The part of quantitative tightening that trips people up is that it sounds like the Fed is doing something active, when mostly it is doing nothing on purpose. Nobody is out selling bonds into the market. The balance sheet shrinks because the Fed stops replacing what matures. A Treasury note rolls off, the Treasury pays it back, and the Fed just does not turn around and buy a new one. That is the whole mechanism. Runoff, not sales. Once you internalize that, a lot of the fear around QT gets more precise and less scary, and the questions that actually matter come into focus.
The reason I keep coming back to this topic is that the effect on markets depends almost entirely on plumbing that most commentary skips. Two rounds of QT can be identical on paper, same monthly pace, same securities, and one barely registers while the other breaks something. The difference is where the drain lands. So it is worth walking through the machinery slowly.
Runoff caps and the Treasuries versus MBS split
The Fed does not let the whole balance sheet run off at whatever speed maturities happen to fall. It sets monthly caps. Below the cap, everything that matures rolls off. Above the cap, the Fed reinvests the excess so the shrinkage stays capped. This is why QT feels mechanical and predictable in a way that rate decisions never are. You can look at the maturity schedule and get a decent sense of the pace in advance.
Treasuries and mortgage-backed securities behave very differently under this setup, and that matters more than the headline number. Treasuries mature on fixed dates, so the runoff is smooth and roughly hits the cap month to month. MBS do not. Homeowners prepay mortgages when they refinance or move, and refinancing collapses when mortgage rates are high. So in exactly the environment where the Fed is tightening, MBS runoff tends to come in well under its cap because nobody is refinancing. The practical result is that the Fed usually shrinks its Treasury holdings close to plan while the mortgage book drains slowly and stubbornly. If you only watch the total, you miss that the composition is drifting, and the composition is part of the point for a central bank that would rather not hold mortgages long term.
The nuance that decides whether QT hurts: RRP or reserves
Here is the part I wish someone had explained to me plainly years ago. When the Fed lets a bond roll off, money has to come from somewhere on the liability side to balance it. The balance sheet is an accounting identity, so a smaller asset side means a smaller liability side. The two liabilities that can absorb the shrinkage are the overnight reverse repo facility, the RRP, and bank reserves. Which one gives way changes everything.
Think of the RRP as a parking lot for cash that money market funds did not have a better home for. It is idle liquidity, sitting overnight at the Fed, doing nothing for markets. When QT drains that first, it is pulling down money that was inert anyway. You barely feel it. Reserves are the opposite. Reserves are the working cash of the banking system, the stuff banks use to settle payments and meet their own liquidity needs. When QT starts eating into reserves, you are draining liquidity that was actually doing work, and that is when funding markets get twitchy, repo rates spike on quarter-end, and something occasionally snaps.
So the honest one-line summary of QT is this. While the RRP is draining, tightening is close to painless. Once the RRP is near empty and the drain shifts to reserves, every additional dollar of runoff costs more than the last. The same monthly pace that was invisible becomes the thing that tips a funding market into stress. Historically that transition is where prior tightening cycles ran into trouble, and it happened not because the pace changed but because the buffer that had been absorbing the pace ran out.
What to actually watch: reserves relative to GDP
If the pain is about reserves, then the useful signal is the level of reserves, not the size of the balance sheet. And the level that matters is not an absolute dollar figure, because the economy and the banking system grow over time. A reserve balance that felt abundant a decade ago could be tight today. So the cleaner gauge is reserves as a share of GDP. It normalizes for growth and gives you a rough sense of how much cushion is left before the system moves from abundant to merely adequate to scarce.
Nobody knows the exact line where adequate becomes scarce, and the Fed does not either. It is a range, and they tend to find the floor by bumping into it. But you can watch the approach. Here is the rough workflow I use to think about where a QT cycle stands.
- Check whether the RRP still has meaningful balances. If it does, QT is mostly draining idle cash and has room to run without much market stress.
- Once the RRP is near zero, shift your attention entirely to reserves. This is the regime change that matters.
- Track reserves as a percentage of GDP rather than the raw number, so growth in the economy does not fool you into thinking the buffer is bigger than it is.
- Watch short-term funding markets for early tells. Repo rates poking above the Fed's administered rates, especially around quarter-end and month-end, is the system telling you reserves are getting tight before any official acknowledges it.
- Expect the Fed to slow the pace of runoff before it stops, then stop entirely, well before reserves reach true scarcity. They would rather taper early than repeat a funding blowup.
The reason this framing is worth the trouble is that the endgame of QT is usually the setup for the next easing. When reserves get tight enough that the Fed has to stop draining, it is admitting the passive tightening has gone as far as it comfortably can. That does not mean rate cuts are imminent, but it removes one source of ongoing pressure and it often precedes a friendlier stance on liquidity. Watching reserves relative to GDP gives you a lead on that pivot that the balance-sheet headline never will.
Where people go wrong
The most common failure I see is treating QT as a single dial with a single effect. People read that the balance sheet shrank by some amount and assume the liquidity hit scales linearly with it. It does not. The first trillion coming out of the RRP and the last few hundred billion coming out of reserves are not the same event, even though they look identical on the asset side. A related mistake is fixating on the total pace while ignoring that MBS runs off slower than planned, so the mortgage book lingers and the effective drain is more Treasury-weighted than the caps suggest.
If you want one durable rule to carry out of this, it is that QT's bite is a function of what liability absorbs it. Cash parked at the Fed is a shock absorber. Bank reserves are load-bearing. Keep an eye on which one is being drained, measure the cushion as a share of the economy rather than in raw dollars, and let the funding markets tell you when the cushion is thinning. That combination will usually have you thinking about the QT endgame while everyone else is still quoting the headline runoff number.