When “Cash” Stops Being Cash

It is a quiet day on the screen.

Stocks barely move. The index closes near flat. Credit looks steady. Nothing feels urgent.

Then a different number hits.

Overnight repo rates jump. Not by a hair. By enough to stand out. The move looks wrong next to the calm tape. If “cash” is everywhere, why does the cost of borrowing it spike?

That mismatch is the clue. Repo can scream while stocks whisper, because repo is not mood. It is plumbing.

The usual reasons sound smooth, but they do not reach the pipe.

“Risk-off” is one. Yet risk markets did not sell. “People wanted cash” is another. Yet cash-like assets did not look scarce. Sometimes the blame lands on “a lack of reserves.” That can matter, but it often explains the season, not the moment.

Repo spikes can appear without headlines. They can fade fast. They can show up on days when nothing else moves.

If the spike were mainly fear, it would leak into prices across markets. If it were mainly opinion, it would drift as opinion does.

Instead, repo often moves like a valve. It runs fine, then it doesn’t. That pattern points to constraints. Something hits a limit, and the price jumps to clear the flow.

A repo is a short loan that uses securities as collateral.

One side gives cash and takes a bond. The other side gives a bond and takes cash, with an agreement to reverse the trade soon, often the next day. The bond is the safety. The rate is the price.

Most repo runs through dealers. Dealers sit in the middle. They borrow cash from cash-rich lenders, then lend cash to cash users. They also hold big piles of bonds to make markets.

That middle role is balance-sheet heavy. A balance sheet is the firm’s inventory of assets and the funding used to carry them. Even a “safe” bond takes space on it.

Two limits shape the repo more than most stories admit.

Leverage limits cap how large a dealer’s balance sheet can grow, even if the assets are low risk. Liquidity limits push dealers to hold cash and avoid funding that looks fragile. These limits come from rules, supervisors, and internal risk models. They do not need panic to bind.

Repo also depends on collateral velocity. That is how fast the same bond can support funding as it moves through the system.

The tool that raises velocity is rehypothecation. Rehypothecation means a firm that receives collateral can pledge it again in its own borrowing. The same bond can back more than one cash loan as it passes along.

When velocity is high and balance sheets are open, repo feels like a utility. It feels like cash.

When either one tightens, repo stops feeling like a utility.

A repo spike is often the system admitting one fact: there is not enough usable balance sheet at the old price.

That can happen even when the bond itself is “cash-like.”

“Cash-like” is not the same as “ready to finance”

A Treasury is high quality. But in repo, quality is not the only input. The dealer still must take the bond, book the trade, and fund it overnight. That uses the balance sheet.

When a dealer is near its leverage or liquidity edge, it rations. Not because of a new belief about the bond. Because the firm cannot expand the book without breaking a limit.

Rationing shows up as a higher rate. It can also show up as a bigger haircut.

A haircut is the discount applied to collateral value in a repo loan. A 2% haircut means $100 of bonds raises about $98 of cash. When haircuts rise, borrowers must post more bonds for the same cash. That turns “safe collateral” into “not enough collateral,” fast.

Rehypothecation is a chain, and chains slow under load

In smooth conditions, collateral moves quickly.

A dealer takes in a Treasury from one trade, then re-uses it to secure its own borrowing, then uses the cash to finance another client. Netting and clearing reduce how much cash must move. The system leans on reuse.

When the balance sheet is tight, that reuse slows. Dealers keep more collateral idle, because re-pledging it creates extra exposures, extra settlement needs, and extra intraday risk. The same stock of Treasuries is still in the system, but it turns fewer times per day.

That is a key shift. The squeeze can be about velocity, not about total supply.

Think of it as traffic. The number of cars has not changed. The road has narrowed. A jam forms.

The “right bond” can matter more than “a bond”

A repo is not always one pool. Some collateral is more demanded than others.

General collateral is the broad set of acceptable bonds used in plain funding. A “special” is a specific bond that is hard to borrow, often because it is needed to settle trades or cover shorts.

When a bond goes special, the repo rate for that bond can move in strange ways. The rate becomes a price for access to that exact security, not just a price for cash.

If settlement needs rise and dealer’s balance sheet is already full, specials can pull hard on the system. Firms pay up for the bond they must deliver, even if equities are flat.

Intraday liquidity turns timing into stress

A repo is “overnight,” but the stress can be intraday.

Large settlement systems move cash and bonds on a clock. Cash must arrive at certain hours. If it arrives late, firms draw on lines, hold buffers, and delay other payments. That behavior is defensive, but it is also mechanical.

Dealers manage daylight risk limits, too. A book that looks fine at the end of the day can look risky at noon, when payments are still pending. To avoid daylight breaks, firms hoard cash. That cash is then not lent into repo. The price rises.

None of this requires a big risk move in stocks. It requires only that the balance sheet, collateral reuse, and timing needs collide.

So the repo spike is not a mystery bolt from the sky. It is the clearing price when the system has less capacity to intermediate than the flow demands.

This framework explains repo spikes that come from tight balance sheets, lower collateral velocity, and settlement timing pressure. It also explains why the move can be sharp. Thresholds matter in constrained systems.

It does not explain every spike.

Repo can jump because of one-off cash drains, large bill payments, tax dates, or sudden shifts in central bank operations. It can jump because of a settlement fail, a clearing issue, or a concentrated position in a specific bond. In those cases, the “pipe” may be blocked by a local break, not a system-wide squeeze.

It also depends on today’s market layout, where dealers are key and leverage rules bind at the margin. If the structure changes, the pattern changes.

Still, the core point holds across many calm-day spikes: in repo, “cash” is not a thing sitting in a pile. It is a service provided through balance sheets.

When that service runs short, the price can jump even while everything else looks quiet.