Banks don’t lend out reserves. Or do they?

Over the last weeks/months there has been a lot of agitation in some circles of the financial blogosphere: are banks lending out their reserves? The matter might sound trivial, but it theoretically impacts various monetary policies, such as quantitative easing. I won’t speak about that here. For those of you who don’t know what reserves are, here is a quick reminder:

The money supply comprises several elements. To make things simple the monetary base is the ‘real’ money (remnants of the gold standard and since then, created by the central bank) and the rest of the money supply is some sort of credit money created by the banking system. The ‘real’ money (monetary base) is also what we call high-powered money, and comprises both physical cash and… reserves held by banks at their respective central banks. Credit money created by banks is actually a claim on high-powered money, or high-powered money substitute. Reserves are convertible into physical cash and vice versa. In the UK, the monetary base represents about 17% of M4 (a measure of the money supply that does not include reserves). Reserves represent 82% of the monetary base.

Frances Coppola keeps reiterating again, and again, and again, and sometimes again, that banks don’t lend out reserves. She wrote her latest piece on the topic last week on the Forbes website. The title couldn’t be clearer: “Banks Don’t Lend Out Reserves”

A simple question that could come to anyone’s mind is: what do banks lend out then?

To be honest, I think that a large part of the misunderstanding comes from semantics. I actually agree with Frances on a number of points she makes. I have some objections to her ‘extreme’ view, or at least ‘extreme semantics’. Things aren’t that clear-cut.

While the endogenous money view suffers in my opinion from fallacy of composition, it looked to me at first that the view that banks don’t lend out their reserves suffered from the converse fallacy, the fallacy of division. But it doesn’t look like Frances falls in that trap, given what she made clear in the comment section of her article.

To be clear, banks do need reserves. If a (single) bank makes a loan to a customer, the bank is going to increase both its assets and its deposits (liabilities) by the amount of the loan. So far it does seem that the bank has created money out of thin air. But what happens if the customer withdraws the money from the bank? Deposits will get back to their previous amount, creating a discrepancy with the total loan figure. At the same time, there will be a drain on the bank’s reserves equal to the amount of the loan, as reserves will have been converted into physical cash (while total high-powered money remained the same). Seen this way, we could probably say that banks do lend out reserves.

What happens if the customer doesn’t withdraw the money but pays with it for a good by bank transfer? The drain on reserves still occurs as the beneficiary bank knocks on the door of the borrower’s bank to get hold of the money. In this case again, we could probably say that banks do lend out reserves.

The borrower’s bank also sees its deposits decrease while, on the other hand, the beneficiary bank experiences a deposit and reserves inflow. As a result, its deposit base grows without lending beforehand. Its now increased reserves will allow it to invest or lend more. This clearly nuances Frances’ claim that

when a bank creates a new loan, it also creates a new balancing deposit. It creates this “from thin air”, not from existing money: banks do not “lend out” existing deposits, as is commonly thought.

As we can see, reserves are used for two things: cash withdrawals and interbank settlements.

Let’s now consider the banking system as a whole in a closed economy. If lending leads to cash withdrawals, we could say again that banks lend out their reserves. However, let’s consider the following case: nobody ever withdraws cash. In this situation, all payments are made by electronic bank transfer. While individual banks experience fluctuations in their reserves, the total amount of reserves in the system never changes. Here, we could say that, indeed, banks don’t lend out reserves.

Here is what I meant by fallacy of division: what is true of the system as a whole isn’t for individual banks.

But so far, apart from semantic issues, I think that Frances and I agree although I wish she were more precise in differentiating single banks from the banking system as a whole. However, what I (kind of) disagree with is about her treatment of reserves as a source (or not) of lending and deposit creation. Frances says:

The volume of excess reserves in the system is what it is, and banks cannot reduce it by lending. […]

The volume of excess reserves in the system is what it is, and banks cannot reduce it by lending. They could reduce excess reserves by converting them to physical cash, but that would simply exchange one safe asset (reserves) for another (cash). It would make no difference whatsoever to their ability to lend. Only the Fed can reduce the amount of base money (cash + reserves) in circulation. While it continues to buy assets from private sector investors, excess reserves will continue to increase and the gap between loans and deposits will continue to widen.

This is something I cannot agree with. She is right to say that reserves will permanently be higher than before they were created (at least until the central bank tries to destroy them one way or another). But she overlooks the multiplier effect on lending and deposit expansion.

Assuming no reserve requirements, a fractional reserve bank will estimate how much high-powered money it needs to retain in order to face withdrawals and settlements. It is well-documented that banks in free banking systems naturally maintained reserves. Let’s assume that a whole banking system estimated that it needed to maintain 10% of the amount of its deposits as reserves to face settlements and withdrawals without endangering its existence. This would de facto become an internally-defined reserve requirement. The system can now create a maximum amount of demand deposits (claims on high-powered money) equal to 1/RR = 1/0.10 = 10 times the amount of reserves in its vaults*.

Now let’s get back to the situation described in Frances’ article. Reserves have been created, but have not been ‘lent out’ apparently. Of course, as Frances said, those reserves will not leave the system, unless they are withdrawn. But, they theoretically do provide banks with the ability to create extra deposits ‘out of thin air’ (the pyramiding process), allowing them to maintain their pre-crisis reserves to deposits ratio. And, as deposits expand, the reserves would not disappear but would simply not be considered ‘in excess’ anymore…

The question becomes: why didn’t banks expand their loans and deposit base in line with the increase in reserves? According to the chart below, the M2 multiplier (obtained by dividing all demand and saving deposits by the monetary base), declined significantly when the Fed started its quantitative easing policies.

I don’t think anybody has a clear answer to this. Low demand for loans, the Fed now paying interests on excess reserves, maintenance of precautionary excess reserves to face a possible future liquidity squeeze as long as the crisis isn’t completely over, are some of the possible underlying reasons. From my experience, the third reason is highly likely. I’ve heard many banking executives over the past few years who made clear that they were temporarily hoarding extra cash and reserves not to lend but… “just in case”.

* This multiplier is obviously a very simple one. Chester A. Phillips, in its famous 1921 book Bank Credit, identified the maximum amount x that can be lent out as:

, with c being the amount of reserves deposited, r the reserve/deposit ratio and k the derivative deposit/loan ratio (derivative deposits being the amount of newly-created deposits by lending that are not withdrawn or transferred by the depositors).

In 1984, Alex Mcleod also wrote a brilliant, but barely readable, book called Principles of Financial Intermediation. He covers pretty much all possible and theoretical cases involving credit expansion based on the public’s acceptance ratio for claims (notes and deposits) on ‘money proper’, reserve to deposit ratio of financial intermediaries, external drain in case of an open economy, simple, cross and compound pyramiding (when the claims on money of one or several institutions can be used as reserves by other institutions)… There is no way I can reproduce his massive equations here…

Moreover, things get even more complex when you add in secondary reserves (very liquid low-yielding financial instruments almost instantly convertible into cash), as well as technology and financial innovations that allow banks to economise on reserves.