The World's Biggest Pawnshop

At ReadOn, we don’t just report the markets. We help you understand what truly drives them, so your next decision isn’t just informed, it’s intelligent.

Join ReadOn's WhatsApp Community

On December 31, 2025, borrowing cash overnight against US government bonds, the safest collateral on earth, cost as much as 4.0%. That was well above where the US Federal Reserve wanted it.

Nobody defaulted that day. Nothing got downgraded.

Banks simply borrowed $75 billion from the Fed at 3.75% and lent it straight back into the market at a higher rate. Free money, for one night.

Why? Because US government bonds could not find a place to sleep.

That sounds absurd, but it happens more often than you’d think.

The market we’re talking about is called repo, short for repurchase agreement. Think of it as a one-night pawnshop for bonds.

You own a government bond. You need cash today. So you hand the bond to someone, take their cash, and promise to buy the bond back tomorrow at a slightly higher price. That small extra amount is the interest you paid.

Tomorrow comes, you do it again. And again. This is how the global financial system funds itself, one night at a time.

It is unimaginably large. The US repo market runs over $12 trillion in daily outstanding volume. That is roughly three times India’s entire annual GDP, turning over every single day.

Sitting in the middle are the big bank trading desks, called primary dealers. Money market funds hand them cash. Hedge funds and bond investors take that cash. Dealers earn the difference.

Here’s the thing. Dealers cannot do this for free. Every trade they pass through has to sit on their own books for the night, and their books have a size limit set by regulators.

So think of a dealer as a warehouse owner. The bonds are boxes. The warehouse is a fixed size. And storage has rent.

A new IMF working paper by Kleopatra Nikolaou is really about one question. Who ends up paying that rent, and when?

Most research on this asks what pushes repo rates up on average.

Nikolaou does something smarter. She splits the last seven years into calm days, normal days, and bad days, and asks the question separately for each.

It’s the difference between asking how a city’s traffic behaves on average, versus asking what jams it up at 3am versus 9am. Same roads. Totally different culprits.

And that’s exactly what she finds.

One thing works in every weather. Bank reserves, meaning the spare cash banks keep parked at the Fed, always calm the market down. More cash sloshing around, cheaper overnight borrowing. Simple.

But the effect is not constant. On calm days, extra reserves shave a small amount off borrowing costs. On the worst days, they shave off roughly twice as much. Cash matters most exactly when there isn’t enough of it.

Now the odd part. Dealer activity, meaning how much of that warehouse is being used, does not behave in a straight line at all.

On calm days, busier dealers mean higher rates. In the middle, busier dealers actually mean lower rates, because they’re smoothing out mismatches like good middlemen do. On the worst days, busier dealers mean sharply higher rates again.

Up, down, up.

Same measure, three different behaviours. Something is hiding inside it. Two somethings, actually.

On calm days, the pressure comes from leverage.

There is a trade hedge funds love called the cash-futures basis trade. A bond and a futures contract on that same bond should be priced almost identically. Sometimes there’s a tiny gap. A fund buys the bond, sells the futures, and pockets the gap.

The gap is minuscule. So the only way to make real money is to do it with borrowed cash, at enormous size.

Where does the borrowed cash come from?

The repo market. Every single night.

This has grown huge. The Fed estimates these positions reached roughly $830 billion by September 2025, about double their previous peak in early 2020. Hedge funds’ net repo borrowing hit roughly $1.8 trillion by the end of 2025, more than doubling in just two years.

All of that eats warehouse space. And when space gets tight, the warehouse owner raises the rent.

The paper’s numbers back it up. On calm days, a busy dealer book alone nudges borrowing costs up a little. Add a big jump in hedge fund positions, and that nudge roughly triples.

Then, as conditions get worse, the effect fades out entirely.

Why? Because the trade stops paying. Higher borrowing costs wipe out a gap that was tiny to begin with. So the funds stop adding, or start selling out.

So on the bad days, who’s squeezing the market?

Uncle Sam.

Government bond issuance barely shows up in calm conditions. But once things tighten, it becomes the main pressure point. When dealer books are already stretched, a wave of new issuance roughly doubles the upward push on borrowing costs compared to normal times.

The reason is boringly mechanical. Every newly auctioned bond has to be stored and financed by somebody before it finds a long-term owner. That somebody is a dealer.

And the mismatch here is brutal. US government debt outstanding has gone from roughly $5 trillion in 2008 to over $25 trillion today. Dealer warehouses have stayed roughly the same size since the 2008 crisis.

More boxes. Same warehouse.

The cleverest part of the paper is a test she runs to check she isn’t imagining things. She reruns the entire model using mortgage-backed bonds instead of government bonds.

The effect vanishes. Completely.

That’s how you know this is specifically about the flood of government debt, and not just a general money market story.

The author is upfront that she’s showing strong patterns, not proof of cause and effect.

There’s a second condition. The warehouses may be getting bigger. After US regulators eased a key capital rule, dealers’ government bond holdings climbed to roughly $550 billion in 2026 on Financial Times calculations, the highest since 2007.

If the bottleneck was the rulebook, changing the rulebook moves the bottleneck. Whether that makes things safer or just shifts the risk somewhere less visible is genuinely unsettled.

Here’s why this matters for us.

India has the opposite problem. Our banking system is drowning in spare cash, not starved of it. Surplus liquidity crossed ₹4 trillion in early April 2026. Overnight rates fell below the RBI’s own policy rate instead of climbing above it.

The RBI has been running auctions to mop up the excess. It has also been openly debating which overnight rate it should even be targeting, since collateral-backed lending now makes up about 98% of India’s overnight money market volume.

In plain terms, India’s version of this pawnshop is already where the action is.

And our pile of government paper is growing fast. FY27 gross market borrowing was budgeted at ₹17.20 lakh crore, trimmed to ₹16.09 lakh crore after adjustments. That’s up roughly 16% on the previous year.

Right now, a cash surplus hides the strain. It won’t hide it forever. Someone still has to store all that paper.

Every central bank unwinding a big balance sheet is hunting for the same thing. The right amount of cash to leave in the system. Too much and policy loses its grip. Too little and December 31 happens.

The Fed stopped shrinking in December 2025, having reversed only half of its pandemic-era expansion.

There is no right number. It depends on how much borrowed money is floating around and how much paper the warehouses are being asked to hold. Both change constantly.

So the honest question was never how much cash is enough.

It’s enough for whom, on which night?

Until we find out, ReadOn!