Guide

Fed balance sheet explained

Harbor Credit Union's asset-liability committee rehearsed every federal funds hike scenario for 2023 but treated the Fed's weekly H.4.1 statistical release as background noise. Then Treasury rebuilt its cash balance at the Fed — the Treasury General Account (TGA) rose by roughly $200 billion in two months while the policy rate sat unchanged. Bank reserves fell by a similar amount, SOFR–OIS spreads widened, and Harbor's wholesale funding desk paid 15 basis points more for overnight cash than the ALM model predicted from the dot plot alone.

The Federal Reserve balance sheet is the plumbing ledger behind modern monetary policy: what the Fed owns (mostly government securities), what it owes (reserves, currency, Treasury deposits, reverse-repo balances), and how those stocks move when officials run QE, QT, or adjust liquidity facilities. After Harbor rebuilt its ALM sleeve around H.4.1 line items and reserve-regime thresholds, funding-cost forecasts improved and the desk stopped treating balance-sheet mechanics as a footnote to rate decisions. This guide covers the balance-sheet identity, major asset and liability buckets, abundant versus scarce reserve regimes, the Harbor refactor, a technique decision table versus rate-only models, pitfalls, and a production checklist.

The balance-sheet identity

A central bank balance sheet must balance: total assets equal total liabilities plus capital (surplus). For the Fed, the simplified identity investors track weekly is:

Securities held (SOMA) + loans and other assets = Reserves + Currency in circulation + Treasury General Account + Reverse-repo (RRP) + other liabilities

Every dollar of Fed asset expansion creates a matching liability. When the Fed buys Treasuries in QE, it credits the seller's bank with reserves. When those securities mature during QT without reinvestment, reserves are extinguished. Shifts between liability buckets — reserves moving into the TGA when Treasury issues debt, or into the overnight reverse-repo facility when money funds park cash at the Fed — can tighten bank liquidity even when total assets are flat.

The Fed publishes this breakdown every Thursday in the H.4.1 “Factors Affecting Reserve Balances” release. Markets that watch only the policy rate miss half the liquidity story encoded in those tables.

Asset side: what the Fed owns

System Open Market Account (SOMA)

The SOMA portfolio is the dominant asset line. It holds U.S. Treasury securities across the maturity spectrum and, since 2008, agency mortgage-backed securities (MBS). SOMA size drives the Fed's influence on term premia and the duration the private sector must absorb. During QE, SOMA expanded from under $1 trillion pre-crisis to more than $8 trillion; QT and runoff programs shrink it passively as bonds mature.

Gold, SDRs, and foreign currency

Legacy asset lines — gold certificates, Special Drawing Rights, and foreign-exchange holdings — are small relative to SOMA. They matter for completeness when reading H.4.1 but rarely move markets.

Loans and liquidity facilities

In stress episodes the Fed activates discount-window lending, repo facilities, and emergency programs (as in March 2020). These spike temporarily on the asset side and can add reserves on the liability side. Monitoring facility usage signals funding stress before it appears in equity or credit spreads.

Liability side: where liquidity lives

Bank reserves

Reserve balances are deposits commercial banks hold at the Fed. They satisfy regulatory requirements, settle payments, and collateralize short-term funding. Post-2008 the U.S. system operates in an abundant reserves regime: reserves far exceed minimum requirements, and the Fed steers the policy rate with administered rates (interest on reserve balances, IORB) rather than by rationing scarce reserves.

Currency in circulation

Physical cash is a Fed liability. Currency growth is driven by public demand, not FOMC votes. Seasonal and crisis-driven spikes in cash usage drain reserves as the public swaps bank deposits for bills.

Treasury General Account (TGA)

The U.S. Treasury keeps its operating cash at the Fed. When Treasury issues debt and builds cash (TGA rises), reserves leave the banking system — a fiscal drain on liquidity. When Treasury spends down the TGA, reserves return. Debt-ceiling episodes that force TGA drawdowns can temporarily flood reserves; rebuild phases tighten them again.

Overnight reverse-repo facility (RRP)

The RRP absorbs cash from money-market funds and government-sponsored enterprises when safe private repo rates fall below the Fed's offered floor. High RRP usage signals abundant liquidity seeking a Fed backstop; RRP decline during QT can coincide with reserve scarcity as cash moves back into bank balance sheets or funds private repo markets.

Abundant vs scarce reserve regimes

The regime matters more than the absolute dollar level of reserves. In abundance, modest QT or TGA builds may barely move funding spreads. As reserves approach the lowest comfortable level (estimated by Fed staff and market participants, often discussed in FOMC minutes), money markets become elastic: SOFR–fed funds bases widen, repo fails rise, and banks hoard liquidity.

Signs of transition toward scarcity include:

Elevated SOFR–OIS or GC repo–OIS spreads on month-end and quarter-end.
Rising demand for Fed standing repo facility usage.
Bank bidding aggressively for Treasury bills as HQLA substitutes for reserves.
FOMC discussion of slowing or pausing QT to preserve buffer.

The 2019 repo spike and the 2024 QT slowdown both followed periods when aggregate reserves had fallen substantially while liability redistribution (TGA, RRP) masked the pace of drain. Harbor's ALM error was assuming abundance would persist because the policy rate had stopped rising.

How balance-sheet policy connects to markets

QE and QT

QE expands SOMA and reserves, compressing term premia and easing financial conditions beyond what rate cuts alone achieve. QT does the opposite — but not symmetrically, because fiscal issuance, RRP stocks, and market expectations interact. Balance-sheet size is a stock; rate policy is a price. Both matter.

Yield curve and portfolio effects

Fed holdings remove duration from private hands. Shrinking SOMA forces investors to absorb more Treasuries and MBS, often lifting long-end yields via higher term premia even when short rates are stable. ALM desks holding mortgage securities face additional convexity risk when the Fed's MBS portfolio runs off slowly due to low prepayment speeds.

Remittances to Treasury

The Fed earns interest on SOMA securities and pays interest on reserves and RRP. Surplus net income is remitted to Treasury; mark-to-market losses on securities acquired at low yields can temporarily reduce or eliminate remittances when rates rise — a political talking point but not a constraint on monetary policy operations.

Reading H.4.1 in practice

Focus on week-over-week changes in four lines:

Total assets / SOMA — QE purchases or QT runoff pace.
Reserve balances — banking-system liquidity stock.
TGA — fiscal flows in and out of the Fed.
RRP take-up — non-bank cash parked at the Fed.

A flat SOMA week with falling reserves and rising TGA is a liquidity drain driven by Treasury, not FOMC action. A falling SOMA with stable reserves may mean RRP is declining as cash migrates to banks — a different transmission path to watch in funding markets.

Harbor Credit Union ALM sleeve refactor (worked example)

Harbor's legacy ALM stack keyed off the financial conditions index and fed funds futures. H.4.1 parsing was manual and quarterly at best.

Refactor steps:

Automated H.4.1 ingest — weekly pull of reserves, TGA, RRP, and SOMA with four-week moving averages and z-scores versus 2019–2024 history.
Liability redistribution module — decompose reserve changes into QT runoff, TGA flow, RRP flow, and currency drift.
Reserve regime flag — tag weeks as abundant, transitional, or scarce based on spread and level thresholds.
Funding cost overlay — map regime flags to wholesale deposit beta and FHLB advance spread assumptions.
Scenario library — pair Treasury issuance forecasts with TGA path scenarios independent of FOMC rate decisions.
Committee dashboard — single slide comparing model implied reserves to actual H.4.1 each ALM meeting.

Outcome: the 2023 TGA rebuild was flagged three weeks before Harbor's wholesale desk widened its funding buffer; overnight borrowing costs came in within 3 basis points of forecast versus a 15 basis-point miss on the legacy rate-only model.

Technique decision table

Your goal	Prefer	Avoid
Forecast bank funding costs	H.4.1 reserve + TGA + RRP decomposition	Fed funds path alone
Estimate long-end yield pressure	SOMA runoff + net Treasury supply	Dot plot without QT caps
Stress-test liquidity	Scarce-reserve regime with spread shocks	Assuming infinite excess reserves
Explain sudden repo volatility	Liability redistribution week	Blaming only corporate tax dates
Retail portfolio timing	Balance sheet as background regime	Trading equities on every H.4.1 tick

Common pitfalls

Confusing stocks and flows — a one-week reserve drop is not permanent QT; check TGA and RRP.
Ignoring RRP as a liquidity buffer — RRP decline can offset reserve drain temporarily.
Treating pre-2008 models as current — scarce-reserve mechanics do not apply in today's floor system until reserves are genuinely tight.
Assuming QT is linear — caps change, MBS runoff is lumpy, and pauses are policy options.
Overweighting remittance headlines — deferred asset accounting does not stop open-market operations.
Single-point reserve threshold — scarcity is market-determined and state-dependent.
Neglecting currency in circulation — cash demand spikes drain reserves silently.
Reading H.4.1 without Treasury fiscal calendar — tax dates and debt-ceiling flows dominate many weeks.

Production checklist

Subscribe to weekly H.4.1 release with automated reserve, TGA, RRP, SOMA parse.
Maintain four-week and thirteen-week moving averages for key liability lines.
Tag weeks with fiscal events (tax receipts, coupon issuance, debt-ceiling milestones).
Track SOFR–OIS and GC repo spreads against reserve level and regime flag.
Model QT caps separately from Treasury net issuance in duration forecasts.
Document lowest-comfortable reserve estimate with explicit spread triggers.
Include balance-sheet scenarios in ALM committee decks alongside rate paths.
Cross-check Fed balance sheet with M2 and bank deposit data for consistency.
Review FOMC minutes for balance-sheet policy language each cycle.
Reconcile funding cost forecasts to actual spreads monthly.

Key takeaways

The Fed balance sheet links SOMA assets to reserves, currency, TGA, and RRP liabilities — every asset move has a liability counterpart.
Reserve scarcity is regime-dependent; TGA builds and RRP flows can drain bank liquidity without any rate change.
H.4.1 weekly data is the operational source of truth for balance-sheet plumbing, not the FOMC statement alone.
QE and QT operate on stocks; rate policy operates on prices — investors need both lenses.
Harbor Credit Union improved funding forecasts by decomposing liability redistribution instead of tracking only the policy rate.