Reader question: A bank can pledge Treasury securities, agency MBS, corporate bonds, mortgages and business loans to a Federal Reserve Bank. These assets have different market prices, maturities and credit risks. How does the Federal Reserve turn such a mixed collateral pool into one borrowing-capacity number?
The basic algorithm is:
Collateral Value = Estimated Fair Market Value × Applicable Margin.
The difficulty lies in both terms. Securities are generally valued using market prices from external vendors and receive margins based on asset type, credit rating and duration. Pledged loans are valued using information supplied by the bank—cash-flow and credit characteristics—together with market data and Reserve Bank valuation models. The margin then protects the Reserve Bank against the risk that collateral loses value during a liquidation period.
As of July 1, 2026, the Federal Reserve updated its collateral margins and its public display. The current methodology remains explicitly risk-sensitive: a short U.S. Treasury can receive a much higher percentage of market value than a long or riskier asset, and collateral with insufficient pricing or required information can receive zero collateral value.
What this page owns — and what it does not
This page owns:
pledged asset → eligibility → fair-value estimate → margin → lendable collateral value → aggregate borrowing capacity.
It does not replace repo pricing, daylight-overdraft algorithms, collateral optimisation, or the discount window’s credit-eligibility decision. This page explains how acceptable pledged assets are transformed into collateral value.
This is public central-bank collateral mathematics, not an assessment of any real bank or a recommendation to borrow from the discount window.
Start with legal and operational eligibility
Before an asset can contribute value, it must be acceptable collateral under Federal Reserve law, Operating Circular 10, Reserve Bank procedures and the relevant lending arrangement.
An asset can have a clear market price and still contribute zero to borrowing capacity if it is legally ineligible, improperly pledged, subject to an unacceptable lien or otherwise fails the Reserve Bank’s requirements.
The algorithm therefore begins with a gate:
Eligible and perfected collateral interest? → yes/no.
The distinction between market value and collateral value
Suppose a security has fair market value of $10 million.
If the applicable Federal Reserve margin is 95% of market value:
Collateral Value = $10m × 0.95 = $9.5m.
The economic haircut is:
Haircut = 1 − Margin = 5%.
It is important to read the Federal Reserve table correctly. The public table reports margins as percentages of market value. A 95% margin means the Reserve Bank lends against 95% of the estimated value, not that it removes 95%.
Why margins exist
If a borrower defaults, the Reserve Bank may need to realise collateral value after market conditions have changed.
A margin protects against:
- price volatility;
- credit deterioration;
- liquidity and liquidation time;
- valuation uncertainty;
- asset-specific risk.
The current Federal Reserve collateral page states that securities margins are designed to account for historical price volatility over liquidation periods determined by the Reserve Banks.
Current 2026 Treasury example
The Federal Reserve’s securities valuation and margins table effective July 1, 2026 shows U.S. Treasuries and fully guaranteed agencies with margins that decline as duration increases.
For the published duration buckets, Treasury and fully guaranteed agency margins are currently:
- 0–1 year: 99% of market value;
- >1–3 years: 99%;
- >3–5 years: 98%;
- >5–10 years: 97%;
- >10 years: 95%.
The table makes the rate-risk logic visible. Longer-duration securities can move more for the same yield shock, so the lendable percentage generally falls.
A duration example
Suppose two Treasury positions each have a market value of $20 million:
- Security A is in the 0–1 year bucket at 99%;
- Security B is in the >10 year bucket at 95%.
Collateral values:
A = 20 × 0.99 = $19.8m.
B = 20 × 0.95 = $19.0m.
Same issuer family, same market value, different borrowing capacity because interest-rate risk differs.
Securities valuation: external market prices first
The current Federal Reserve collateral page states that securities are generally valued using prices supplied by external vendors.
This creates a clean pipeline:
security identifier → external price → fair-market-value estimate → margin lookup → collateral value.
If no external vendor price is available under the current approach, the security receives zero collateral value.
This zero-value rule is important. A missing price does not mean the economic asset is worthless. It means the Reserve Bank will not assign lendable value under the stated collateral methodology when it lacks the required pricing basis.
Zero value is a risk-control state, not an accounting write-off
A bank can carry a bond at $5 million on its balance sheet while the Reserve Bank assigns zero discount-window collateral value because an acceptable price is unavailable.
Those statements are not contradictory. Accounting value and central-bank lendable value answer different questions.
Asset type changes the margin
U.S. Treasuries, agency securities, corporate securities, ABS, MBS and municipal securities have different historical volatility, credit and liquidity characteristics.
The Federal Reserve therefore does not use one universal haircut.
A margin engine typically keys off:
(asset class, rating/risk class, duration bucket, effective-date table).
This is a rule-table problem as much as a valuation problem.
Credit quality changes lendable value
For credit-sensitive securities, rating categories can affect margins.
The rationale is intuitive: a bond can lose value because benchmark yields rise, but it can also lose value because the issuer’s credit spread widens or default risk increases.
A collateral engine that uses duration but ignores credit class for a corporate security can materially overstate collateral value.
Loans cannot be valued like exchange-traded securities
A commercial loan or mortgage often has no continuously observable market price.
The Federal Reserve therefore estimates fair market value for pledged loans using internal valuation models informed by:
- cash-flow characteristics;
- credit characteristics;
- information supplied by the pledging institution;
- market data analysed by the Reserve Banks.
The current public methodology states that loan fair-market-value estimates are calculated on a recurring basis and that margins reflect cash-flow and credit characteristics plus historical price volatility over liquidation horizons.
Stylised loan valuation
A simple educational loan model is:
FMV = Σ Expected Cash Flowt × Discount Factort.
Expected cash flows can reflect:
- contractual principal and interest;
- prepayment;
- default probability;
- recovery assumptions;
- fees where relevant;
- remaining maturity.
The Reserve Banks’ production models are not public in full detail. A public article should explain the observable methodology without pretending to reconstruct proprietary valuation code.
Missing loan data can destroy lendable value
The Federal Reserve’s public loan-collateral guidance notes that it may make general assumptions where information is unavailable, but missing required information can also cause a pledged loan to receive zero collateral value.
This creates an important data-quality invariant:
economic loan exists ≠ sufficient collateral data exist.
A collateral system needs both the asset and the fields required to value and margin it.
Loan margins reflect two uncertainty layers
Loan collateral can lose value because:
- the borrower’s expected cash flows weaken;
- market discount rates change;
- the loan is difficult to sell or service during liquidation;
- valuation inputs are uncertain.
The fair-value model captures an estimate of current economic value. The margin then adds protection against future movement and liquidation uncertainty.
This is why:
Collateral Value ≠ Par × simple haircut
for many loan categories.
Monthly revaluation prevents stale borrowing capacity
Current Federal Reserve material describes recurring loan valuation based on updated institution and market data.
If a borrower deteriorates, a previously pledged loan can receive a lower fair-value estimate or margin. Collateral capacity can therefore fall even though the contractual principal balance has barely changed.
A bank relying on old collateral values can overestimate emergency borrowing capacity.
Aggregate borrowing capacity
For pledged assets i = 1…n:
Total Collateral Value = Σ FMVi × Margini.
Subject to legal, program and Reserve Bank constraints, that aggregate helps determine the amount of secured Federal Reserve credit the institution can support.
But the number is not always identical to “amount immediately borrowable.” Program eligibility, outstanding advances, daylight overdraft usage, operational readiness and Reserve Bank discretion can matter.
Collateral can support both discount-window lending and payment-system risk capacity
The Federal Reserve’s collateral valuation page states that assigned collateral value can be used to determine both discount-window borrowing capacity and overdraft credit capacity for Reserve Bank account activity.
This connects the collateral engine to the existing daylight-overdraft page.
One collateral pool can therefore matter across more than one central-bank credit mechanism.
Secondary credit can receive additional margin
The current Federal Reserve collateral page states that institutions borrowing under the secondary credit program generally face an additional margin on pledged collateral other than U.S. Treasury and Agency securities.
This is a program-level overlay:
base collateral value → additional secondary-credit adjustment where applicable.
An engine that calculates only the asset table and ignores the lending-program overlay can overstate capacity.
Operating Circular 10 is part of the control layer
Federal Reserve Operating Circular 10 contains the terms under which an entity may obtain advances, incur obligations to or pledge collateral to a Federal Reserve Bank.
The valuation table therefore does not stand alone. A production collateral system needs:
- asset valuation rules;
- margin tables;
- legal pledge/perfection status;
- program terms;
- effective-date controls.
A mixed collateral example
Suppose a bank pledges:
- $20m short Treasury at 99% margin;
- $15m long Treasury at 95% margin;
- $10m corporate security at an illustrative 85% margin;
- $25m loan portfolio with modelled FMV of $23m and illustrative 70% margin.
Collateral values:
19.8 + 14.25 + 8.5 + 16.1 = $58.65m.
Gross accounting principal can be much larger than lendable value. The algorithm converts heterogeneous assets into a common secured-credit measure.
The corporate and loan margins in this example are illustrative rather than claims about the current Federal Reserve table for a specific category.
Counterexample: the safest-looking par value can have lower collateral value
Suppose Loan A has par 100 but modelled fair value 80, while Security B has par 90 and market value 90.
Even if both receive the same 80% margin:
A collateral value = 64.
B collateral value = 72.
Par value is not the common basis. Fair value comes first.
Counterexample: more collateral does not always mean more capacity
If a bank pledges another security that cannot be priced under the current Federal Reserve approach, the asset can add zero collateral value.
Operationally, the collateral file becomes larger while borrowing capacity does not increase.
Counterexample: a price increase can be offset by a margin change
Suppose a bond rises 2% in market value but moves into a longer duration bucket or worse risk classification after a rules update.
The higher price can be partly or fully offset by a lower lendable percentage.
Collateral value depends on both fair value and the current margin table.
Counterexample: accounting-performing does not guarantee full collateral value
A loan can still be classified as performing while market spreads widen, expected cash flows weaken or required data are incomplete.
Central-bank collateral value can fall without the bank having charged the loan off.
Inputs and outputs
A Federal Reserve collateral engine can require:
- asset identifier;
- asset type;
- legal eligibility and pledge status;
- market price or loan valuation inputs;
- par/outstanding balance;
- credit rating or risk classification;
- duration/maturity;
- cash-flow characteristics;
- borrower credit characteristics for loans;
- current Federal Reserve margin table;
- lending-program type;
- effective date.
Outputs can include:
- fair-market-value estimate;
- applicable margin;
- collateral value;
- zero-value reason;
- program adjustment;
- aggregate collateral capacity;
- valuation date and provenance.
Evidence polarity: what supports confidence?
Evidence for a reliable collateral result includes security prices traceable to the approved valuation feed, asset classification matching the current margin table, loan fields complete enough for valuation, revaluation dates current, pledge status confirmed, and aggregate collateral value reconciling to Reserve Bank reports.
Evidence against confidence includes stale external prices, duration buckets derived from wrong dates, missing loan credit fields, a security with no approved price still receiving positive value, obsolete pre-July-2026 margins in a current run, or collateral value that cannot be reconciled to the Reserve Bank’s record.
Weak links in implementation
margin/haircut inversion. A 95% margin is interpreted as a 95% haircut.
stale table. Old margins remain active after a Federal Reserve update.
duration misclassification. A long security receives a short-duration margin.
rating mapping error. Security credit quality is mapped to the wrong table row.
price-source mismatch. Internal market price is used where the Reserve Bank assigns value from its own vendor feed.
loan-data sparsity. Required fields are absent but the internal system still assigns full lendable value.
double counting. One pledged asset appears in two collateral accounts.
program overlay omission. Secondary-credit adjustments are ignored.
Diagnostics: how to test the engine
- Treasury-duration test: use the July 1, 2026 99/99/98/97/95 published Treasury margins across duration buckets.
- margin arithmetic test: $10m × 95% must equal $9.5m collateral value.
- zero-price test: remove the approved external security price and require zero collateral value under the current public rule.
- loan-missing-data test: remove critical valuation fields and require conservative treatment or zero value.
- revaluation test: worsen loan credit data and verify fair value/margin response.
- effective-date test: switch from June 30 to July 1, 2026 and load the new margin table.
- duplicate-pledge test: one asset must not contribute twice.
- secondary-credit test: apply the additional program margin where required.
- aggregate reconciliation: sum asset collateral values independently and compare with total capacity.
- Reserve Bank reconciliation: compare institution-calculated lendable values with official collateral reports.
What would falsify confidence?
Confidence should be withdrawn if the institution cannot reproduce the Reserve Bank’s assigned values; if old margins remain active after an effective-date change; if zero-priced securities receive value; if loan valuation depends on unavailable fields without a conservative fallback; or if legal pledge status cannot be tied to the collateral record.
Alternatives and limits
Private repo lenders use their own collateral eligibility, pricing, haircuts and margining rules. A central counterparty uses another framework. The Federal Reserve discount-window system is institution-specific and governed by Reserve Bank terms and public central-bank rules.
The published tables are informative but do not supersede Reserve Bank requirements or create a binding entitlement to a particular advance amount.
How this connects to the surrounding knowledge estate
Repo pricing shows how private funding rates respond to collateral quality and scarcity. Collateral optimisation decides which eligible assets to deploy. Daylight-overdraft algorithms consume collateral capacity for payment-system risk. The discount-window collateral engine owns the central transformation from pledged assets to Federal Reserve lendable value.
Verification and update triggers
Preserve Operating Circular version, collateral-eligibility rules, valuation date, pricing source, loan-input schema, asset classification, margin table, program type and Reserve Bank reconciliation result. Revalidate after every Federal Reserve margin update, collateral-policy amendment, material market move, loan-data migration or borrowing-capacity discrepancy.
Primary and high-quality references
- Federal Reserve Discount Window & Payment System Risk, Collateral Valuation, last updated July 1, 2026.
- Federal Reserve Discount Window & Payment System Risk, Securities Valuation and Margins Table — effective July 1, 2026.
- Federal Reserve Discount Window & Payment System Risk, Updated Collateral Margins Including New Display Now in Effect.
- Federal Reserve Financial Services, Operating Circular 10 — Lending.
- Board of Governors of the Federal Reserve System, Discount Window Lending and related public lending framework material.
Educational boundary: This article explains public Federal Reserve collateral valuation principles. It does not determine the borrowing capacity of any real institution and does not provide legal, supervisory or personalized financial advice.