U.S. IORB–SOFR Spread

The reserve-conditions spread gauges, in a single number, how ample or scarce the reserves of the U.S. banking system are. It is computed in basis points by subtracting the secured overnight rate at which cash is borrowed against Treasury collateral from the administered rate the Federal Reserve pays on reserve balances:

Here $\text{admin}_t$ is the Interest Rate on Reserve Balances (IORB) or its predecessor administered rate, and $\mathrm{SOFR}_t$ is the Secured Overnight Financing Rate (SOFR).

A wide positive spread indicates the normal regime in which reserves are ample and the administered-rate floor is effective, whereas a spread that approaches zero or turns negative indicates a regime in which secured funding has become scarce relative to reserves. A lower reading therefore signals tighter funding conditions, the opposite of the intuition for most rate spreads, where a widening spread signals stress.

The spread is a direct arithmetic transform of two daily published rate series, with no estimation.

(1) Administered-rate chaining. The Interest Rate on Reserve Balances (IORB) exists only from 2021-07-29, the date on which a single rate replaced the separate Interest Rate on Excess Reserves and Interest Rate on Required Reserves (Board of Governors of the Federal Reserve System 2021). To extend the spread back to the SOFR start in April 2018, the predecessor administered rate IOER is spliced on before that date to form a continuous administered rate $\text{admin}_t$:

IOER was paid on excess reserves whereas IORB is paid on all reserve balances, so the two are not the same instrument, and they are linked only because each served as the administered rate at the top of the corridor during its own regime. This is standard practice, and the pre-2021-07-29 segment is accordingly labelled as IOER rather than presented as if IORB had existed before that date.

(2) Spread in basis points. Both the administered rate and SOFR are reported in percent, so the two are inner-joined on common dates and their difference is multiplied by one hundred to express the spread in basis points. The inner join makes the series begin at the SOFR start and leaves no gaps.

(3) Sign convention. A positive spread means SOFR trades below the administered rate, the normal ample-reserves regime in which the floor is effective. A spread that narrows toward zero or turns negative means SOFR has risen to or above the administered rate, which is the scarce-reserves signal the series exists to surface. Lower readings therefore indicate tighter funding conditions, the opposite of the reading for most rate spreads.

This spread reads in the opposite direction from most rate spreads, since a lower value rather than a wider one signals tighter conditions. Once paying interest on reserves became the primary tool of monetary-policy implementation, the administered rate took on the role of a floor for the policy corridor (Board of Governors of the Federal Reserve System 2008), and whenever reserves are ample the overnight market rate settles at or slightly below that floor (Bech and Klee 2011). A positive spread reflects this cushion of ample reserves, whereas a spread that approaches zero or turns negative signals that secured funding has become scarce relative to the reserves the banking system holds (Ihrig, Senyuz, and Weinbach 2020).

The canonical illustration is the money-market disruption of mid-September 2019. Corporate tax payments and Treasury settlement drained reserves below the level the system needed, and SOFR spiked far above the administered IOER rate so that the spread turned sharply negative (Afonso et al. 2021; Anbil, Anderson, and Senyuz 2020). Copeland, Duffie, and Yang (2021) argue that the episode revealed reserves to have been less ample than supposed, and because the distribution of reserves across banks mattered as much as the aggregate, the spread can compress abruptly once the system approaches the point of scarcity.

Monitored over time, the spread offers a market-based gauge of where the banking system sits on the reserve-demand curve. A wide, stable positive spread is consistent with abundant reserves and an effective floor system, while a persistent drift toward zero is the kind of early warning that informed the Federal Reserve's decisions to slow and then halt balance-sheet runoff. Because the indicator is built from administered and secured benchmark rates rather than from any model, it is best read alongside the U.S. net liquidity series (USNETLIQ) and reserve-balance measures (USFEDRSV) to distinguish a genuine scarcity signal from transient month-end or quarter-end frictions.

For positioning, the sign convention is what matters first. A spread that compresses toward zero or turns negative is a funding-tightness signal that tends to coincide with upward pressure on repo and other secured overnight rates, so a falling reading is the cautionary one rather than the reassuring one. The September 2019 episode showed how quickly the spread can move once reserves approach scarcity, and Copeland, Duffie, and Yang (2021) emphasize that such moves can be abrupt and nonlinear (Afonso et al. 2021).

Desks that fund positions in the repo market read the spread as a proxy for the cost and reliability of secured overnight funding. A wide positive spread indicates that cash is plentiful and that financing positions against Treasury collateral is cheap relative to the administered floor, while a compressing spread warns that financing may become more expensive and more volatile, which bears directly on the carry of leveraged Treasury and basis trades.

Because the indicator is a clean, model-free read on dollar funding conditions, it is also useful as a global risk barometer for non-U.S. investors, including those positioning in Korean rates, since stress in U.S. secured funding markets transmits through cross-currency bases and global dollar liquidity. It is best used together with the U.S. net liquidity series (USNETLIQ) and the Federal Reserve balance-sheet components, with the reading scaled to reflect that month-end and quarter-end dates routinely produce transient spikes that do not signal a durable change in reserve conditions.

Over 2044 observations from 2018-04-03 to 2026-06-09, the unit-root reading of the IORB-minus-SOFR reserve-conditions spread is ambiguous, since the Dickey and Fuller (1979) and Phillips and Perron (1988) tests reject a unit root at p = 0.0081 and p = 0.0 while the Kwiatkowski et al. (1992) test also rejects stationarity at p < 0.01, the near-unit-root disagreement of a persistent bounded signal, so no clean order is assigned. On the level the first-order autocorrelation is 0.560 with an implied half-life of about 1.2 observations, a descriptive persistence summary and not an integration claim, the Ljung and Box (1978) portmanteau on the level rejects white noise at lags 10 and 20, Q = 3500.18 and Q = 6022.75 at p = 0.000 and p = 0.000, and the Bai and Perron (1998) procedure, by the Bai and Perron (2003) algorithm, finds 4 breaks in the mean at 2019-12-03, 2021-02-24, 2023-10-31, 2025-03-19 (Andrews 1993; Perron 1989).

This signal, the IORB-minus-SOFR reserve-conditions spread, has no daily seasonal period, so the seasonal-unit-root and seasonal-stationarity machinery of Hylleberg et al. (1990) and Canova and Hansen (1995) is deliberately not run (Beaulieu and Miron 1992; Ghysels and Osborn 2001).