A Federal IT Fund Required Savings Estimates but Not Savings

← Back · ACT II · Chapter 9: The Value Control Loop

In July, the Government Accountability Office published its latest review of the Technology Modernization Fund, the money federal agencies can draw on to replace aging IT systems. As of June 2025 the fund had invested about $1.03 billion in 68 unclassified projects across 40 agencies and offices. Twenty-four of those projects expected to save money, about $1.06 billion in all. Eleven had realized some savings so far, about $13.5 million between them. Thirty-seven others expected no savings and were funded for other reasons, such as reducing security risk.

Most of the coverage stopped at that gap, and fairly so. Modernization takes years to pay off, and 21 of the 24 expect to finish achieving their savings in fiscal 2027 or later. A shortfall this early does not prove much on its own.

The part I would want a buyer to read is a sentence in the report’s background section, about the rules the fund started with.

The Estimate Came First

When OMB issued the funding guidelines in March 2018, it required every proposal to include “a reliable estimate of any project-related cost savings or avoidance relative to pre-modernization activities.” The estimate had to undergo “appropriate due diligence and concurrence from the requesting agency’s Office of the Chief Financial Officer” before it went to the board that approves the money. That is a serious entry requirement, and more than I have seen many corporate capital committees ask for.

Agencies also had to pay the money back, and at first a fee as well. Then the report adds: “While the guidelines noted that reimbursement was not contingent upon achieving project-related savings, agencies could use the project’s generated cost savings to repay the investment.”

Read plainly, the estimate helped decide whether a project got funded, and the repayment owed afterward did not depend on whether the estimate came true. Later guidance changed how much has to be repaid, and the report notes that partial repayment may now be considered for projects that tackle urgent cybersecurity and modernization problems. The terms are set in a written agreement with the General Services Administration (GSA), which administers the fund, and nothing the report describes adjusts them for whether the estimated savings arrive. Savings could help an agency pay the fund back, and the agency owed what it owed if they never came.

Most buyers will know the shape, repayment aside. The number does its job on approval day, and once the money is released very little in anyone’s budget, bonus or contract depends on it again.

What Happened to the Estimates

Seven of the projects that expected savings had finished by June 2025, and six of them had been complete for at least a full fiscal year, which was long enough for GAO to assess them. Of those six, two met their targets or are on track to, and four did not. One, a mainframe migration at the Department of Housing and Urban Development, was expected to save about $48.16 million. HUD did see lower annual costs once it left the old platform. But in March 2026, officials told GAO the project “did not and would not achieve any cost savings” compared to its baseline costs, because moving the mainframe environment to the cloud, and the transition itself, brought increased costs.

For the projects that fell short, the report says officials pointed to “the removal of planned functionality and higher system migration costs than planned.” Both are things that happened after approval. Removing functionality shrinks what the project delivers, while the estimate that justified the money was built on the larger version.

None of this says the projects were badly run. GAO found the savings data “sufficiently reliable for our purposes,” and several of these programs may deliver later. The fund’s program office does monitor projects, and GAO reports on savings every two years. My point is narrower. The estimate was examined with care before the money was released. After that, nothing in the repayment terms gave anyone a financial reason to keep it true.

Which Saving Counts

The report closes with a disagreement. In written comments on the draft, GSA’s administrator objected that significant amounts agencies did not have to spend “are categorized as cost avoidance rather than included in realized savings totals,” and that separating the two “understates the full financial benefit generated by TMF investments.” GAO agreed the totals would be higher with cost avoidance included, and answered that the law requires it to report on cost savings, “and that is what this report covers.”

Both positions are reasonable. Look at when the question is being settled, though. The 2018 guidelines asked for an estimate of savings “or avoidance.” Eight years and about a billion dollars later, the fund’s administrator and its auditor are on record with different answers about whether avoided costs belong in the results.

Anyone who has sat through a post-implementation review knows this meeting. The business case promised hard savings. The review finds some productivity, some costs avoided and a lot of opinion, and the hour goes to arguing about categories instead of checking results. It happens late because nobody wrote down early which kind of saving the purchase was meant to produce, how it would be counted against the starting baseline, and who would do the counting.

Write the Definition Before the Money Moves

Chapter 9 of The Value Standard describes the value control loop, where the result is defined before the decision, then delivered, verified and acted on. The fund’s rules built a careful first step. Nothing in the repayment terms connected that step to the ones after it, which is the open loop I described in Most Companies Set Goals for Purchases and Never Check Them.

None of this needs repayment tied to results, which may not suit a government fund at all. What a buyer can do is settle, before approval and in its own words, which kind of saving counts, how it will be measured, and when it will be checked, and then keep that definition as the buyer’s record afterward. It belongs in the document a buyer writes before suppliers bid, the one stating the result the purchase has to produce. That document is a Request for Value. The contract can carry the same definition through the Verification Line. Both have Word copies on the Guides page.

A supplier that delivers gains from this as much as the buyer does. On a migration that hits every milestone and still misses the savings figure in the funding request, the vendor’s work gets judged against a number it may never have agreed to. A definition settled up front gives good work something fair to be measured against.

The estimate had a template and the concurrence of a CFO’s office before the money moved. Eight years later, the fund’s administrator was making the case in a comment letter to GAO that avoided costs should count toward the result.

References: U.S. Government Accountability Office, Technology Modernization Fund: Small Savings Achieved So Far, but Substantial Future Savings Expected, GAO-26-107737, July 23, 2026.

Keep reading: ACT II · Chapter 9: The Value Control Loop