What makes a repo rate spike, and how do funding spreads warn you?

Repo rates move for two completely different reasons that look similar on a headline and mean opposite things. Learning to tell them apart is the single most useful skill in this unit.

Cause one: cash is scarce (a general funding spike)

Everyone wants cash and there is not enough of it. Reserves have drained — a quarterly tax date, a heavy settlement day, a quarter-end when dealers shrink balance sheets for regulatory reporting — and the price of cash rises for everybody.

Signature: general collateral repo rates rise together, SOFR rises relative to unsecured rates like EFFR, and the whole complex pushes toward or through the top of the policy target range. This is a cash story.

Cause two: one bond is scarce (a specials squeeze)

Now the mirror image. Suppose a particular Treasury issue is heavily shorted. Anyone short must borrow that specific bond to deliver it, and repo is how you borrow a bond: you lend cash and take that bond as collateral. When demand for one issue is intense, cash lenders compete by accepting a lower interest rate on the cash they lend, just to be given that bond.

So the repo rate on that issue falls below the general collateral rate — sometimes to zero, occasionally negative. The issue has "gone special." In the extreme, the bond cannot be sourced at all and trades fail to deliver.

Note the direction reversal: a cash shortage pushes repo rates up; a collateral shortage pushes the affected issue's repo rate down. Same market, opposite signs, different causes.

The arithmetic of specialness. General collateral trades at 4.30%; a heavily shorted on-the-run issue trades special at 3.30%. The specialness is 100 basis points. A holder of that bond can lend it out and fund itself 100bp cheaper than everyone else. On $100 million for 30 days:

100,000,000 × 0.0100 × 30 / 360 = $83,333

That is what bond scarcity is worth, in cash, per month. It is also why a bond that is expensive to borrow is expensive to be short.

Reading the gauges

Money markets are the shortest-dated, highest-turnover claims in the system. Everything rolls daily, so a loss of confidence appears immediately as a refusal to roll rather than as a slow drift in price. That is why funding spreads are watched as an early-warning gauge: they move before slower markets have finished noticing.

The spreads watched most closely map directly onto the two causes above:

• EFFR minus SOFR — unsecured minus secured. Which of the two sits higher is not a fixed fact: EFFR ran a few basis points above SOFR through 2021–2023, when reserves were abundant, while since mid-2024 SOFR has fixed at or above EFFR on most days as collateral supply grew. The ordering is structural rather than a credit judgement — EFFR is a thin market arbitraged close to what the Fed pays on reserves, while SOFR is a multi-trillion-dollar repo market whose level answers to collateral supply and dealer balance sheet capacity. So read this spread against its own recent range and watch for a change in it. A negative print is not by itself a stress reading.
• SOFR against the bottom and the top of the target range — where the secured rate sits inside the policy range. Drifting toward the top of the range is the classic cash-scarcity tell.
• TGCR minus BGCR — two general collateral rates across different venues and collateral mixes. Normally near-identical; a gap says the venues are pricing differently.
• OBFR minus EFFR — a broad unsecured measure against the interbank one.

Two disciplines go with these. First, distinguish calendar effects from stress: quarter-ends and year-ends produce reliable, benign spikes, because European banks report balance sheets on the final day and dealers step back. Second, and more important: these are gauges to read, not signals to act on. A widening spread tells you that somebody is paying up for cash today. It does not tell you what any policymaker will do, and this course never treats it as though it does.

In the data

The range gauge is one subtraction at each edge. Here it is on an ordinary day, with both sides of each subtraction:

On 24 September 2026 SOFR fixed at 3.88%, 13 basis points above the bottom of the 3.75–4.00% range and 12 below the top: close to the middle. On ordinary days both gaps sit in the single or low double digits, which is what makes a reading in the hundreds legible without any further context. A cash squeeze shows up as the first number growing and the second shrinking toward zero.

Try it now

1. Below are SOFR minus the bottom of the range, and EFFR minus SOFR, on the same twenty-one days, picked from the six months to 24 September 2026. Mark every quarter-end, and the two US tax dates, 15 April and 15 June.

2. Separate the calendar spikes from everything else. Which rows fall on no month-end, quarter-end or tax date, and did SOFR move up or down on those days?
3. Write one neutral sentence describing the largest move you found — what happened to the price of cash, and nothing about what it implies.