I've been reading through this paper and found it fascinating, but have a few lingering questions which I can't quite reconcile.

The paper really stresses the point that banks are not reserve constrained in our current environment, and that central bank money (to be precise, M0) is not "multiplied up" into broad money (M2 for this purpose). The paper asserts broad money is created primarily by banks via loans (makes sense), and that this process is not constrained by reserves (doesn't make sense, to me at least).

Pg.16, ¶3: "In no way does the aggregate quantity of reserves directly constrain the amount of bank lending or deposit creation"

I don't understand this assertion. They then provide a very helpful graphic which walks through the balance sheet changes from the perspective of a buyer and seller's bank when a loan is made.

Loan Creation

The customer takes out a loan, their bank increases both loans/deposits by the amount of the loan. Then the customer makes a purchase - so their bank reduces the customer's deposit on the liabilities side, and transfer's reserves on their assets side to the seller's bank. The buyer's bank must have reserves sufficient to facilitate this transaction, as discussed in the bottom right graphic.

How then, is the lending process (and corresponding broad money creation) not limited by reserves? Sure, a bank can create a loan and deposit pair, and assuming no regulatory reserve requirement, they can do this without needing any increase in reserves. However, when it comes time for the customer to withdraw that deposit and make a purchase, they absolutely need reserves. If they do not have sufficient reserves, they either need to borrow them, or they would be insolvent. If the banking sector in it's entirety had reserves less than the amount of the loan being created, how can this lending process continue?

The fact that M2 is about 3.93X M0 as of 2022-06-01 seems to indicate that clearly loans/deposits can easily exceed M0, but where does that difference come from? One explanation could be that deposits which are not associated with loans bring cash into the system, for example those brought in by savers.

Pg. 16 ¶4: "This description of money creation contrasts with the notion that banks can only lend out pre-existing money, outlined in the previous section."


Pg.15 ¶3 "One common misconception is that banks act simply as intermediaries, lending out the deposits that savers place with them."

The Bank of England thinks not!

Maybe they don't "lend out" deposits of savers, but without some form of deposits coming outside the loan creation process, how can there be enough cash to facilitate deposit withdraws associated with loans? Yes, not all customers withdraw their deposits at once, but when we are strictly talking about deposits created via loans, that is precisely what they do.

As a theoretical exercise - imagine a scenario in which the only deposits in the system were those created by loans. If M0 did not equal or exceed the system wide loan amount, no withdraws could occur from the deposits created from such loans. So the notion that "banks create money via loans" seems suspect to me. They create deposits, and yes deposits are M2 (money), but without M0 which loans don't impact, how is there not a hard limit?

Assume for examples sake a simplified banking system with 2 denovo institutions, where the M0 is fixed at \$500 prior to any loans being made, and the only deposits created come from loans. If Bank 1 makes a loan of \$500, then it creates a loan/deposit pair on it's balance sheet of \$500 each. Then the customer withdraws the deposit, placing it with bank 2 to make a purchase. Assuming bank 1 had those reserves, it then eliminates the deposit, and wires the reserves to bank 2. Bank 2 now has \$500 reserves and \$500 deposits, and Bank 1 has the loan and \$500 less reserves. Bank 1 can no longer create loans because it would not be able to honor the deposit withdraw associated with that loan, unless it borrowed \$500 reserves from Bank 1, which could then no longer lend for the same reason. The total amount of loans are constrained by the pre-existing reserves in the system.


3 Answers 3


Very good question. Banks do indeed need reserves, and if short they can borrow them, either from each other or from the central bank. The reason banks can get away with far fewer reserves than deposits is all to do the the fact that banks are large institutions with many customers (so toy examples where you have tiny banks with only a customer or two will give misleading answers). Consider two large banks A and B and the transfer of reserves needed to settle payments between them. Every day there will be thousands customers of bank A transferring deposits to customers of bank B and vice versa, but the corresponding transfer of bank reserves corresponds to the net transfer of deposits. The net transfer will usually only be only a small fraction of the total deposits transferred.

So additional loans of size X will generally necessitate additional reserves (but only a small fraction of X). The reason the BoE paper says this does not place any restriction on bank lending is because the central banks have a policy of never refusing to lend banks additional reserves. Indeed if this policy was not in place then there would indeed be a constraint on money creation.

  • $\begingroup$ Thanks for the reply. My question is really specifically about the claim that deposit creation via loans is equivalent to commercial banks creating money - and critically, that this money creation process does not require the existence of reserves in advance. It would seem to me that this assumption only holds true if there are savers (directly contradicting the paper) who bring in deposits (along with reserves) and may not withdraw them at once (as you mention). If you take out a loan, you are going to withdraw that corresponding deposit. $\endgroup$
    – Solaxun
    Aug 2, 2022 at 17:44
  • $\begingroup$ Not 100% sure I understand your question... When a bank makes a loan...A) under some circumstances there may be no need for any new reserves in the system at all. B) even when some new reserves will be needed, the landing bank can get them after the loan has taken place. Does that help? $\endgroup$
    – Mick
    Aug 2, 2022 at 20:20
  • $\begingroup$ I guess the most concise way I can ask it is - without the deposits of savers, wouldn't loan creation be constrained by the amount of reserves in the system prior to the loans being made? If there are no savers, every deposit created via a loan has no M0 behind it. $\endgroup$
    – Solaxun
    Aug 2, 2022 at 20:56

As Mick touched on in their answer last year, "central banks have a policy of never refusing to lend banks additional reserves".

I believe this answers your very good queries about bank loans and reserve constraints (or lack thereof).

When a bank decides that a loan is likely to be profitable and legal, they will make it. New asset-liability pairs are created for both the bank and the borrower in the form of the loan and new deposits.

As you correctly state, loans often very quickly get used to buy something - eg. a house. If the seller banks with the same bank as the borrower/buyer, reserves are not involved. Bank A would just debit one deposit account and credit another. Bank A's liabilities have not changed and neither have their assets.

However, if the seller banks at Bank B, then to settle the payment, the central bank must debit Bank A's reserves and credit Bank B's of the amount purchased. This results in Bank A's balance sheet contracting (losing both deposit liabilities and reserve assets of equal amount) and Bank B's balance sheet expanding (gaining both deposit liabilities and reserve assets of equal amount).

The clincher is that Bank A never needs to worry about access to reserves in this scenario. If they do not have sufficient funds, then they, as you say, can borrow on the highly liquid inter-bank market (for a % interest fee). If the banking system in aggregate does not have sufficient reserve funds to settle all transactions then it becomes the Bank of England's problem (in the UK since you used that example).

The Bank of England has a legal responsibility to "protect and enhance the stability of the financial system" and to "maintain price stability... [and] to support the economic policy of Her Majesty’s Government, including its objectives for growth and employment." due to the Bank of England Act 1998. This is crucial because it means the BoE would never allow the aggegrate level of reserves in the system to fall to such a level as to render payment settlements null and void. There could be bank runs and economic collapse.

Therefore, to avoid such a scenario which flies in the face of the BoE's explicit objectives in law, they utilise a number of tools to ensure reserves are always available. They act as 'lender of last resort'. These liquidity management tools include Open Market Operations (OMOs) which involve the BoE trading in UK gilts to achieve desired monetary policy

A subset of OMOs is the Short Term Repo (STR) facility. This is when banks can borrow reserves from the BoE (at a premium above bank rate) against highly liquid collatoral such as sovereign debt (eg. gilts or US Treasuries). These loans mature in 7 days but there are longer term reserve loan facilities on offer as well

The bottom line is the demand for reserves is continuously monitored by observing indicators such as LIBOR and the BoE will act reflexively to make them available in all circumstances. This is why the BoE concludes that banks are not reserve constrained when deciding on whether to extend loans or not.

Ultimately, we are in a period of excess reserves where Quantitative Easing (BoE buying up gilts like nobody's watching) has flooded the Sterling monetary framework with them. This is predicted to reverse in the coming years but for now, there are way more reserves than required to settle payments. But note that flooding the banking system with reserves over the past decade did not result in loans being multiplied up. The relevant constraint here (explained in the BoE paper you reference) is customer behaviour and demand for loans, even when rates were low. Now of course, rates have increased and the BoE has started reversing its asset purchases


Pg.16, ¶3: "In no way does the aggregate quantity of reserves directly constrain the amount of bank lending or deposit creation"

I read this as:

The aggregate quantity of reserves only indirectly constrains the amount of bank lending or deposit creation

Indirect constraints may include:

  • the effect on the owner's ego if the bank goes insolvent
  • the fiduciary duty of the bank towards shareholders (regulation!)
  • the fiduciary duty of the bank towards customers (also regulation!)
  • the fiduciary duty of the bank towards creditors (guess what!)

... other than regulation, there is nothing that actually prevents a bank from going insolvent if it really wants to.


Not the answer you're looking for? Browse other questions tagged or ask your own question.