The refunds that will add to America's borrowing

The Treasury switched on the Customs Automated Protest Entity (CAPE) on 20 April. It is a centralised electronic clearing system built to speed up the tariff refunds that the courts have ordered. It will work, which is the trouble. Paying those refunds quickly means the government will have to sell a great deal more short-term debt, and do so far sooner than anyone had planned.

From paper protests to a portal

Refunds of customs duty have traditionally moved at the pace of customs law. An importer lodges a protest with Customs and Border Protection, an officer reviews it, and the money follows once the claim is allowed. Handled that way across the enormous volume of entries filed since the tariffs began, the court-ordered repayments would have stretched across several fiscal years. CAPE replaces the queue with a single electronic intake, so claims that would have crawled through port offices one at a time can now be cleared in bulk.

Speed was the political point. Importers, many of them small firms that paid the duties out of working capital, have spent months lobbying for their money, and a refund that arrives in three years is worth a good deal less to a borrower paying bank rates than one that arrives this summer. The same speed that pleases them moves the cost of the refunds out of some vague future budget and into the Treasury's cash forecasts for the coming weeks.

Refunds drain the cash pile

Automating claims through the CAPE portal has sharply raised the speed at which cash leaves the government. Payments that were expected to trickle out over several years now look like a drain of several billion dollars a month from the Treasury General Account. The Treasury Office of Debt Management cannot let its cash balance fall below the level it needs to operate, so it will have to lift sales of Treasury bills, and probably by a wide margin.

Bills are the natural tool. The Treasury runs its note and bond programme on a regular and predictable schedule, changing auction sizes only at its quarterly refundings and only gradually, so any unexpected hole in the cash position is filled at the short end. The timing is awkward. Mid-April tax receipts have just swollen the account, and the Treasury normally uses that seasonal windfall to pay down bills in late spring. Refund outflows will eat into that cushion, turning what would have been a quiet stretch for bill supply into a busy one.

The Treasury also has a faster instrument than the weekly calendar. Cash management bills can be announced and sold within days, at whatever size and maturity the cash forecast demands, and they tend to appear exactly when outflows surprise. They usually clear at a small premium to regular bills of similar maturity, because investors are being asked to find cash at short notice. If CAPE payouts run ahead of the Treasury's estimates, these off-cycle sales will be the first sign.

Where the cash comes from

A refund paid out of the Treasury General Account lands as a deposit at the importer's bank, which adds to bank reserves at the Federal Reserve. The bill sold to replace that cash pulls money back the other way, from whichever investor buys it. When the Treasury last rebuilt its account at speed, after the debt-ceiling deal of June 2023, money-market funds paid for the flood of new bills largely by moving cash out of the Fed's overnight reverse repo facility, which had held more than $2 trillion at its peak in 2022. That buffer had been run down almost to nothing by late last year, so this time the cash for new bills has to come out of bank reserves or out of the repo market itself.

The net effect on reserves is not zero even when refunds and bill sales match dollar for dollar. Refund cash lands with importers and their banks, while bill buyers draw on a different pool, largely money funds that lend in repo. The system's total reserves may barely move, yet the cash ends up in the wrong places at the wrong times, and that mismatch is what shows up in overnight rates on settlement days.

Bills pile up at the front end

Tens of billions of dollars in new bills will push up short-term borrowing costs. Primary dealers already hold plenty of government paper. They will need cash to absorb more, and their hunt for it is likely to drag repo rates above policy targets.

There is recent precedent for how quickly this can go wrong. In September 2019 a corporate tax date coincided with the settlement of a batch of Treasury auctions, drained reserves in a matter of days and sent overnight repo rates to more than double the top of the Fed's target range before the central bank stepped in with emergency operations. The system is better armoured now. The Fed's standing repo facility, set up in 2021, lends against Treasuries at the top of the policy range, and the Fed stopped shrinking its balance sheet at the start of December. Both limit how far repo can spike, though neither stops it drifting to the top of the range and staying there.

That drift matters more than any spike. Overnight rates that sit a few basis points higher for months feed straight into the Secured Overnight Financing Rate, the benchmark for trillions of dollars of loans and derivatives. Borrowers whose floating-rate debt resets off that benchmark pay more, and the Fed faces an awkward choice between tolerating a policy rate that has slipped from its intended spot in the range and adjusting its facilities to pull it back.

Who pays for speed

The losers are the borrowers who live on overnight funding. Hedge funds running leveraged Treasury positions financed in repo would see their carry squeezed as funding costs rise against the yields on the bonds they hold. Dealers face balance-sheet limits that make each extra bill more expensive to warehouse, particularly around quarter-ends. Money-market funds and their savers are the winners, since more bills at slightly higher yields is exactly what they want to buy.

Importers themselves sit on both sides. The refund cash improves their liquidity at once, and many will park it in money funds or pay down short-term credit lines until they decide what to do with it. Some of the money that leaves the Treasury will therefore circle back to buy the very bills sold to replace it, which softens the pressure at the margin without removing it.

What would change the call

The case against is that the sums are manageable next to a bill market measured in trillions, and that the Treasury can spread extra issuance across many weekly auctions. That is fair as far as it goes. What would change the call is the pace of CAPE payouts in its first weeks, the bill guidance in the May refunding statement, and whether the Fed chooses to buy bills itself to keep reserves ample. CAPE gets importers their money fast, but the bill falls on the money markets: expect heavier bill auctions and dearer repo, with dealers taking the strain, probably for months.