Reader question: What happens when one Treasury security becomes unusually scarce in the repo market and dealers need that exact CUSIP to settle trades? The Federal Reserve may own billions of that issue in the System Open Market Account. How does it lend those securities without simply handing them to whichever dealer asks first?
The New York Fed operates a daily SOMA Securities Lending Program. Participating primary dealers submit competitive lending-fee bids for specific Treasury or agency debt securities. Each security is auctioned using a multiple-price process, subject to issue-level and dealer-level borrowing caps. The programme is designed to support smooth clearing and settlement and to reduce scarcity-related market frictions without transferring permanent ownership of SOMA securities.
Current key parameters include:
- auction each Bank business day at 12 noon ET;
- minimum bid rate: 5 basis points;
- bid increment: $1 million;
- two bids per issue per dealer;
- dealer maximum per issue: 25% of theoretical amount available;
- dealer total outstanding loan cap: $5 billion par;
- theoretical supply: generally 90% of SOMA holdings for securities with at least 14 days to maturity;
- loan term: overnight under market conventions;
- eligible pledged Treasury collateral currently margined at 102% of loaned-security market value.
These are current programme terms and should be treated as effective-date parameters, not permanent constants.
What this page owns — and what it does not
This page owns:
SOMA issue inventory + dealer bids + issue/dealer caps → multiple-price loan awards → collateral pledge → overnight fee/return → extension/fail state.
It does not replace repo pricing and specialness, general securities-lending algorithms, or Federal Reserve outright Treasury operations. SOMA lending is a specific central-bank securities-loan auction.
This is public market-infrastructure mathematics, not a recommendation to borrow, lend or trade any Treasury security.
Why the programme exists
The New York Fed states that SOMA securities lending promotes the smooth clearing of Treasury and agency debt securities in support of monetary-policy implementation.
The key problem is issue-specific scarcity.
Suppose a dealer has sold a Treasury note and must deliver that CUSIP today, but the issue is extremely scarce in repo.
A temporary SOMA loan can add that exact security to settlement supply without changing the total size of the Federal Reserve’s long-run Treasury holdings.
Lending fee versus repo rate
The dealer submits a lending fee rate, not a repo rate.
The New York Fed FAQ explains that under the borrow-versus-pledge structure the fee can be viewed as roughly analogous to:
Lending Fee ≈ General Collateral Repo Rate − Special Repo Rate.
If a Treasury trades very special, its special repo rate is far below general collateral, so the implied scarcity spread is large.
A dealer can therefore rationally bid a larger lending fee for a scarce issue.
Why the minimum bid is 5bp
The current minimum lending bid is 5 basis points.
This creates a threshold intended to focus the programme on securities with some degree of specialness rather than replacing normal private-market financing for securities trading near general collateral.
A 0bp bid is invalid under current terms.
Step 1: calculate theoretical supply
For most eligible issues with at least 14 days to maturity:
Theoretical Supply = 90% × SOMA Holdings.
If SOMA owns $1.0 billion of a Treasury:
Theoretical Supply = $900 million.
The remaining 10% reserve reduces the chance that the lending programme exhausts the System’s position in an issue.
Actual supply can be lower than theoretical supply
Some of the theoretical 90% can already be:
- out on securities loan;
- unreturned from a prior loan;
- committed as collateral in reverse-repo operations;
- otherwise absent from the SOMA custody account at auction time.
The New York Fed calls the remaining custody amount the actual amount available.
Therefore:
Actual Available = min(Theoretical Supply, Securities Physically Available in SOMA Custody).
New York Fed’s scarcity example
The FAQ gives a useful structure.
If SOMA owns $1.0bn:
theoretical supply = $900m.
If $500m remains out on loan and another $100m is committed elsewhere, only $400m remains in custody.
Then:
actual auction supply = $400m, not $900m.
This is a strong operational lesson: inventory accounting must reflect committed state, not just gross holdings.
Step 2: dealer issue-level cap
A dealer can receive at most:
25% × Theoretical Supply
in one issue, reduced by that dealer’s outstanding loans of the same issue.
For theoretical supply of $900m:
dealer issue cap = $225m.
Outstanding loans reduce available capacity
If the dealer already has $100m of the issue out on loan at auction time:
remaining issue capacity = $225m − $100m = $125m.
If the dealer returns the prior loan before the noon auction, the full $225m capacity becomes available again.
This creates an explicit state dependency between yesterday’s settlement performance and today’s bidding capacity.
Step 3: total dealer cap
Current rules allow up to $5.0 billion par in outstanding SOMA securities loans to one dealer at a time.
Let:
OutstandingDealerLoans = Σ active SOMA loan par.
Then:
RemainingDealerCapacity = $5bn − OutstandingDealerLoans.
A dealer can submit bids across many issues, but awards stop once total capacity is exhausted.
Step 4: bid construction
Each participating dealer can submit up to two bids per issue.
Each bid contains:
- CUSIP/issue;
- par amount;
- lending-fee rate.
Bids must be in $1 million increments and fee rates in percent form to two decimal places under the current FedTrade rules.
Higher fee bids are economically stronger
The New York Fed is lending the security and earning the lending fee.
For otherwise identical bids:
higher fee = better compensation to the lender.
If a dealer submits two bids that together exceed its per-issue cap, the higher-rate bid is awarded first before the lower-rate bid is curtailed.
A two-bid cap example
The current FAQ gives:
- theoretical supply = $900m;
- dealer issue cap = $225m;
- Bid 1: $150m at 1.10%;
- Bid 2: $100m at 1.30%.
Total requested = $250m, above the cap.
The 1.30% bid is economically stronger and receives $100m first.
Only $125m of the 1.10% bid can then fit.
Total dealer award = $225m.
Step 5: multiple-price auction
The New York Fed states that loans are awarded through a multiple-price auction for each security.
Multiple-price means an accepted dealer pays the fee rate on its own accepted bid rather than every winner receiving one uniform stop-out fee.
If Dealer A is accepted at 0.20% and Dealer B at 0.35%, the accepted portions retain those different contracted lending fees.
This is distinct from the single-price Treasury issuance auction.
Issues themselves are ordered when dealer capacity binds
The FAQ states that the issue with the highest overall weighted-average bid rate is auctioned first, with remaining issues processed in descending weighted-average-rate order.
Why does this matter?
A dealer’s total $5bn borrowing capacity is reduced as awards are made.
If the dealer reaches its total cap after high-demand issues are processed, its bids on lower-ranked issues are eliminated.
So auction order affects which securities a capacity-constrained dealer ultimately receives.
Weighted-average rate
For accepted bids i on one issue:
Weighted Average Rate = Σ(qiri) / Σqi.
The New York Fed publishes the weighted-average lending rate by issue in the daily operation result.
The statistic summarizes accepted fee levels but does not mean every dealer paid that exact fee.
Current operation example
A recent August 12, 2026 New York Fed operation reported more than $43bn of securities-lending awards across issues.
Several Treasury bills in that operation showed weighted-average lending rates at the current 5bp minimum.
That indicates the programme can process a large number of ordinary low-fee loans while also allowing higher bids where particular issues are more special.
Step 6: collateral pledge
SOMA securities lending is not unsecured.
The dealer must pledge eligible collateral to the New York Fed.
The current public FAQ lists direct U.S. Treasury obligations as eligible pledged collateral with a margin percentage of 102%.
The margin is defined as:
Pledged Collateral Market Value / Loaned Security Market Value.
So a $100m market-value loan requires approximately $102m of eligible Treasury collateral under the current published schedule.
Why 102%?
The extra 2% is overcollateralization.
It protects against small market-value changes and settlement exposure between the loaned security and pledged collateral.
This is conceptually similar to a repo haircut but expressed here as a collateral-margin percentage above 100%.
Step 7: calculate lending fee
The New York Fed says the lending fee is applied to the market value of the loaned security using an Actual/360 convention.
For loan market value V, annual fee rate f, and d calendar days:
Fee = V × f × d / 360.
If V = $100m and fee rate = 0.25% for one day:
Fee ≈ $100m × 0.0025 / 360 ≈ $694.44.
The loan is overnight
The current programme permits overnight borrowing according to market conventions.
Weekend/holiday conventions can extend the number of calendar days used in fee calculations even though the loan is still operationally classified as overnight.
Why securities near maturity are excluded
Under most circumstances, issues with less than 14 days to maturity are excluded from auction.
Near-maturity securities create operational complexity because:
- redemption cash flows approach;
- settlement windows shrink;
- return and maturity events can collide.
The exclusion simplifies risk and settlement control.
Fails on return
If a dealer does not return the loaned security on maturity date, the loan becomes a fail.
The New York Fed states that failure to return can produce:
- a penalty fee equivalent to the prevailing general-collateral rate in lieu of the contracted lending fee;
- any applicable Treasury Market Practices Group fails charge;
- extension/rebooking of the failed loan;
- recollateralization requirements.
The punishment is intentionally stronger than simply “keep paying the cheap lending fee.”
Failure to deliver pledged collateral is also penalized
If the dealer cannot deliver required collateral on the loan date, the New York Fed can hold cash collateral overnight and assess additional charges.
So the state machine requires both legs:
loaned security delivered ↔ eligible collateral received.
A dealer cannot treat collateral as an optional afterthought.
How SOMA lending supports settlement
Suppose an issue is trading very special because many dealers need the same security for delivery.
Additional temporary supply from SOMA can:
- reduce settlement fails;
- reduce extreme specialness;
- improve clearing efficiency;
- support Treasury-market functioning.
The objective is market plumbing, not investment profit maximization.
The programme does not promise to eliminate specialness
A Treasury can remain scarce even after SOMA lending if:
- private demand is very large;
- SOMA owns only a small amount;
- actual lendable custody inventory is constrained;
- dealer caps bind;
- dealers choose not to bid enough.
The programme adds supply; it does not mathematically fix the special repo rate.
Counterexample: 90% theoretical supply does not mean 90% is actually available
Outstanding loans and other SOMA commitments can reduce actual custody inventory below the theoretical amount.
Counterexample: weighted-average auction rate is not a single-price award
Two dealers can pay different accepted lending fee rates even though the New York Fed publishes one weighted-average rate for the issue.
Counterexample: a high fee bid can still be capped
A dealer can submit the highest fee in the auction but still receive no more than 25% of theoretical supply for the issue and no more than its remaining $5bn total capacity.
Price priority does not override concentration controls.
Counterexample: borrowing a special Treasury does not eliminate financing risk
The dealer still must fund or pledge collateral, manage settlement and return the borrowed security on time.
The loan solves security availability, not every balance-sheet constraint.
Inputs and outputs
A SOMA lending engine can require:
- SOMA holdings by CUSIP;
- security maturity;
- outstanding SOMA loans;
- SOMA securities committed elsewhere;
- dealer identity and primary-dealer status;
- dealer outstanding loan total;
- two bids per issue;
- bid par and fee rate;
- minimum fee and increments;
- collateral eligibility/margin;
- market value of loaned security;
- settlement and maturity calendar.
Outputs can include:
- theoretical amount available;
- actual amount available;
- dealer issue cap;
- dealer total remaining capacity;
- accepted bid amounts;
- contracted lending fee by bid;
- weighted-average rate by issue;
- required collateral amount;
- loan fee;
- return/fail state.
Evidence polarity: what supports confidence?
Evidence for a reliable allocation includes theoretical supply equal to 90% of eligible SOMA holdings, actual supply reconciled to custody state, dealer limits reflecting outstanding loans, accepted bids respecting 25%/$5bn caps, multiple-price fees matching accepted propositions, collateral at current margin and published weighted-average rates reconciling to accepted bid amounts.
Evidence against confidence includes securities below 14 days maturity being auctioned routinely without an explicit exception, old $1bn total dealer limits remaining in code, a single uniform fee replacing multiple-price results, outstanding loans not reducing today’s capacity, or a 100% rather than 102% collateral requirement under the current published schedule.
Weak links in implementation
SOMA-inventory staleness. Gross weekly holdings are mistaken for current lendable custody.
theoretical/actual confusion. 90% supply is used even when securities are already committed.
dealer-cap staleness. historical programme limits remain active.
bid-order error. lower fee receives priority over higher fee when dealer bid totals exceed caps.
multiple-price error. all winners are charged one stop-out rate.
fee/repo-rate confusion. bid fee is interpreted as the repo rate itself.
collateral-margin inversion. 102% is treated as a 102% haircut.
fail-state omission. unreturned loans disappear instead of reducing next-day capacity.
Diagnostics: how to test the engine
- 90% test: $1bn eligible SOMA holding → $900m theoretical supply.
- actual-supply test: subtract loans/commitments from custody and cap auction inventory accordingly.
- 25% test: $900m theoretical supply → $225m maximum dealer issue exposure.
- $5bn test: outstanding loans reduce total dealer capacity dollar-for-dollar.
- two-bid test: if bids exceed issue cap, higher fee is allocated first.
- 5bp floor test: reject lower bids.
- multiple-price test: accepted bids retain their own fee rates.
- weighted-average test: recompute the published issue rate from accepted amounts.
- 102% collateral test: pledged Treasury market value equals 1.02× loaned-security market value.
- fail test: unreturned loan triggers penalty/extension and reduces next auction capacity.
What would falsify confidence?
Confidence should be withdrawn if published operation results cannot be reproduced from bids and limits; if custody availability is ignored; if accepted fees do not match the multiple-price rule; if dealer capacity is wrong after an outstanding fail; or if collateral and loaned securities do not reconcile through settlement.
Alternatives and limits
Dealers can borrow Treasuries in private repo or securities-lending markets. SOMA lending is a public operational backstop-like source of temporary issue supply available only through the New York Fed’s primary-dealer programme terms.
The programme is not intended to guarantee any particular repo spread or eliminate all settlement scarcity.
How this connects to the surrounding knowledge estate
Repo pricing explains why exact Treasury CUSIPs become special. General securities lending provides the private-market analogue. SOMA lending adds central-bank inventory, public dealer limits and an issue-specific multiple-price auction.
Verification and update triggers
Preserve current New York Fed FAQ/terms version, SOMA holdings/custody state, minimum fee, theoretical-supply percentage, per-issue cap, total dealer cap, collateral margin, bid file, award file and fail status. Revalidate after programme-term changes, SOMA portfolio changes, collateral schedule updates or material Treasury settlement disruptions.
Primary and high-quality references
- Federal Reserve Bank of New York, Securities Lending.
- Federal Reserve Bank of New York, FAQs: SOMA Securities Lending Program.
- Federal Reserve Bank of New York, SOMA Securities Lending Program Terms and Conditions.
- Federal Reserve Bank of New York, Securities Lending Operation Results.
- Treasury Market Practices Group, current fails-charge trading practice referenced by the New York Fed.
Educational boundary: This article explains public Federal Reserve securities-lending mechanics. It does not advise bidding, repo trading or borrowing a particular Treasury security and does not provide personalized financial advice.