At the discount window, $100 of collateral is rarely $100 of cash

The Federal Reserve Bank of New York at 33 Liberty Street in December 2025. The building is the public face of a backstop whose usable liquidity is determined behind the scenes by agreements, collateral valuation, and operational readiness. Photograph by Kidfly182, Wikimedia Commons, CC BY 4.0.[10]

Priced: a bank with a large securities or loan book has a large emergency-liquidity buffer. New: the Federal Reserve's collateral tables, updated on July 1, show why that shortcut can fail. The discount window lends against lendable value—market value after a category-specific margin—not face value, accounting value, or the amount printed in an investor presentation.[1][2]

That distinction is the live gap in bank-liquidity analysis. A Treasury can be moved and valued quickly but still receive less than its market value. A loan can be economically sound yet convert into much less cash because its maturity, coupon, repayment structure, and risk rating all affect the margin. And an eligible asset produces no same-day liquidity if the legal agreement, lien, data file, custodian movement, or test transaction is not ready.

One asset, three values

Bank balance sheets encourage three numbers to blur together:

Face or carrying value is the accounting starting point.
Market or model value is what the Federal Reserve estimates the collateral is worth now.
Lendable value is the second number multiplied by the applicable margin.

For securities, the Reserve Banks generally start with fair-market-value estimates supplied by external vendors. If no vendor price is available, the security receives zero collateral value. They then apply a margin intended to absorb historical price volatility over an assumed liquidation period. Asset type, credit rating, and duration determine the result.[2]

The mechanism is simple even when the table is not:

face value → current valuation → Federal Reserve margin → lendable value → available borrowing capacity.

Every arrow can shrink the answer. A bond bought at par may trade below par after rates rise. A longer-duration security receives a lower percentage of that reduced market value. A loan pool may require more data and time to pledge than a Treasury. An asset already pledged elsewhere is not a spare backstop merely because it remains on the balance sheet. Interagency liquidity guidance therefore tells banks to monitor collateral by legal entity, jurisdiction, currency, encumbrance, and operational location—not as one undifferentiated pile.[5]

Five numbers that constrain the read

The July tables make the conversion visible.[2]

A Treasury with up to one year of duration receives 99% of market value.
A Treasury with more than ten years of duration receives 95% of market value.
A fixed-rate, fully amortizing first-lien mortgage with 30 years to maturity receives 67% in the displayed in-scope schedule.
Primary credit was priced at 3.75% on July 23, only 12 basis points above the 3.63% effective federal-funds rate.[6]
In a Federal Reserve study, pre-pledged securities increased the likelihood of primary-credit use after a reserve shock by 10.5 percentage points for institutions starting with below-median reserve-to-asset ratios.[4]

The first three numbers are capacity; the fourth is price; the fifth is evidence about execution. They should not be combined into a single score.

Consider a long Treasury whose face value still reads 100 but whose current market value has fallen to 85. At a 95% margin, its lendable value is 80.75—not 100. The margin did not create the first loss of capacity; the market price did. It then added a second buffer. Conversely, a short Treasury near par can convert almost dollar for dollar because both valuation and margin are favorable.

Loans widen the dispersion. The new presentation shows minimum, weighted-average, and maximum margins observed on historically pledged loans, but the Fed warns that the actual treatment of a particular loan can differ. The displayed ranges are informative, not binding quotes.[1][2] A bank cannot safely turn a category average into a committed line.

Pre-pledged does not mean pre-funded

The Federal Reserve's readiness sequence has three separate gates. An institution needs signed borrowing documents with its regional Reserve Bank. It needs eligible collateral pledged and valued. It also needs people and systems able to request, receive, reconcile, and repay an advance under stress. The Board explicitly encourages periodic transactions during normal times because a dormant contingency line can fail at the operational step.[3][5]

That is why the aggregate amount of pledged collateral is reassuring but incomplete. The Board's readiness data show that loans make up the larger share of lendable value already positioned at the window.[3] That is sensible: marketable securities are easier to sell or repo elsewhere, while loans benefit more from being prepared in advance. It also means the system depends on slower, data-heavy collateral whose value is more institution-specific.

The Fed's own research supports the operational argument without proving that pre-pledging alone causes resilience. Across roughly 80,000 observations in which a depository institution experienced a large one-day reserve decline, about 4% involved discount-window borrowing. Having collateral in place the prior day was associated with a higher probability of borrowing, and the relationship was strongest when starting reserves were relatively scarce.[4] Preparedness appears to make the backstop more usable precisely when timing matters most.

Yet use and readiness are not the same signal. Zero borrowing can mean a bank is liquid, market funding is cheaper, or managers remain reluctant to approach the window. Heavy borrowing can mean the facility is working as designed or that private funding has already failed. The H.4.1 balance is therefore a stress thermometer, not a public inventory of every bank's spare capacity.

The strongest counterweight: the haircut is meant to be conservative

It would be equally wrong to read every gap below par as hidden insolvency. Discount-window margins are protection for the Reserve Bank in a secured loan, not a forecast that the collateral will lose the withheld amount. A long mortgage can receive a lower lendable percentage because it is difficult to value and liquidate under stress even when the borrower remains current. The asset may ultimately pay in full.

Nor is the window currently priced like an obviously punitive source of cash. Primary credit sat close to the effective federal-funds rate on July 23.[6] For an eligible institution facing a temporary outflow, secured central-bank funding can be economically cleaner than dumping assets into a falling market or cutting credit to customers.

The appropriate conclusion is narrower: the discount window can stop a liquidity problem from becoming a fire sale, but it cannot make one dollar of gross assets equal one dollar of immediately usable cash. It also cannot repair weak capital, poor asset quality, or a deposit franchise that keeps shrinking. Those are solvency and business-model questions, not collateral-operations questions.

Falsifier

The thesis is that gross pledged collateral overstates decision-ready liquidity when the valuation and execution layers remain undisclosed. That reading is falsified for a bank if a recent, documented test shows that its unencumbered collateral—marked under a severe but plausible rate and credit shock, run through the Federal Reserve's applicable margins, and moved through the correct legal entity—still produces enough lendable value to cover its modeled cash outflows on time.

If banks begin publishing that post-haircut, stress-valued, operationally tested capacity in a consistent form, and it remains stable as market values move, the hidden conversion gap largely disappears. Until then, investors should treat “collateral available” as the start of the liquidity calculation, not its answer.

Watchlist

July 29 — FOMC decision: watch the primary-credit rate alongside the federal-funds target. A rate change alters the cost of using the window; it does not change the collateral conversion mechanics.[6][7]
July 30, 4:30 p.m. ET — H.4.1 release: watch primary-credit balances, but resist reading a quiet weekly number as proof that every institution is operationally ready. The release measures borrowing outstanding, not unused lendable value.[8]
July 30 — second-quarter Call Reports due from most banks: use the new balance-sheet data to screen securities duration, loan mix, pledges, and funding concentration, then apply the July margin logic. Call Reports still will not provide a standardized bank-by-bank measure of tested discount-window capacity.[2][9]

The July update is not a signal that the Federal Reserve suddenly became less willing to lend. It is a clearer ruler. The ruler shows that liquidity is a conversion process: value the asset, apply the margin, clear the lien, move the collateral, test the line. A bank that has completed those steps owns a real option. A bank that has merely counted the assets owns a spreadsheet.

cronfeed.work