Appendix A: Federal credit and the cost of investment
The lending programs, their budget costs and the proposed financing package.
Supports financing the mobilization.
What federal lending can pay for
Federal credit pays for different parts of the national work. The federal government can lend directly, or guarantee repayment of financing supplied by another lender. A guarantee can bring private capital into a project whose risk a bank would otherwise refuse. In either case, the agency must find the borrower and project eligible under the program Congress created.
The programs’ estimated 2025 lending shows their scale; it is not an inventory of money waiting for the next president. The amount available for new work in 2029 depends on each program's authority, commitment limits, credit subsidy or borrower payments, and amounts already committed to other projects.
In the FY2026 Federal Credit Supplement, issued in June 2025, OMB estimated about $192 billion in direct loan obligations excluding student loans in 2025, and about $1.16 trillion in gross guarantee commitments. The guarantee total includes mortgage-backed securities and can count guarantees at more than one level of the same financing. It is not a measure of unique new credit, much less a fund available for industrial renewal.1
The catalog compares the June 2025 estimates with FY2025 actual obligations or commitments subsequently reported in the FY2027 budget. An obligation commits the government to financing a loan; disbursement occurs as funds are paid out. The actual column therefore does not measure cash already received by borrowers. Its subsidy rates still estimate lifetime credit costs, even though they are recorded beside a completed year’s lending activity. Both columns express credit cost in cents per dollar lent: one cent per dollar is a 1 percent subsidy rate.
| Program | FY2025 estimate published in June 2025 | Estimated subsidy cost per dollar | FY2025 actual obligations or commitments; credit cost in cents per dollar |
|---|---|---|---|
| Energy Infrastructure Reinvestment, Section 1706, before its redesign as EDF | $44.9bn | Negative. Expected net receipts | $44.144bn; −3.77 cents |
| Small Business Administration 7(a) guarantees | $35.0bn | Zero. Fee-funded | $33.372bn; zero |
| Advanced Technology Vehicles Manufacturing | $26.5bn | 4.1 cents | $25.787bn; 3.71 cents |
| SBA 504 debentures | $12.5bn | Zero. Fee-funded | $6.883bn; zero |
| Innovative Technology loans, Section 1703 | $11.1bn | 1.2 cents | $6.699bn; 0.92 cents |
| CHIPS direct loans | $10.0bn | Negative | $5.500bn; −1.07 cents |
| Water Infrastructure Finance and Innovation | $8.8bn | 0.9 cents | $1.491bn; 0.64 cents |
| Rural Utilities Service, electric loans and guaranteed underwriting | $7.4bn | Negative for all three components | $9.024bn; negative for each component |
| Transportation Infrastructure Finance and Innovation | $5.7bn | 1.4 cents | $6.401bn; 1.46 cents |
| Rural business and industry guarantees | $1.7bn | 0.2 cents | $2.112bn; 0.20 cents |
| Maritime Administration ship financing | $1.1bn | 4.1 cents | No new obligations or commitments; no rate reported |
| Office of Strategic Capital, Department of Defense | $0.9bn | 4.5 cents | $0.150bn; 13.64 cents |
| Community Development Financial Institutions bond guarantees | $0.5bn | Zero | $0.100bn; −4.15 cents |
| HUD Section 108 community development guarantees | $0.4bn | Zero | $0.210bn; zero |
The actuals come from OMB’s FY2027 Analytical Perspectives credit tables, Tables 4-2, 4-3, 4-8, and 4-9, and the detailed program schedules in the FY2027 Budget Appendix, printed pages 150, 406, 572, 951, and 1143–1144. The RUS row consistently combines Treasury electric loans, FFB electric loans, and FFB guaranteed underwriting: $5.416 billion, $2.708 billion, and $900 million in actual 2025 activity, at rates of −1.61, −4.30, and −2.89 percent. It excludes other rural electric and telecommunications lending. SBA 504 excludes its separate commercial-real-estate refinance program; the CDFI row excludes other CDFI assistance.
Some programs committed much less than projected. WIFIA’s $1.491 billion was roughly one-sixth of its $8.8 billion estimate; SBA 504 reached $6.883 billion against $12.5 billion; ship financing recorded no new commitments against $1.1 billion. These differences justify finding out what held projects up. They do not establish that the difference remains available to spend, or that every missed loan was caused by administrative delay.
Several of these carry a negative estimated subsidy rate: expected receipts exceed estimated credit costs. That is a budget estimate, not a promise that every loan will repay or that administration costs nothing. Federal lending for rural electricity dates to 1935; the Rural Utilities Service is an established instrument the president can put to work.
The newer budget also reports about $163 billion in direct loan obligations excluding federal student loans. Its summary reports about $576 billion in guarantee commitments after removing Ginnie Mae and SBA secondary-market guarantees. These figures use a different basis from the older gross guarantee estimate above, so the difference is not a measure of shrinking program capacity. Both aggregates contain large housing and other programs outside this plan. Neither is a fund that an incoming president can assign to new industry. OMB, FY2027 credit tables, Tables 4-2 and 4-4.
The opportunity is to get eligible projects through programs that have often been underused. The transition must establish the actual uncommitted capacity in each one; subtracting projected lending from a statutory ceiling is not enough.
The Federal Financing Bank supplies financing for authorized federal programs, including by making loans and purchasing obligations. It can provide Treasury-backed financing, but it does not expand the participating agency's authority to approve projects. It helps fund an authorized loan; it cannot make an ineligible project eligible.2
The agencies retain their lending decisions. The Federal Credit Reform Act generally requires budget authority covering estimated credit costs before a program makes new loans or guarantees, unless an appropriations act otherwise authorizes the commitments. Zero or negative estimated subsidy costs do not remove statutory loan limits. Some statutes also permit borrowers to pay credit costs, including Title 17’s energy programs. That route needs both an eligible project and usable commitment authority. 2 U.S.C. §661c(b); 42 U.S.C. §16512(b).
OMB’s subsidy estimates and apportionment decisions affect how much and when agencies can commit. The RFC team should obtain, program by program, the legal commitment limit, usable subsidy, commitments already made, and remaining capacity. Where funding is available but an application is stuck, its job is to organize a decision. Where funding or authority is missing, its job is to identify the specific congressional action needed. The renewal calendar and eligibility account explain those conditions. OMB Circular A-129 sets the government-wide credit-management requirements.
The financing Congress is asked to provide
The national work begins with the authority and funds the administration inherits. These requests would replenish or enlarge that capacity. They cover different instruments: credit subsidy supports loans and guarantees, capital at community lenders revolves as it is repaid, and Defense Production Act Title III funds production capacity directly.
Credit subsidy is the up-front budget provision for the government's estimated net cost of a loan or guarantee. That cost reflects expected repayments, interest, fees, and defaults. Where the cost is positive, dividing the subsidy by the estimated cost per dollar gives the lending volume it can support, subject to the program's legal limits. A billion dollars supports $50 billion at 2 cents per dollar and about $11 billion at 9 cents.
The figures below are financing scenarios. Some use the estimated 2025 subsidy rates in the catalog above; the Section 1703 request uses the IRA's historical package as a precedent; the EDF row assumes a 2-cent rate. They are not promises about future credit costs, unused balances or loans already available. Congress must provide the necessary commitment authority as well as the funding, and programs with sunsets must be extended for new lending.
| Move | Funding requested | Illustrative financing or investment supported |
|---|---|---|
| Provide Section 1703 subsidy and additional commitment authority | $3.6bn | $40bn of guarantee authority, using the IRA package as a legislative precedent |
| Refill Advanced Technology Vehicles Manufacturing | $1–3bn | $24–73bn of direct loans, at 4.1 cents per dollar, subject to available authority |
| A successor to the Greenhouse Gas Reduction Fund | $10–30bn | $10–30bn of capital at state and community lenders, revolving as it is repaid |
| Renew and fund Defense Production Act Title III | $1–10bn | $1–10bn of production capacity. This is direct spending and does not multiply |
| Extend EDF commitment authority and provide further subsidy | $1–5bn | $50–250bn, assuming 2 cents per dollar and enough room within the $250bn ceiling. Energy and critical minerals, including processing and refining, but not general manufacturing |
| Extend and capitalize the Office of Strategic Capital | $1–2bn | $22–44bn of industrial-base lending, at 4.5 cents per dollar |
| Total, rounded | $18–54bn | $148–448bn of combined financing and direct investment |
The ATVM and OSC scenarios assume credit costs of 4.1 and 4.5 cents per dollar. The actual 2025 rates illustrate why future loan volumes must be recalculated when projects and credit terms are known.
The total combines loans, guarantees, revolving capital, and direct production spending; it is not all federal lending. Amounts are rounded from the underlying ratios. Actual credit costs and statutory limits could change the volume supported by any of these requests.
The distinction between debt and deficit matters here. For a direct federal loan of $1 billion with a 2-cent subsidy rate, the estimated $20 million subsidy is the up-front budget cost. The government finances the principal by borrowing and holds the borrower's repayment obligation as an asset. A guarantee of private lending is different: it does not necessarily require Treasury to borrow the principal when the guarantee is issued. Both instruments carry credit risk and are budgeted for their estimated cost. A negative subsidy rate means the government expects receipts to exceed costs; it is an estimate, not a guarantee of profit.3
The EDF row makes the conditions concrete. At the assumed rate, $5 billion of subsidy could support $250 billion of guarantees only if that much commitment authority were still available and Congress extended the deadline. The $1 billion appropriated in 2025 is neither proof of a remaining balance today nor a basis for fresh 2029 commitments after the current deadline. Likewise, $3.6 billion for Section 1703 would not automatically recreate the IRA’s $40 billion tranche after its September 30, 2026 expiration: the legislative request must supply both pieces.
Reconciliation is the proposed route for provisions that satisfy its budget rules. Extensions or other changes that cannot qualify need another legislative vehicle. The political work is to keep existing instruments usable and win more financing through the Congress the president has, while building support for the full RFC charter.
Federal credit’s relatively small recorded budget cost helps explain why maintaining these instruments is a plausible congressional objective. In January 2026, CBO estimated that $184 billion of commercial loans and guarantees would carry $3.5 billion in lifetime cost under federal credit accounting, an average 1.9 percent subsidy rate. Twenty-nine of the 66 commercial programs had zero or negative rates. Its fair-value estimate, which includes market risk, was higher: $22.2 billion, or 12.1 percent. Administration costs are separate. These are comparisons for CBO’s program mix, not a promised rate for this plan. CBO, Estimates of the Cost of Federal Credit Programs in 2026.