Banks don’t need deposits to lend. They just really want them.

To steal from the great Cosma Shalzi: Attention Conservation Notice: this is me worrying away at modern monetary theory. Unless that’s your bag, move along.

I want to pick nits with two articles of MMT faith. What concerns me is whether any policy prescriptions follow from them. MMTers portray themselves as having reached new policy conclusions because they are the only ones who really understand how the modern financial system works.

I will take one article of faith at a time. This post is about the idea that banks don’t need deposits to lend. Another post Governments don’t need taxes to spend. They just really want them is about the claim taxes do not fund public expenditure. Caveat: [1].

The critics are right and the standard money multiplier story from first-year undergrad textbooks is wrong: banks don’t need deposits before they can lend. They can lend first, and finance later. Where I differ from MMTers is by seeing this as a rather minor modification to the story, not the bombshell it is portrayed to be. Banks expend a great deal of effort and money to attract deposits. Banks may not need deposits before then can lend, but they do need them [2].

The observation that bank lending is not constrained by the need to have new deposits on hand has somehow gotten translated into the conclusion that banks do not intermediate between savers and borrowers [3]. I think that’s misplaced: a bank’s balance sheet has savers on one side and borrowers on the other. Banks do not need to take deposits first and find borrowers second, but I don’t think the concept of intermediation rests on the order in which things happen. My savings are in a bank earning (pitiful) interest. I did not go out and find the borrowers who are the source of that income. The bank did it for me.

One of the papers that often gets cited as official acknowledgement that mainstream economic has got money creation wrong actually describes things perfectly well [see page 17]. The question to focus on is what happens when a bank expands its balance sheet by lending. When a bank agrees to lend and ‘creates a demand deposit at the stroke of a virtual pen’ it is also undertaking to pay the borrower (or make a payment on their behalf). Without that, the money it has created in that deposit account is useless. Creating money out of thin air is only the first step in a process that relies on banks being able to get their hands on money they have not created.

When a bank lends, it simultaneously creates an asset (the debt) and a liability (the borrower’s deposit account) out of thin air, so the two sides of the balance sheet expand in tandem, but things do not stop there. If the borrower only ever spent that money with other customers of the same bank, no external financing would be needed. But that’s not how things work. Looking at a single marginal transaction, when that loan is drawn down externally, the bank must settle in reserves it cannot create. As an individual bank grows through credit expansion and loans are spent outside the bank, it needs money flowing in from depositors. An individual bank cannot entirely self-finance its own expansion. The banking system does, so long as the central bank plays ball and supplies reserves as demanded. You can only say banks don’t “need” deposits to lend if you think the liability side of a bank’s balance sheet could consist entirely of equity. Possible in theory, wrong in practice.[4]

The money multiplier

The observation that banks create money from thin air is usually seen to contradict the money multiplier story, told in most economics textbooks. I am not sure about that either. Here’s how I was taught it.

First, there is the money multiplier formula itself, which is derived by rearranging definitions and accounting identities. The mistake, which some textbooks make, is to treat an accounting identity as a behavioural equation and proceed as if the parameters in this expression are constants, so that the central bank can choose M by choosing H. This is an anachronism: central banks long ago abandoned any attempt to target H, and any fool can look at historical data to see the ratio H:M is not a constant.

Second, there is a story to accompany the equation. Here’s how it goes. Someone walks into a bank and deposits £100. The bank then keeps £20 in reserve, and lends out £80. That £80 is deposited in another bank, which keeps £16 in reserve and lends out £64. And so on.

In this story, banks create money from thin air. Someone deposits £100, the bank lends £80 so the total amount of money is now £180. If we are ‘following the cash’ £20 is on reserve and £80 in the hands of the borrower, so where has the extra £80 come from? Out of thin air. It is the £80 in the depositor’s account that is not backed by cash on reserve. This only works because people are prepared to accept bank liabilities as money, because they believe that when asked to stump up the cash, banks will be able to deliver.

We need not interpret this story as an assertion that a bank can only lend after somebody makes a deposit. It could be taken merely as a story that communicates in a very simple way how fractional reserve banking creates money, without the implication that this sequencing of events is necessary. The sequence of events it describes is perfectly possible. But we could start the story in a different place. Someone walks into a bank, agrees a loan, and goes off to buy a £80 pair of trousers with their debit card. Then you could describe how the bank’s reserves and liabilities management works. Maybe the growing bank starts an advertising campaign to attract retail deposits because it does not want to rely on wholesale financing (also a form of deposit), and the story could end with somebody walking into the bank to deposit £100. If you could winnow story that down to something comprehensible by the average undergraduate, that would be preferable to the traditional story. But I don’t see that the traditional money multiplier story is so bad, so long as you emphasize it’s just an illustrative story, and that in reality banks do not actually need to obtain the deposit first, before they can lend. Which is what I did, when I taught it to undergraduates as a teaching assistant.

So what?

Let’s accept that mainstream economics made the mistake of thinking that individual banks need to attract more deposits before they can lend more. What mainstream policy prescription rests on that error, and what MMT policy prescription flows from correcting it?

As far as I know, most mainstream economists were unsurprised when QE resulted in banks merely accumulating reserves without expanding lending, because they did not imagine banks were up against a reserves constraint so would expand landing when it was relaxed. A few economists warned of an exploding money supply and hyper inflation but they were mostly met with hoots of derision.

As is often pointed out, most mainstream macro abstracts away from all this stuff in any case. I guess the question is, when they do include banks in their models, what do mainstream economists get wrong, and how would modeling banks in way that got money creation right, change?

UNFINISHED What I need to do is re-read that BoE paper and ask both MMTers and mainstream macroeconomists what results hang on the idea that banks can lend first and finance later.

[1] It is hard for anyone outside the tent to write critically about MMT without being told they haven’t bothered to understand what they are critiquing. Despite having spent a decade reading MMTers explain themselves, that accusation can probably be levied at me. I haven’t read all the Ur texts. So better to read this post as addressing the online disciples of the MMT, not the high priests. An aside, if the online disciples are unreliable sources concerning the true tenets of MMT, I’m not sure why I so rarely see better informed MMTers setting them straight.

[2] If you regard a loan from the central bank as just another form of deposit, then all else equal a bank may need a deposit on the very same day it extends a loan which is spent externally. I think.

[3] That paper argues that a correct understanding of bank money creation has large implications for macroeconomic policy, because many attempts to model banking wrongly impose deposits first lend later. In that sense a correct understanding of bank money creation could be described as a bombshell. I can’t claim to fully understand that paper and what I write above may contradict it — I don’t think a single bank can finance itself through the creation of deposits in the way that paper describes, although I do think banks can instantaneously and discontinuously expand lending.

[4] My current employer CDC Group lends large amounts of money to banks, so that they can in turn expand their lending operations. Why can’t these banks just finance themselves because lending creates deposits?