We’ll send you a myFT Daily Digest email rounding up the latest Markets news every morning.
This article is an on-site version of our Unhedged newsletter. Sign up here to get the newsletter sent straight to your inbox every weekday
Good morning. A strong morning followed by a bad afternoon for stocks yesterday. A report of slower hiring and investing at Apple was blamed for the change of mood, but we’re not buying it. An ugly market is gonna ugly. Today, we take a trip into the monetary policy weeds. It may not be for all readers, but geeks are gonna geek. Email us: email@example.com and firstname.lastname@example.org.
QE, bank lending and inflation
Did quantitative easing contribute to inflation, and how much? Will quantitative tightening have the opposite effect?
Important questions: it is generally thought that low rates did contribute to inflation, and the Federal Reserve is hoping like hell that higher rates will slow it. The role of QE and QT, as we’ve noted before, is trickier. There is little consensus among practitioners, academics and Fed officials about what central bank bond buying does to the economy and how it does it. It would be a lot better if we had a clear idea of what was going on.
Benn Steil and Benjamin Della Rocca, economists at the Council on Foreign Relations, have argued in a recent post that QE had a very important, perhaps even determining, role in creating inflation. I suspect they’re wrong, but it’s an important argument to consider.
Here’s how the argument goes (picky financial details incoming!).
When, in the course of QE, the Fed buys a Treasury bond from investors, the transaction is completed through an intermediary — a bank. The proceeds from the sale become a customer deposit at that bank, a liability. At the same time, the bank is credited with a reserve deposit at the Fed for the same amount, an asset.
Because a bank will always have fewer reserve assets at the Fed than deposit liabilities, the addition of the same amount to each will push the bank’s reserve/deposit ratio up. QE improved the bank’s liquidity in that sense.
A bank that has more liquidity has an incentive to lend. They don’t have to lend more, but they have reason to, in order to optimise their balance sheet. This is what Steil and Della Roca call the “credit channel” of QE.
In 2020, the Fed bought many billions in Treasuries, but banks’ reserves fell (they call them “excess reserves”; more on this shortly). This, Steil and Della Rocca say, is evidence that “QE was working as hoped” — that is, it was encouraging lending. When lending goes up, they argue, excess reserves fall.
But the gap between the Fed’s holding of Treasuries and reserves grew wider and wider, allowing Steil and Della Rocca to predict that inflation was coming. Here is their chart:
“As that gap [between Fed Treasury holdings and reserves] began rising again in May 2021], with Core PCE inflation running at 3.5 per cent, the Fed should clearly have declared victory and ended its bond buying. Instead, it continued the [buying] binge for another 10 months. By that time, March 2022, Core PCE inflation was up to 5.2 per cent, and the Fed should have been well into hiking rates.”
I suspect that this argument is wrong for two reasons, one involving how banks behave, and one about financial plumbing.
I’ve been talking to bankers for several years, as an analyst and a reporter, and I have never heard of talk about the decision to lend or not in terms of having ample liquidity or being liquidity constrained. Instead, the effective constraint is loan demand — the availability of creditworthy borrowers who want money.
I asked my favourite bank analyst, Brian Foran of Autonomous Research, about this, and he confirmed my suspicions. “Ninety per cent of bank lending decisions are, do my customers want to borrow and are they in good shape to do so,” he told me. “I’ve never sat in a meeting with a bank CEO who said ‘I’ve got all these deposits and I have to figure out a way to lend them out’. They might say — I have a high loan to deposit ratio and I have to work that out over the long run,” for example by retaining more of the loans the bank makes rather than selling them on to the secondary market.
It could be that bankers either are unaware or hesitant to admit the role that liquidity plays in lending decisions. But there is another point, made to me by former Fed trader and the “Fed Guy” blogger Joseph Wang. Banks’ lending isn’t constrained by the volume of liquidity, but rather by its cost. If banks need cash they can always borrow it at some price. Reserves at the Fed, if they ever did matter, don’t matter now, because as Steil and Della Rocca point out in a footnote, the Fed eliminated all reserve requirements in March of 2020.
Now the financial plumbing point. Here is a version of the Steil and Della Rocca chart, with two other series added: total bank credit creation, and the balances in the Fed’s reverse repo programme (more on what that is momentarily).
Now, one thing in this chart fits very nicely with the Steil/Della Rocca account. The total amount of Treasuries the Fed has added to its balance sheet in the Covid era, $3.3tn (light blue line), at the moment very closely matches the new lending created by US banks (fuchsia line).
Here’s the problem, though. Lending does not track reserves (blue line) at all. Now, as argued above, I don’t think there is much reason that it should. And if those two don’t track, the Steil/Della Rocca argument does not work, because it depends on the idea that higher reserves, driven up by QE, incentivise lending.
And there is another explanation, other than higher lending, for why banks’ reserves have come down. It was, again, explained to me by Joseph Wang. The reserves are being funnelled, somewhat circuitously, into the Fed’s growing reverse repo programme (yellow line).
Here is how that funnelling would take place (now we are getting really technical, so feel free to skip the next two paragraphs). The Fed uses the reverse repo programme to sop up excess liquidity in the banking system that would otherwise force the overnight rate below the Fed’s target. Market participants, mostly money-market funds, can give the Fed their cash and receive an interest-paying Treasury security in return. It’s a collateralised overnight loan to the Fed.
In recent years banks have had more deposits than they wanted, which caused problems with their capital requirements. So they have pushed clients towards money market funds. The money market funds have, in turn, put more and more money into the RRP. The way that transaction takes place is that the money market fund that wants to participate in the RRP makes (another) deposit at a bank, and then that same sum is taken out of the bank’s reserve account at the Fed, and placed in a Fed RRP account. The bank’s reserves at the Fed fall.
I’m not totally confident in any of this, and look forward to hearing what Steil and Della Rocca have to say in response. But it seems to me that if QE encourages lending, the mechanism is much more indirect than the one they suggest.
Zooming back out to why all this matters. Unhedged’s best guess is that QE works primarily by injecting/sopping up liquidity in financial markets, not by encouraging lending. So QT will have its effect by withdrawing liquidity from markets, making them more volatile, diminishing investor risk appetites and increasing demand for cash and risk-free assets. That, rather than weaker lending, is what we should prepare for as QT proceeds.
One good read
Claes Oldenburg was the man.