Take a twenty out of your wallet. Cotton and linen, about a gram, green ink, a serial number, and across the top the words Federal Reserve Note.
That last part is not decoration. The note is a liability of the Federal Reserve. You are holding an IOU issued by the institution whose balance sheet was the actual news on Wednesday, and the fact that almost nobody knows this is why the story was reported as though nothing happened.
The rate was held. That was expected and it is the least interesting half. The other half was an announcement about how fast the balance sheet shrinks, and this is a primer on what that means.
A central bank has a balance sheet like anybody else
Two sides, and they have to match.
On the asset side sit the things the Fed owns: overwhelmingly Treasury securities and mortgage-backed securities bought over the last fifteen years.
On the liability side sit the things it owes. Currency in circulation, which is the note in your pocket. And reserves, which are the balances commercial banks hold at the Fed, and which are the single most important number in this whole subject.
When the Fed buys a bond, it does not spend anything. It creates a reserve balance and credits it to the seller’s bank. The asset side grows and the liability side grows by the same amount, which is what people mean when they say a central bank creates money, and it is far less mystical than the phrase suggests.
What quantitative tightening actually is
The reverse, and it happens by doing nothing.
Bonds mature. When a Treasury security the Fed holds comes due, the Treasury repays it, and the Fed can either reinvest the money in a new security or let it go. Letting it go shrinks both sides of the balance sheet: the asset disappears, and so do the reserves that were created to buy it.
That is the whole mechanism. No selling, no announcements, no market impact on any given day. The Fed sets a monthly cap on how much it allows to run off, reinvests anything above the cap, and the balance sheet shrinks at a pace it chooses.
What was announced
From June the monthly cap on Treasury runoff falls from sixty billion dollars to twenty-five billion. The cap on mortgage-backed securities stays at thirty-five billion.
So the balance sheet keeps shrinking and shrinks more slowly. Nothing is being bought. Nothing is reversing. The brake is being eased, not released.
The operating instruction that implements it goes to the New York Fed’s trading desk, which is where this is actually executed, and it is written in the flat procedural language of a standing order rather than as an announcement.
How the balance sheet got this large
Two episodes, and it is worth separating them because they are constantly conflated.
After 2008 the Fed bought bonds because the policy rate had reached zero and could go no lower. Buying long-dated assets was a way of continuing to ease when the ordinary instrument had run out of room, and the balance sheet went from under a trillion dollars to about four and a half.
In March 2020 it bought on a different rationale. The Treasury market, which is supposed to be the most liquid market in the world, stopped functioning over a period of days, and the Fed bought at enormous scale to restore it. That was a market-functioning intervention rather than a stimulus one, and it is the reason the total went to nearly nine trillion.
Those two purposes imply different exits. Assets bought to ease policy should come off as policy tightens. Assets bought to fix a market can come off whenever the market is fine, which it has been for years. The runoff since 2022 has been unwinding both at once, without distinguishing them.
Why they did it, which is about reserves rather than about the economy
Here is the part that explains the timing, and it has almost nothing to do with inflation.
Reserves are what banks use to settle payments with each other. A banking system with abundant reserves settles smoothly. A banking system with scarce reserves does not, and the transition from abundant to scarce is not gradual. It is a cliff, and nobody knows in advance where it is.
That is not a hypothetical. In September 2019 the overnight funding market seized up with no warning, rates spiked, and the Fed had to intervene within hours. The cause was reserves falling below the level the system needed, at a point where everybody had believed there was room to spare.
Slowing the runoff is a direct response to that memory. Going slower means the approach to the cliff edge happens over a longer period, with more data, and with more chance of noticing the ground changing before anybody falls off it.
There is a third consideration and it is fiscal rather than monetary. A Fed holding fewer Treasuries is a Fed remitting less to the Treasury, and a Treasury issuing into a market with one fewer large buyer pays a little more. Neither effect is large on its own and both are permanent.
What this is not
It is not easing, and it will be described as easing.
An institution reducing its holdings more slowly is still reducing them. The stock of reserves still falls every month. Nothing about Wednesday puts money into the economy or lowers anybody’s borrowing cost.
It is also not a signal about rates. The two instruments are deliberately separated: the rate is the policy tool and the balance sheet is being normalized on its own track, and the Fed has been explicit for two years that the second should not be read as information about the first.
What to watch
Not the balance sheet total, which is large and falls smoothly and tells you nothing about stress.
The total is a stock, and the question is entirely about whether the stock has fallen below what the system needs, which is a level nobody can identify in advance and everybody identifies immediately once it is crossed.
Watch reserve balances as a share of bank assets, which is the measure the Fed itself uses to judge whether reserves are still abundant. And watch the overnight funding rates, the ones that spiked in 2019, because those are where a shortage shows up first and they move before anything else does.
A reserve shortage announces itself in a bank’s cost of funding drifting upward on days when nothing else has moved, roughly a week before anybody gives it a name. That drift is the thing to watch, and by the time it is reported, the institution that issued the note in your wallet will have been on the phone about it for some days.
