A recent David Beckworth interview of Bill Nelson provided a number of fascinating observations, including this comment:

Nelson: . . . A banker, a chief investment officer at one of the largest banks gave me an example that I like to use to explain how this works. You can be in compliance with all of the regulations by holding three days worth of cash for your emergency situation. But when reserve balances were cheap, meaning market rates were below the rate that the Fed was paying on reserves, they decided they would hold five days worth of cash rather than holding alternative types of liquid assets.

Nelson: But then when the Fed started to shrink its balance sheet back in 2018 and market rates moved up above the Fed’s rate that it was paying, sooner than they expected, as an illustration of what I’m talking about, they thought, “Well, we’ll reexamine this. It might not be cheaper. Let’s hold three days of cash and then hold a couple days of money in reverse repos, because that’s earning a little bit more.” And they looked at it and they thought about it and they decided, “Well, but then we’d have to explain this to our examiner and they would want to know why they were doing it.” And it just wasn’t worth the hassle to make the change.

Nelson is concerned that this might make it more difficult for the Fed to reduce the size of its balance sheet:

Nelson: The Fed was, I think, following the right course of action, when you look back in 2018, and they started to reduce the size of their balance sheet by letting it roll off. It looks like they’re going to be doing that soon in May, or June, July, probably more quickly than before. The proof is going to be in the pudding in terms of what I’m saying about structural demand. If I’m wrong and that flat part of the curve is the right way to look at it, the Fed will be able to shrink its balance sheet quite a long way before it sees any response in rates. But if I’m right, then what will happen is that as it starts to shrink much sooner than it would anticipate, you’ll see rates moving up. And in that paper, there’s kind of an exhibit of how the responsiveness of rates is much different when the balance sheet is shrinking than when it’s increasing.

Nelson: However, they’ll start reducing the size of the balance sheet, market rates will move up a bit above the interest rate that they pay on reserve balances, that will create an incentive for banks to reduce their holdings, find alternative ways to meet their liquidity needs. Bank supervisors, examiners will start getting used to the idea that not every problem will be solved by the banks holding more reserve balances. The Fed needs to encourage that process by educating examiners that they shouldn’t be building in this preference for reserve balances.

Prior to 2020, reserve requirements were one of the Fed’s tools for controlling monetary policy (albeit not used very actively).  Higher reserve requirements increased the demand for base money, and hence were contractionary (ceteris paribus).  Today, banks no longer face explicit reserve requirements, but it seems that bank examiners are imposing a sort of implicit reserve requirement on banks.  I have two problems with this:

1. It’s not clear why bank examiners should care about reserves.  I can see why they might be interested in bank capital, or bank holdings of safe assets, but it’s not clear why bank reserves are important.  Prior to 2008, banks did just fine with extremely low levels of bank reserves, barely 1% of current levels.  If it is default risk that is the concern, T-bills are an equally safe asset.  As for liquidity, the Fed needs to be willing to fully meet the banking system’s demand for reserves in a crisis.  But that fact is true even if banks hold large amounts of reserves.  Nelson points out that the more reserves are injected into the system, the greater the bank demand for reserves to meet the preferences of bank examiners.  There is a sort of ratchet effect. Large reserve holdings do not solve the liquidity problem.

2.  Even if regulators should be focused on bank reserves, it is the level of reserves that ought to matter.  And yet according to Nelson (first quote above), regulators often respond to a change in reserves, not the level.  Thus one bank might find its reserve ratio of 25% to be acceptable, while another bank might face greater scrutiny from examiners if it reduced the ratio from 30% to 27%.  Why?

Ideally, we’d go back to the pre-2008 “corridor system”, where banks held relatively small amounts of reserves and the Fed’s balance sheet was only about 6% of GDP.  A large balance sheet increases the risk of Fed policy becoming politicized, as when it engages in credit allocation.