Guide

Fed discount window explained

Harbor Credit Union's $4.8B balance sheet ran a liquidity stress drill in February 2026 after a peer regional bank reported accelerated uninsured deposit outflows. The treasury team's model tracked fed funds, SOFR-OIS, and SRF take-up — but not discount-window borrowing. When weekly H.4.1 data later showed primary-credit usage jumping from near zero to $8.4B across the banking system, Harbor's week-ahead funding-cost forecast missed by 14 bp. The desk had treated widening TED-style spreads as a tradeable blip rather than a signal that banks were tapping the Fed's direct lending backstop.

The team added primary-credit outstanding, discount-rate spreads over interest on reserve balances (IORB), and same-day borrowing velocity to their daily dashboard alongside bank reserves and commercial paper spreads. Forecast error on wholesale funding costs fell from 14 bp to 4 bp over Q1–Q2 2026. This guide explains what the discount window is, how primary credit differs from secondary and seasonal programs, how the discount rate relates to IORB and the fed funds target, why borrowing stigma persists, how discount-window stress differs from repo-market stress, the Harbor Credit Union refactor, a technique decision table, pitfalls, and a production checklist for bank treasury and macro desks.

What the discount window is

The discount window is the Federal Reserve's direct lending facility for depository institutions. Banks borrow from their regional Federal Reserve Bank, posting collateral (typically loans and securities) and paying an administered rate called the discount rate. Unlike open-market operations that move reserves through primary dealers, discount-window lending is bilateral: any eligible institution can request credit when it needs reserves or liquidity.

Modern discount-window policy centers on three programs:

Primary credit — for financially sound institutions with short-term funding needs. No questions asked about why you borrow; terms are overnight (extendable to 90 days in stress). This is the program macro desks watch during bank stress episodes.
Secondary credit — for institutions with severe financial difficulties. Supervisory approval required; priced above primary credit. Signals distress, not routine liquidity management.
Seasonal credit — for small banks with predictable seasonal deposit swings (e.g. agricultural cycles). Limited eligibility; rarely relevant to system-wide stress monitoring.

The Fed publishes aggregate discount-window borrowing in the weekly H.4.1 release (typically Thursday afternoons). Primary credit outstanding is the headline number; spikes from near-zero baselines are among the clearest real-time signals that banks are struggling to fund in private markets.

Discount rate vs IORB vs fed funds

Three administered rates anchor short-term bank funding, and confusing them leads to bad forecasts:

Federal funds target range — the FOMC's policy rate. Overnight unsecured lending between banks typically trades inside this band.
Interest on reserve balances (IORB) — the rate the Fed pays on excess reserves. Acts as a soft ceiling on fed funds in abundant-reserve regimes: banks won't lend below what they earn sitting at the Fed.
Discount rate (primary credit) — traditionally set above IORB, creating a penalty spread that discourages routine discount-window use. Post-2003 reforms aimed to make primary credit a backup, not a first resort.

The penalty spread matters for monitoring: when private funding costs spike above the discount rate, borrowing at the window becomes economically rational even before a bank is insolvent. A widening gap between market unsecured rates (fed funds effective, LIBOR successors, or TED-style proxies) and the discount rate often precedes visible discount-window take-up by days.

Contrast with secured funding: the Standing Repo Facility rate caps repo borrowing for dealers and clearing banks. Discount-window stress and repo stress can diverge — March 2023 saw massive primary-credit usage while Treasury repo markets remained orderly for well-collateralized borrowers.

Why stigma persists

Economists describe discount-window borrowing as carrying stigma: banks avoid the window because markets interpret usage as a sign of weakness, which can accelerate deposit flight and counterparty pullbacks. Historical episodes (pre-2003 when the window was often the cheapest source of funds, and 2008 when stigma prevented timely borrowing) motivated the primary-credit framework.

Stigma creates a monitoring paradox for macro desks:

Low reported borrowing may mean banks are healthy or they are desperately avoiding the window while paying punitive rates elsewhere.
Sudden spikes often mean stigma broke — private markets closed or priced borrowing above the discount rate, making the window the least-bad option.
Regional concentration in H.4.1 data (which district borrowed how much) can signal idiosyncratic bank stress vs broad system pressure.

Harbor Credit Union's refactor treated discount-window data as a lagging confirmation indicator, not a leading one. Leading signals included uninsured deposit ratios, held-to-maturity securities losses vs capital, FHLB advance growth, and CP/CD spreads for bank names. The window print confirmed when those signals aligned.

Discount window vs repo and SRF

Treasury and liquidity desks need a clear map of Fed backstops:

Facility	Borrower	Collateral	Typical stress signal
Primary credit (discount window)	Depository institutions	Broad loan/security collateral	Bank funding stress, deposit outflows
Standing Repo Facility	Primary dealers, clearing banks	Treasuries, agencies, MBS	Repo market dysfunction, dealer balance-sheet limits
Fed funds / reserves	Interbank (market)	Unsecured	Reserve scarcity, quarter-end reporting
FHLB advances	Member banks/credit unions	Member collateral	Regional bank liquidity (often pre-Fed window)

A bank can face deposit stress (discount window) while repo markets for high-quality collateral remain calm (no SRF take-up). Conversely, September 2019 repo turmoil spiked GC rates and SRF-style interventions without broad discount-window usage because the problem was securities-dealer balance sheets, not retail deposit runs. Monitoring both legs of the plumbing is non-optional.

Harbor Credit Union refactor

Harbor's liquidity desk rebuilt its early-warning stack after the February drill miss:

Daily H.4.1 primary-credit line — level, week-over-week change, and share of total Fed assets. Alert threshold: > $2B system-wide when baseline < $500M.
Penalty-spread monitor — fed funds effective plus 25 bp vs discount rate; alert when market rate approaches discount rate from below.
Cross-asset confirmation — bank CP spreads, KRE equity moves, and FHLB advance proxies must co-move before upgrading stress regime from “watch” to “active.”
Regional peer screen — match H.4.1 district borrowing against uninsured-deposit-heavy peer list; flag when peer geography borrows disproportionately.
Funding plan playbook — pre-approved wholesale funding lines and securities-sale ladders triggered by regime shift, not by discount-window headlines alone.

Week-ahead forecast error on Harbor's marginal funding cost fell from 14 bp to 4 bp. False-positive stress alerts dropped from 6 per quarter to 1 because the desk stopped treating every TED wiggle as a crisis.

Technique decision table

Monitor / tool	Use when	Skip when
Primary-credit outstanding (H.4.1)	Banking-sector stress, deposit-flight headlines, regional-bank equity selloffs	Pure dealer-repo or MMF reform stories with no bank liability stress
Discount rate vs IORB spread	Assessing whether window borrowing is economically attractive	Secured-funding-only stress (watch SRF/repo instead)
TED / unsecured CP spreads	Leading signal before window usage prints	Post-usage confirmation only (too late as sole trigger)
SRF take-up	Treasury repo specials, dealer balance-sheet constraints	Retail deposit run narratives without repo market disruption
FHLB advance growth	Regional bank liquidity; often precedes discount window	Money-center banks with deep CP programs only
Bank reserves level	System-wide liquidity abundance vs scarcity	Idiosyncratic single-bank stress with ample system reserves

Default for multi-asset treasury desks: run unsecured bank stress (TED, CP, discount window, FHLB) and secured market stress (SOFR-OIS, SRF, GC repo) as parallel tracks. Merge only when both fire or when Fed balance-sheet flows ( TGA, QT) explain the divergence.

Common pitfalls

Weekly frequency blindness — H.4.1 is weekly; intraday stress can peak and fade between prints. Pair with daily market spreads.
Stigma underestimation — assuming banks will borrow as soon as spreads widen; they often delay until private markets shut.
Conflating discount window with SRF — different borrowers, collateral, and stress channels.
Ignoring FHLB — regional banks often tap Home Loan Bank advances before the Fed window; missing FHLB growth delays detection.
Single-threshold alerts — $1B borrowing means different things in 2019 vs 2023 vs abundant-reserve 2026; normalize to GDP or bank assets.
Secondary credit noise — small secondary-credit lines in H.4.1 signal supervisory distress, not routine liquidity; do not blend with primary-credit trends.
Discount rate calendar lag — the Board sets the discount rate; it can trail FOMC moves. Recompute penalty spreads after each FOMC week.

Production checklist

Pull H.4.1 primary-credit outstanding every Thursday; archive week-over-week delta.
Chart discount rate, IORB, and fed funds effective on one axis; alert on penalty-spread compression.
Run parallel unsecured (TED, bank CP) and secured (SOFR-OIS, SRF) stress panels.
Track FHLB advance growth for regional-bank peer set.
Define regime states: normal, watch (spreads widen), active (window usage or market shut).
Pre-write funding contingency playbooks per regime; do not improvise during headlines.
Separate primary-credit from secondary-credit in dashboards.
Normalize borrowing levels to historical percentiles, not fixed dollar thresholds.
Cross-check discount-window spikes against equity (KRE) and CDS if available.
Re-evaluate thresholds after QT pace, TGA, or reserve-scarcity regime changes.

Key takeaways

The discount window is the Fed's direct unsecured lending backstop for banks — primary credit is the macro-relevant program.
Discount rate sits above IORB by design; when market rates approach it, window usage often follows.
Stigma delays borrowing — low usage does not prove calm markets; spikes confirm stress that often started days earlier.
Discount-window stress and repo/SRF stress are different animals; monitor both tracks.
Harbor Credit Union cut funding forecast error 14→4 bp by pairing H.4.1 data with leading spread indicators.