QT2 — Shrinking the Balance Sheet

By the spring of 2022, the word “transitory” had become a punchline. Consumer prices were rising at 8.5% year-over-year — the fastest clip since 1981. Used car prices were up 35%. Gasoline had doubled. Rent, groceries, and airline tickets were all surging. The $6 trillion in pandemic money creation, which we traced in Episode 6, was crashing headlong into a supply chain that still hadn’t fully recovered. The Fed had spent most of 2021 insisting the inflation would pass on its own. By March 2022, even the Fed’s most patient voices had stopped pretending.

The pivot came in two parts. First, rate hikes: the Fed raised the federal funds rate from near zero in March 2022 to 5.25–5.50% by July 2023 — the fastest tightening cycle since Paul Volcker’s campaign in 1980–81. But rate hikes alone couldn’t address the elephant in the room: the $8.97 trillion balance sheet. The Fed owned roughly one-quarter of all outstanding Treasury debt and one-third of the MBS market. Simply raising rates while holding $9 trillion in assets was like stepping on the brake while a boulder sat on the accelerator.

On May 4, 2022, the FOMC announced its plan for quantitative tightening. Starting June 1, the Fed would allow up to $47.5 billion per month in securities to mature without reinvestment — $30 billion in Treasuries and $17.5 billion in mortgage-backed securities. After three months, the caps would double: $60 billion in Treasuries, $35 billion in MBS, for a combined maximum runoff of $95 billion per month. That cap took effect September 1, 2022. It was the most aggressive balance sheet unwind ever attempted.

The mechanics were deceptively simple. When a Treasury bond on the Fed’s books matured, the Treasury Department would pay the Fed the principal. Normally, the Fed would reinvest that money by buying a new bond. Under QT, it simply let the payment extinguish the liability — the money disappeared from the system. No selling on the open market, no disruption to bond prices. Just quiet attrition, month after month. The Fed was draining the bathtub one bucket at a time.

The first year of QT2 was textbook. The balance sheet fell from $8.94 trillion in April 2022 to $8.34 trillion by March 2023 — a decline of $600 billion, or roughly $55 billion per month. The actual pace fell short of the $95 billion cap because MBS prepayments slowed dramatically as mortgage rates rose. When rates went from 3% to 7%, nobody was refinancing, so fewer MBS matured. The MBS side of the balance sheet barely moved; Treasuries did the heavy lifting.

Markets absorbed the initial runoff without drama. The S&P 500 fell 19% in 2022, but the decline owed more to rate hikes and multiple compression than to QT. Bond markets remained orderly. The overnight reverse repo facility (ON RRP) — a parking lot where money market funds stash excess cash at the Fed — absorbed much of the liquidity drain. Reserves in the banking system remained comfortable, hovering around $3 trillion. The plumbing held.

Then, in March 2023, the plumbing broke — but not because of QT.

Silicon Valley Bank didn’t fail because of quantitative tightening. It failed because of the oldest mistake in banking: borrowing short and lending long. SVB had taken in $189 billion in deposits — mostly from tech startups flush with venture capital — and invested heavily in long-dated Treasuries and MBS. When the Fed raised rates from 0% to 4.5%, those bonds lost enormous market value. SVB was sitting on $15 billion in unrealized losses against $16 billion in equity. When depositors noticed, the run was instantaneous.

On March 9, 2023 — a Thursday — depositors tried to withdraw $42 billion in a single day. That’s roughly one-quarter of SVB’s entire deposit base. No bank in history had survived a run that fast. By Friday morning, the FDIC had seized it. Two days later, regulators closed Signature Bank in New York, which had $110 billion in assets and heavy exposure to crypto. The following weekend, Credit Suisse — a 167-year-old institution — entered a forced merger with UBS.

The Fed’s balance sheet told the story of the emergency response. On March 8, WALCL was $8.342 trillion — quietly declining per the QT plan. One week later, on March 15, it was $8.639 trillion — up $297 billion in seven days. The balance sheet jumped by more in one week than QT had shrunk it in the previous three months. By March 22, it peaked at $8.734 trillion, as the Fed continued pumping emergency liquidity.

The tools were familiar from 2008. The discount window — the Fed’s emergency lending facility for banks — saw borrowing explode from $4.6 billion on March 8 to $152.9 billion on March 15 — a 33-fold spike in one week. Additionally, the Fed created a brand-new facility: the Bank Term Funding Program (BTFP), which allowed banks to borrow against their Treasury and MBS holdings at par value — not market value. This was crucial. Banks could pledge bonds that had lost 20% of their market value and borrow the full face amount. It was an elegant way to stop the contagion without forcing fire sales.

The discount window is the Federal Reserve’s original emergency tool, established in 1913. Banks can borrow directly from the Fed by pledging collateral — but they almost never do, because using the discount window carries a stigma. Going to the discount window is an admission that no one else will lend to you. In normal times, total discount window borrowing runs around $2–5 billion, most of it routine overnight lending by small banks.

On March 15, 2023, that number hit $152.9 billion. To put that in context: it was higher than the peak during the 2008 financial crisis. Banks that couldn’t get funding in the private market — or feared they wouldn’t be able to — flooded the window. The stigma evaporated when survival was at stake. The Fed, for its part, actively encouraged the borrowing, signaling that using the discount window was not a sign of weakness but a sign of prudence.

The spike was sharp but brief. By the last week of April, discount window borrowing had fallen to $74 billion. By May 3, it collapsed to $5.3 billion — back to normal. The crisis lasted roughly six weeks from ignition to all-clear. Banks repaid their emergency borrowing, the BTFP stabilized confidence, and the contagion that many feared — a cascading run through the regional banking system — never materialized. Two banks failed. A third, First Republic, was seized in May and sold to JPMorgan. But the system held.

The remarkable thing about QT2 is that it survived the SVB crisis. The Fed never paused QT — not for a single month. Even as it pumped $297 billion into the banking system through emergency facilities, it continued letting Treasuries roll off the portfolio. The message was deliberate: emergency lending is not QE. The balance sheet grew temporarily because of crisis tools, but the underlying QT machinery never stopped.

By June 2023, the SVB-related borrowing had drained away and the balance sheet was back to $8.39 trillion — right where the pre-SVB trajectory would have placed it. From there, the decline resumed its steady grind. The balance sheet fell below $8 trillion in October 2023, below $7 trillion in November 2024, and hit $6.54 trillion in December 2025 — the trough.

The pace did slow. In June 2024, the Fed announced it would reduce the Treasury runoff cap from $60 billion to $25 billion per month, while keeping the MBS cap at $35 billion. The stated reason was to “ensure a smooth transition” as reserve balances declined. The real reason was caution: the repo crisis of September 2019 (Episode 5) had taught the Fed that reserves can go from “abundant” to “scarce” faster than anyone expects. Nobody wanted to repeat that experience.

By late 2025, the runoff had slowed to a trickle. Monthly declines of $15–40 billion replaced the $80–100 billion drops of early QT. In December 2025, the balance sheet hit $6.536 trillion. Then it turned. January 2026: $6.574 trillion. February: $6.606 trillion. March: $6.629 trillion. April: $6.675 trillion. The balance sheet was growing again. QT2, the longest quantitative tightening in Federal Reserve history, was over.

QT1, which we covered in Episode 5, ran from October 2017 to September 2019 and shed $700 billion before the repo crisis forced the Fed to reverse course. QT2 shed $2.4 trillion — more than three times as much. QT1 lasted 23 months; QT2 lasted 42 months. QT1’s maximum pace was $50 billion per month; QT2’s was $95 billion. By every measure, QT2 was bigger, faster, and more sustained than its predecessor.

But the comparison also reveals how much the financial landscape changed. When QT1 started in 2017, the balance sheet was $4.5 trillion and reserves were about $2.2 trillion. QT1 ran until reserves fell to $1.4 trillion — and the system cracked. QT2 started with a $9 trillion balance sheet and reserves around $3.3 trillion. It ran until reserves fell to roughly $2.9 trillion — and the system held. The minimum comfortable level of reserves, it turned out, was somewhere around $2.5–3 trillion. The banking system simply couldn’t function on pre-2008 reserve levels anymore. There were too many regulatory requirements, too much demand for safe collateral, too many moving parts.

The other key difference: QT1 ended in panic, with the repo rate spiking to 10% and the Fed rushing to inject reserves. QT2 ended quietly, with the Fed gradually slowing the pace and letting the program expire. No crisis. No disruption. No overnight rate spikes. The Fed had learned from its mistakes and managed the landing with considerably more skill the second time around.