first for cost-saving activities approved by the CIO. It let them reprogram and transfer savings from those activities into the fund, consistent with reprogramming rules. That created a sequence: fund savings projects first, capture the savings, recycle them into further modernization.
The sequence made sense. It also depended on measuring savings credibly. A CIO approving a cost-saving activity needed a baseline of current costs, a method for counting savings and a way to show they were real. Agencies without good cost data for their legacy systems would find it hard to prove savings, and therefore hard to justify moving money on the strength of them.
Setting up the account
Establishing a working capital fund was a policy decision and an accounting task. It required a new Treasury account. Agencies intending to set up a fund in fiscal 2018 were to tell OMB by March 26, 2018, through their resource management office and the federal CIO’s desk officer. They were also to begin creating the account. An agency deciding later had to give at least 30 days’ notice before establishing one. It also had to identify the accounts with transfer authority and submit their names to the same offices.
Those steps involved the budget office, the financial systems team and OMB. An agency that wanted a fund had to plan for the account setup, the accounting rules for deposits and the reporting before any money moved. The account was the easy part. The authority to fill it was the hard part.
Using the agency fund to repay the central fund
The two vehicles were linked. One permitted use of a working capital fund was to reimburse the central fund, with the approval of the agency CIO in consultation with the CFO. An agency with both could plan to repay a central fund loan from its own working capital fund. That would let savings, once deposited, flow to repayment within the agency’s IT accounts.
The plan still depended on getting savings into the working capital fund, which needed transfer authority. Without it, the link existed on paper only. With it, an agency had a closed loop: borrow from the central fund, deliver savings, deposit them in the working capital fund, repay. Our reading is that the loop was the act’s intended design, and the missing transfer authority was the break in it.
Reporting
The act required each agency with a working capital fund to report to OMB within one year of enactment and every six months after that. It required a list of each investment funded, with estimated cost and completion date, and a summary of obligations, expenditures and unused balances by fiscal year. OMB would publish them. The memo asked for more: quarterly reports on planned transfers and reprogramming with justifications, IT Dashboard updates, and the obligations summary. It said OMB would align that reporting with existing processes.
Why the limit mattered
The working capital fund was meant to let agencies keep savings and reinvest them in modernization. Without transfer authority, an agency could not move savings into it on its own. It needed Congress to appropriate to the fund or to grant transfer authority. That put the working capital fund back in the ordinary appropriations process the act had been meant to work around. Our reading is that the memo told agencies, plainly and early, that the fund’s promise depended on authority they might not have.
How this differed across agencies and public bodies
The four parts weighed differently on different organizations.
Large CFO Act departments
Large departments had much to gain and the machinery to pursue it. They could set up working capital funds, had budget offices experienced with transfers, and might already hold transfer authorities in their appropriations. Their risk was internal: components competing for fund money, and repayment obligations falling on component budgets that had not agreed to them. A department-level fund needed internal rules on who could draw, who repaid and how savings were credited.
Smaller agencies outside the CFO Act list
The working capital fund authority applied only to CFO Act agencies. The central fund was open to agencies as defined in the Administrative Procedure Act, which is broader. A small independent agency could propose to the central fund but could not set up its own working capital fund under the act. For such an agency, the central fund was the only new vehicle, and its repayment terms fell on a small IT budget. A five-year repayment would be a larger share of that budget than of a department’s.
Agencies with fee or trust fund financing
Some federal programs are financed by fees or trust funds rather than annual appropriations. Whether such programs could repay the fund from those sources, or deposit savings into a working capital fund, would depend on the statutes governing those funds. The MGT Act’s repayment clause referred to appropriations for IT activities. Agencies in that position needed counsel’s view before relying on fee income for repayment.
State and local governments
The MGT Act applied to federal agencies. State and local governments faced the same economics in their own budgets: the cost of maintaining legacy systems, the difficulty of capturing savings, and rules on moving money between accounts. The federal memo offered them a reference design: proposals with repayment plans, incremental funding, milestone reporting and a clear statement of what authority the fund did and did not carry.
Inspectors general and auditors
Fund projects created new things to audit: written agreements, milestone evidence, repayment schedules, transfers into working capital funds and the authority behind them. Inspectors general and financial statement auditors would see these in agency books. A transfer into a working capital fund made without proper authority could raise appropriations law questions. Agencies would benefit from documenting the authority for each transfer at the time, so the audit trail was complete.
Vendors and integrators
For contractors, the fund changed the shape of opportunities. Incremental funding, a preference for commercial products and a requirement for iterative delivery favored firms that could deliver in releases and accept work that might pause between tranches. Contracts for fund projects would need to reflect the funding structure. A vendor expecting a single large award would find the fund’s design pushed toward smaller, staged ones.
Objections, in order
We take five objections, numbered in the order we address them.
1. The fund was small, so its design hardly mattered
The first objection is that an authorization of up to $250 million a year was small against federal IT spending. The authorization was modest, and the memo noted that fiscal 2018 appropriations were pending. But the design was a template. Its repayment model, incremental funding and reporting set expectations for how modernization money should work, whatever the amount. As our hindsight section shows, the fund later received far more.
2. Repayment would come from savings, so it was self-financing
The second objection is that projects would pay for themselves. The memo itself pointed to cost savings as a consideration, and savings could repay the fund. But savings arrive late, land in other accounts and may need transfer authority to move. The act did not make repayment contingent on savings. A repayment plan that relied on savings alone was a forecast, not a source.
3. OMB could approve longer terms, so repayment was flexible
The third objection is that the five-year limit could be extended. The act allowed the Director to approve longer terms. That added flexibility at the margin. It did not change the nature of the obligation, the solvency requirement or the collection remedy. A longer term spreads payments. It does not remove them.
4. Working capital funds would gain authority over time
The fourth objection is that Congress would grant transfer authority as funds proved themselves. That might happen agency by agency. The memo’s point stands for the period it addressed: on February 27, 2018, the act did not confer transfer authority, and each agency had to find its own. Planning a modernization on authority that Congress might grant later was a risk the memo flagged.
5. Technology quality, not budget mechanics, decides modernization
The fifth objection is that good technology choices matter more than funding mechanics. Both matter. The memo’s own considerations stressed technical approach, team strength and commercial solutions. Our argument is narrower: a sound technical plan could still fail to reach funding, or fail to finish, if its budget path did not work. The memo made that path explicit.
What a leader could check
These checks follow from the four parts. They are our suggestions, not requirements.
- Does every modernization proposal name a CFO sponsor and a repayment source, with the account it would come from in each year?
- Are contracts for incrementally funded projects structured with stopping points that match the funding milestones?
- Are milestones set so that each delivers value on its own and can be reported publicly without embarrassment?
- Has the budget office identified, in writing, which accounts have authority to transfer money into a working capital fund?
- Are working capital fund deposits matched to project phases so that money is obligated before the three-year window closes?
Each check is a budget office task as much as a technology one. A CIO who can answer them with the CFO has a modernization plan. One who cannot has a project idea and a funding hope.
The checks also need owners. The repayment source and the budget request belong to the CFO’s budget office. Contract structure belongs to the senior procurement executive’s staff. Milestones and their reporting belong to the project’s executive sponsor and the CIO. Transfer authority belongs to the budget office and the general counsel’s appropriations lawyers. A leader should know who answers each question before the board asks it.
With hindsight
The fund was capitalized slowly at first and then all at once. GAO’s July 2026 review, GAO-26-107737, records that the fund received $100 million in fiscal 2018 and $25 million in each of fiscal years 2019 through 2021. It received $1 billion from the American Rescue Plan Act in fiscal 2021 and $50 million in fiscal 2023. In fiscal 2024, $100 million of the American Rescue Plan money was rescinded. GAO said that from fiscal years 2018 through 2025 the fund received over $1 billion in net appropriations.
The repayment model bent. GAO reported that the original guidelines required agencies to start repaying within one year of the first transfer and finish within five years unless OMB approved otherwise. Repayment did not depend on savings. GSA collected administrative fees to cover the program office’s costs, and GAO said GSA stopped collecting them from new projects in March 2021, after the American Rescue Plan Act. In July 2022, GAO issued a legal decision, B-333396, finding that the authorizing law left OMB and GSA discretion to set rates below full reimbursement.
Policy then moved back toward repayment. GAO reported that in February 2024 GSA set a minimum 50 percent repayment for new investments, with exceptions requiring GSA and OMB approval. In May 2025, GSA announced a shift to prioritize full repayment for new investments. The loan model of 2018 had passed through partial repayment and was moving back toward full repayment. GAO noted that the May 2025 guidance still allows partial repayment arrangements for projects that tackle urgent IT cybersecurity and modernization problems.
GAO also confirmed the original process. It recorded that agencies were allowed to begin submitting initial application proposals on February 27, 2018, under OMB’s guidance, and that the template asked for an estimate of the funding request. It cited separate OMB funding guidelines for agencies receiving disbursements, dated March 12, 2018, the day the board was established.
Savings, the natural repayment source, arrived slowly. As of June 2025, GAO said, the board had invested about $1.03 billion in 68 unclassified projects, and 40 agencies and offices had received fund money. Twenty-four projects expected total savings of about $1.06 billion. Eleven had realized about $13.5 million, and 21 projects, expecting about 98 percent of the total, planned to finish achieving savings in fiscal 2027 or later. Thirty-seven projects expected no cost savings, aiming at other value such as reduced security risk. Of six completed projects that expected savings, two met or were on track to meet them and four did not. Officials attributed the shortfalls to causes such as removed functionality and higher migration costs.
Competition largely held up in the contracts GAO examined. GAO reviewed the 177 contract actions awarded on the 32 projects that received fund money between March 1, 2023 and June 2, 2025. Of those, 154 were awarded using competitive procedures, accounting for about $713.1 million, 96 percent of the funding awarded. In the other 23, agencies documented authorized exceptions.
The fund’s own site, checked today, describes flexible and incremental funding, with transfers unlocked only as projects complete milestones, and repayment flexibility. It reports over $1.1 billion invested in 74 projects across 34 federal agencies. GAO’s count covered unclassified projects as of June 2025, so the two figures are not directly comparable. GAO also wrote that, as of July 2026, the fund’s authorization was set to expire at the end of September 2026, and that without reauthorization there would be no new investments. An official source shows what happened next. The Continuing Appropriations and Extensions Act, 2027, Public Law 119-103, approved September 2, 2026, amended the act’s sunset provision, section 1078(f)(1), by replacing “September 30, 2026” with “December 11, 2026.” As enacted in 2017, that provision barred new awards or transfers from the fund for projects not already in progress once the sunset date passed. Among the public laws posted on govinfo as of November 15, 2026, we found none that changed the date again.
What held up from our argument is the central role of the budget office. Repayment terms, fees and the savings that were meant to pay for projects were central to GAO’s 2026 review. What did not hold up is the strictness of the loan. GAO’s 2022 decision confirmed that OMB and GSA could set repayment below full reimbursement, and GSA’s 2024 policy set a floor of 50 percent rather than requiring full repayment. The incremental, milestone-based transfer model has held throughout, and the fund still describes it as its core.
Spotted an error? Tell us and we will check it and correct the article where needed.