Countdown to QE

The magnitude of April tax revenue this season is very likely to surpass prior years, which means a record amount of banking liquidity will drain from the financial system. A $1 increase in the Treasury’s cash balance due to tax receipts generally means a $1 decrease in bank liquidity. At a time when dealers are already reaching their balance sheet limits to make markets in Treasuries, the drain on bank cash will add sharp upward pressure to repo costs and spill over into the cash Treasury market. The Fed won’t resort to ‘Fed QE’ but will enact a regulatory change that ushers in ‘bank QE.’

In about six months from today on April 15 2025, eligible Americans will be paying Federal income taxes to the Treasury. Individuals pay the Treasury using bank deposits in their checking or savings accounts. Their respective banks settle this payment by shifting an equal amount of bank reserves from the banking system and into the Treasury’s account, called the Treasury General Account or TGA. This means that tax remittances are effectively a drain on bank liquidity – the larger the tax revenue, the larger the drain.

The bank liquidity held as bank reserves (or just reserves) and the Treasury’s cash balance (the TGA) are both liabilities on the Fed’s balance sheet. Unless the Fed is intentionally shrinking or expanding its balance sheet with QT or QE, respectively, the total size does not change. Instead, money is shifted between different accounts – draining one and refilling another – in a liability swap:

Any time deposits are withdrawn from a bank account and sent elsewhere, the bank settles the payment using its bank reserve balance at the Fed. That could mean taxes sent to the Treasury or payments sent to a different party in the private sector. In the example below, I pay Tyler $100: my payment of $100 in deposits will see a $100 decline in my bank’s reserve assets, but Tyler’s bank sees a $100 increase – the total number of reserves doesn’t leave the banking system. Money isn’t created or destroyed; it’s only swapped between entities that can hold, use, and settle payments in reserves:

The Fed’s balance sheet is a closed system, so for the TGA to rise, there must be corresponding declines in other liabilities at the Fed.

The Treasury’s cash balance at the TGA can rise for two reasons: from borrowing or from collecting taxes. When it comes to borrowing (issuing debt), the Treasury can flood the market with a deluge of enough short term Treasury bills to push their yields above the Fed’s RRP rate (see June 2023 replenishment). This will entice money funds to rotate out of lending to the Fed via reverse repo and instead to lend to the Treasury via bills. The RRP is drained, and the TGA is filled:

The Fed’s RRP facility (which is the financial system’s “cash under the mattress”) must of course have sufficient balance to meet the Treasury’s funding needs. For the eight or so years before 2021, RRP balances were nearly zero. A different liability was tapped.

Tax receipts, on the other hand, are settled by bank reserves. An individual’s deposits leave his checking or savings account for the Treasury, and an equal number of reserves are sent from the individual’s bank to the TGA to settle the payment. The magnitude of the changes (denoted by little arrows) is equal:

The liabilities are dynamic, so the swap works in reverse. Emptying a different liability (i.e., the TGA or the RRP) will be a tailwind for reserves, hence why reserves are unchanged since June 2022:

Revenue from April income tax varies from year to year. Capital gains realizations have accounted for a large share of income growth, particularly for the wealthiest taxpayers, with the top 1% of earners reporting around 26.3% of total adjusted gross income (AGI) – most of that is due to capital gains on stocks. These high net worth individuals also contribute about half of all income taxes, most of which are tied to their ~$20 trillion in equities.

In other words, we can say that tax receipts are correlated to asset performance – an epic stock market rally will likewise see large remittances from capital gains, which means a bigger Tax Day.

Tax Day was its largest on record in April 2022 because equities were still riding on the tailwind of Covid stimulus throughout all of 2021 – an epic bull market. With SPX within an earshot of 6000 after an equity rally lasting all year, along with a rally in crypto, gold and even corporate bonds, there’s good reason we should expect 2025 tax receipts to eclipse that from 2022.

And, therefore, that the April 2025 reserve drain will eclipse that from April 2022:

In 2022, the TGA was filled roughly $500 billion over the latter two weeks of April. But pandemic-era tax breaks and restrictions on activity had not yet been completely undone. Citing "coronavirus" the IRS offered several ways to delay April filing on individual income taxes. Both of these suggest that, without a relevant emergency (like Covid still was to the government in late 2021/early 2022), Tax Day in April 2022 would've been much larger.

Zerohedge last week, citing Mark Cabana, pointed to the Q3 quarter-end spike in repo rates as a sign of reserve scarcity and an impending repo crisis. In my follow up article, I explained how repo rate spikes signal that the demand to borrow cash exceeds the market’s capacity to lend cash. We regularly see this happen at quarter-end, when a dealer will “window dress” its bank’s balance sheet by pulling back from deploying reserves in activities that are low in margin but high in regulatory costs. One example of this is repo lending, which they scale back from to re-shore their reserves for regulators at period-end, causing a rate spike...:

...only to immediately return to repo activities, bringing rates back down.

But along with having cash, the market’s capacity to repo lend also depends on the ability for dealers to finance and hold the securities on their balance sheet. Recall that in a classic repo transaction, a dealer will borrow cash from a cash lender (like a money fund) and re-lend the proceeds to a cash borrower (like a hedge fund) at a slightly higher rate. The profit margin is tiny, but the impact on the total balance sheet size is significant:

If there’s a cash shortage, the cost of borrowing Treasuries rises. If there’s less dealer balance sheet space available, the cost of financing Treasuries rises. Both examples of “rising costs” reflect in rising repo rates, and they can be difficult to tell apart unless you look under the hood at other indicators of reserve scarcity. That’s where the Fed’s portfolio manager Roberto Perli looked last month, and noted only one “warrants attention”:

The Fed operates an “ample reserves” framework, where policy is primarily adjusted through interest rate hikes or cuts, and the banking system is left with a bit more reserves than it needs. But there's no way of knowing what level of reserves constitutes "ample." Waller has suggested ampleness is roughly when the level of reserves reaches about 10% of GDP, which is about $2.7t.

"Reserve scarcity" matters because officials don't know where the LCLoR, or Lowest Comfortable Level of Reserves, is. There’s a shared perspective at the Fed that the effects a large balance sheet has on their policy are not fully understood, hence why some influential members prefer to keep it smaller. In a perfect world, officials shrink the Fed's balance sheet with QT until some theoretical point in time where they decide to slow down and then stop.

In September 2022, bank reserves came down to hit below the $3 trillion mark for the first time since Covid. On the eve of the bank panic in March 2023, reserves briefly dipped to below $3 trillion again, the first time since September (the end-2022 number was a case of window dressing):

Did we hit the LCLoR in late September 2022 and again in early March 2023? That would mean the real number is at $3 trillion, within spitting distance of where it is today. The tell-tale sign that the LCLoR has been breached is stress in the banking system, so Credit Suisse would answer that, yes, we did hit it...

... to the point where swap lines came back in force.

Of all the leverage ratios and GFC regulations that bind balance sheets, the most relevant is the SLR or Supplementary Leverage Ratio. A bank must hold a minimum amount of core capital against all on-balance sheet assets (like its portfolio of Treasuries) and off-balance sheet exposures (like repo activity). Not only a minimum of typically 3%, but also a buffer, which is another 2% for the biggest banks.

In other words, a large bank must maintain an SLR of at least 5%. If it were to fall below that figure because it's balance sheet grew too big relative to Tier 1 capital, it would face fines, restrictions on how it redistributes capital, and prevent executive officers from receiving their bonuses…

This means that every Treasury has a balance sheet "cost", and a dealer’s willingness to make markets largely depends on the "costs" of warehousing it. If there is abundant space on his balance sheet, then he can easily finance and hold it. That "space" is dictated by the bank's level of core capital, and when that space is pushed to its limit, the dealer charges more to use it. Prices are raised – "costs" are higher. That could mean higher repo rates, since the first activity they tend to pull back from is Treasury repo. In a more serious case it could mean wider bid-ask spreads.

In April 2020, a temporary SLR exemption that aimed “to ease strains in the Treasury market resulting from coronavirus…” dropped the balance sheet costs for Treasuries down to zero, and banks became limitless buyers. It took a very disruptive event that urged regulators to make the change, and this time will probably be no different.

On January 1^st, the debt ceiling will once again be reinstated. That means the Treasury will resort to ‘extraordinary measures’, spending down its ~$800 billion cash balance at the TGA, in order to keep the government functional and prevent it from default. Remember that draining the TGA is a tailwind for bank reserves, so this may be enough to offset QT. But merely reversing QT is insufficient, since the draw from bank reserves in April will likely be $600 billion or more – and unless stocks were to suddenly nosedive into Christmas, I think that estimate is very conservative.

The overarching point is this: Dealers – who are bank subsidiaries – are required to comply with leverage ratios like the SLR, which obligates they hold a minimum level of capital against assets like Treasuries and exposures like repo activity. An April tax drain of several hundred billion dollars would shock reserves far below where they are today and where they have been since QT began in mid-2022. The reserve drain would transfer balance sheet costs to the repo market where available liquidity dries up and funding costs spike.

It's not only hedge funds that demand repo financing, but the majority of a dealer's Treasury portfolio is repo-financed. If you can't finance a Treasury, you become a forced seller, and that's when the party really starts...

At that point, regulators are left with three basic options:

1. Inject and manage reserves via term repo operations.
2. Restart QE and remove coupons from dealer balance sheets.
3. Exempt Treasuries and Treasury repo from the SLR.

Options 1 will only buy time. We used Option 1 in March 2023 when the BTFP provided banks with one-year term repo loans at par value. Option 2 doesn’t address the problem at all, which is not that banks are short on cash, but that their broker-dealers don’t have balance sheet space to warehouse and repo finance Treasuries (see here). As I said last week, the level of cash (reserves) is abundant…

…which is why missed payments and intraday overdrafts show no signs of stress. In fact, it doesn’t get much better than this:

Option 3 would be tailored, effective, and permanent. It would resolve a host of problems that have haunted the world’s most systemically-important market over the last few years, immediately sourcing a new marginal buyer and giving banks the ability to mediate during a crisis.

Is it misleading to call this QE? I argued in my article from last week that giving banks bottomless “war-time” balance sheet space for Treasuries would mechanically be the same as a QE program. I said that the coming 'bank QE' will look different than the 'Fed QE' we are used to:

“…if dealers have infinite balance sheet space for Treasuries, this is QE. It will look like QE, behave like QE and solve the problems QE is designed to address, but nobody in the official sector will call it that. That is the next step…”

The only difference is that, instead of the Fed removing the supply of Treasuries held by non-banks to depress longer-dated yields, we will have the dealer subsidiaries of G-SIBs removing the supply of Treasuries held by non-banks due to regulatory encouragement.

What does it look like? I think I’ve described this scenario in every one of my zerohedge articles, but the last time balance sheets were overwhelmed with activity enough to warrant SLR relief was during March 2020. Dealers couldn’t repo finance Treasuries not because they didn't have the cash but because they ran out of space for repos, which saw a violent Treasury selloff that unraveled the basis trade. A daisy chain of blowups and forced selling, from equities to corporate bonds, was unleashed.

I'm writing and publishing this directly onto zerohedge and nowhere else. I read the comments and try to answer questions.