[kgwgk]

>> For that bank with $1m in deposits and $1m in reserves before any lending that 10% requirement means that it can not let its reserves go below 100k

> This just doesn’t make any sense. The whole point of the reserve requirement is to guard against the risk that depositors will withdraw enough money at once to deplete the reserves. The bank needs to meet the reserve requirement of deposits on its balance sheet, not a theoretical future balance sheet.

What part doesn't make sense precisely?

A) The bank has $1m in deposits

B) It has to meet the reserve requirement (10%) for the deposits in its balance sheet ($1m)

C) The reserve requirement is $100k

D) The rest are excess reserves

The balance sheet looks like this:

    Assets                      Liabilities
    $100k Required reserves     $1m Deposits
    $900k Excess reserves

> Say the debtor moves all their money to another bank as per your example. Now the bank has 1m deposits and 100k reserves.

Sure, this is the balance sheet now:

    Assets                      Liabilities
    $100k Required reserves     $1m Deposits
    $900k Loans

> Now those other depositors also move 100k to another bank, so the bank has no reserves. Uh oh - making that 900k loan actually allowed the banks reserves to drop below the requirement in this theoretical eventuality!

That's the whole point of fractional reserve! You have enough reserves to cover a fraction of the deposits amount. If the bank has no excess reserves it will be in breach as soon as some depositor decides to get some of their money back and it will need to get more reserves to remain in compliance.

> The bank doesn’t need to wait for this unlikely event to happen to allow its reserves to drop to 10%, it can just make the extra loans in the first place.

The unlikely event that the people who take loans sends the money elsewhere? What would be unlikely is that they didn't.

[ubercow13]

> That's the whole point of fractional reserve! You have enough reserves to cover a fraction of the deposits amount.

Yes exactly, by making 9m loans, the bank has a fraction (10%) of reserves to cover the deposit amount (10m).

> The unlikely event that the people who take loans sends the money elsewhere? What would be unlikely is that they didn't.

You are assuming that 100% of deposits created by loans will be immediately withdrawn. The thing that doesn’t make sense is that you’re treating deposits created from debt as special. The bank needs reserves of 10% of all its deposits.

Why are you considering the eventuality that the loan holder buys something but not that the saver buys something? They are both equally irrelevant as they are eventualities factored into the 10% requirement.

Say there is only one current account holder at the bank with 1m savings. Then the bank gives that customer a 900k loan. Now the customer buys a house. Why do you assume the house will cost 900k? They might buy a 1.2m house, in which case the bank is stuffed, as it only has 1m reserves. There is nothing special about the 900k.

[kgwgk]

> Why are you considering the eventuality that the loan holder buys something but not that the saver buys something?

Because people take loans for something? It could be to invest, definitely not to keep it untouched in a current account.

Do you know of a single case of someone who took a loan for the sake of it, leaving the deposit created untouched at the lending bank, and paying interests for the privilege of having that deposit?

(The eventuality that the saver buys something is why banks keep reserves, with minimums set by regulators in some countries. To allow for a fraction of those depositors to buy something without the whole setup collapsing immediately.)

Edit: and for what it's worth, this "eventuality" is also seen as a basic scenario in the paper under discussion. That's what Figure 2 is about:

"The house buyer takes out a mortgage... ...and uses its new deposits to pay the house seller."

"The mortgage lender creates new deposits... ...which are transferred to the seller’s bank, along with reserves, which the buyer’s bank uses to settle the transaction. But settling all transactions in this way would be unsustainable: [...] the buyer’s bank will in practice seek to attract or retain new deposits (and reserves) [...] to accompany their new loans."

[ubercow13]

Sure it is likely, and the bank will have to take into account the cost associated with future possible changes in its balance sheet (including if it needs to attract or borrow reserves) when deciding whether to make the 900k loan, or a 1.2m loan, or any loan. But the 10% reserve limit at no point directly limited the amount of loans the bank could make to 900k.

The whole premise of the paper seems to be that this way of thinking is backwards (in terms of the order and causality of events) and not really relevant to modern banking.

Another toy example - there are two banks in the banking system with 1m deposits and reserves, and they are let loose making loans at the same time. Why would they only create 900k of loans? The situation is symmetrical, they can expect the net reserve transfer between them to be small if they make similar amounts of loans. In what way is the 10% reserve requirement limiting them to making 900k of loans in this scenario?

[kgwgk]

Fine, as I said I agree that "if a bank has $1 million in deposits the bank can make $9 million in loans as long as the loans remain as deposits in the bank". But that has little interest.

As for the other question, it's clear from the beginning that things can be said about the banking system that are not necessarily true for any individual bank. That paper says as much "Figure 1 showed how, for the aggregate banking sector, loans are initially created with matching deposits. But that does not mean that any given individual bank can freely lend and create money without limit."

So long!