Tuesday, July 10, 2012

Bella and Stella

Sackerson at Broad Oak got my attention with US financial system has now been reset and a graph showing that "For the first time in 40-plus years, the ratio of monetary base to credit in America has returned to 5%."

I felt as if his post was written just for me. I got out five paragraphs of reply before stopping to catch my breath.

The next time I checked my email a copy of my comment was there, plus a comment by Don't have one, linking to The base money confusion by Izabella Kaminska at FT Alphaville.

I jumped right into that.

Kaminska's article combines her own remarks with remarks from Peter Stella, "formerly the head of the Central Banking and Monetary and Foreign Exchange Operations Divisions at the International Monetary Fund."

"He got in touch with FTAV," Kaminska writes,

because of what he feels is a gross misunderstanding in policy and journalistic circles regarding the nature of central bank reserves, and the myth that banks are not lending because they prefer not to.

Stella finds confusion in the idea that

somehow bank reserves at the central bank ought to be “lent out”, i.e. should exit the “vault” of the BOE, Fed or ECB and begin circulating in the economy. The obverse of this is that an increase in excess reserves at the central bank reflects commercial banks “hoarding” liquidity rather than lending it “out”.

He writes:

My frustration lies in my inability to explain to “sophisticated” people why in a modern monetary system–fiat money, floating exchange rate world–there is absolutely no correlation between bank reserves and lending. And, more fundamentally, that banks do not lend “reserves”.

Commercial bank reserves have risen because central banks have injected them into a closed system from which they cannot exit. Whether commercial banks let the reserves they have acquired through QE sit “idle” or lend them out in the interbank market 10,000 times in one day among themselves, the aggregate reserves at the central bank at the end of that day will be the same.

But Mr. Stella ignores differences between excess reserves and required reserves. FRED shows there were essentially *no* excess reserves for half a century, and then suddenly there was a lot of excess.

If lending increased, there might be the same amount of total reserves, but required reserves would increase and excess reserves would fall. It's no mystery, and it should not be overlooked or hidden away or trampled on or left out of the picture.

I have no trouble with the notion that reserves "are not lent out". But reserves have to be kept in reserve, so to speak, when loans are made. Instead of being "excess", some of the reserves switch over and become "required". So, the aggregate reserves at the end of the day will *not* be the same.

Call it a minor point, if you want, but "the same" and "not the same" are not the same.


As Stella points out, the US banking system could increase lending a thousandfold this year without any change in their deposit holdings at the Fed.

A thousandfold? Possibly. But only because the level of excess reserves is so high.


The only exception to this is if lending constitutes a sharp rise in physical banknote withdrawals from the system. Or for that matter if commercial banks begin to store reserves in physical banknote form rather than on deposit at the central bank so as to avoid negative charges:

See that? There are exceptions.


It is true that banks could reduce their excess deposits at the central bank by exchanging them for physical banknotes and then lend those banknotes out to retail customers.

But "deposits at the central bank" are "reserves". So Peter Stella's claim that "there is absolutely no correlation between bank reserves and lending" must be incorrect.

I need some definitions.

M1 is money in circulation or (as the St. Louis Fed puts it) "funds that are readily accessible for spending." It includes checks and circulating currency. It excludes currency that is not circulating -- currency in bank vaults, basically.

Currency in bank vaults is money in reserve. So are banks' account balances at the Federal Reserve. So: Your cash and your checking account balance count as money in circulation, unless you are a bank. If you are a bank, they count as money in reserve.

When I go to the bank and withdraw $10 from my account, the bank takes the $10 from "the vault" and pays it to me, and then that money is "in circulation". This is the moment the money moves from "in reserve" to "in circulation". It is the moment the money moves from Row 2 to Row 1 and from MB to M1 on this Wikipedia table fragment:

Reserves are money the Fed must pay out on demand, because it is somebody else's money. That's not the same as "open market" operations, where the Fed gets to decide whether or not to buy things. The Fed can choose not to pay out money, if it wants, by deciding not to buy stuff. But it has no choice when it comes to existing reserves.


As we’ve noted before, bank reserves do not leave the doors of the central bank when credit expands, because one man’s loan is another man’s asset. Every time a bank lends, it actually creates brand new credit. This is done by creating a liability for the borrower on one side, and an asset for itself, which can then be sold on, on the other side. Every increase in credit thus comes with an equivalent increase in savings vehicles. Credit is created with one hand and absorbed by the other hand.

Now there's a jumble of notions.

Kaminska opens the paragraph by pretending to talk about reserves. But then she talks about credit expansion instead. The fact that she fails to show how reserves fit into the picture does not mean they don't have a place in it.

The paragraph sounds like it might be the whole story. It is not. Reserves "do not leave the doors of the central bank when credit expands" because reserves are what must be kept in reserve when credit expands.

A few paragraphs later, Kaminska claims "There is no limit to how much banks can expand credit in this way." But then she follows up:

In systems that carry minimum reserve requirements, base money must at the very minimum cover these required ratios. In these cases, when credit rises base money must also rise or else banks could fail to meet such ratios. However, the central bank almost always provides enough base money to ensure that these ratios are met.

In fact, this is how the central bank enforces policy.

So it seems that reserves and reserve requirements *do* limit how much banks can expand credit. To be sure, the connection is not simple and straightforward. Here's what I was taught: Economic policies in the US are designed not to be coercive.

We induce and encourage, because we have a demand economy, not a command economy. But the fact that our policies work by means of guidance, instead of punitive force, does not mean reserves and reserve requirements serve no purpose. To reiterate Kaminska:

The central bank almost always provides enough base money to ensure that these ratios are met. In fact, this is how the central bank enforces policy.

Almost always, she says. Remember the Volcker squeeze?

No comments: