I am having difficulty following your explanation. You described 4 accounting en...

pash · on Sept 22, 2020

Sorry, re-reading my comment I can see that it is likely only to add to the confusion. (The downvotes suggest so also! Apologies all around.)

I meant to make two points. First, that IMO it is misleading to say that a bank simply creates new money ex nihilo when making a loan, counter to what you wrote ("the bank does not need to have $10k sourced from somewhere else"). There is in fact a bound on the amount of new lending a bank can undertake, one related to its assets, and in that sense the money lent does "come from somewhere." My attempt at a (poorly) stylized accounting was meant to illustrate what sort of thing makes up a bank's cash assets and how they relate to lending, but I'll try a new approach below. Second, I was trying to clarify what "bank money" means, to explain how it gets created, as much of the broader confusion in this thread seems to be around what "[bank] money-creation" really entails.

Let me try to make my first point again. In modern banking systems, fractional reserve has been mostly replaced by capital requirements, which limit a bank to holding an amount of (risk-weighted) assets no greater than some multiple of its capital. A bank's capital is its assets less its liabilities, which we'll write A - L. Let's call the regulatory capital multiple M, then capital requirements essentially say that A <= M(A - L). When a bank makes a new loan, its assets and liabilities increase equally in the way you described. That means that for a loan of X, if the lending bank is not already bounded by capital requirements, that the capital adequacy inequality becomes X + A < M(A + X - (L + X)) = M(A - L). That is, the left side of the inequality increases but the right side of the inequality is unchanged, so the capital requirement binds more tightly. In particular, it must be that the loan amount X < M(A - L) - A; in other words, new loans are constrained to be less than the excess of regulatory capital over pre-existing assets. So a bank cannot just create arbitrarily large new loans; loans are constrained by capital, the positive component of which is assets. That's where loan money "comes from": a bank's assets.

The second point I meant to make about bank money-creation was not directed at your comment specifically. But I wanted to point out that "bank money" simply means the sum of all bank deposits, and that it, too, is constrained to be a finite number. In this sense also banks cannot poof into existence arbitrarily large loans (with corresponding deposits) in the way I read your comment to imply.

rahimnathwani · on Sept 22, 2020

Thanks for following up!

Your point about capital requirements is well made. This is what I was hinting at (but did not explain) in my other comment here: https://news.ycombinator.com/item?id=24538786

> The obvious next question is: what, if any, mechanism > puts a limit on bank X's ability to lend? If it > considers a borrower credit-worthy for a $1 billion > unsecured loan, can it just make the loan, even if the > loan is much larger than the bank's existing balance > sheet? What will happen when the borrower tries to spend > some of the money with a customers of bank Y or Z?

The answer is, of course, the capitalization requirements you described above mean that (i) no, it can't make arbitrarily large loans, and (ii) the bank should hopefully have enough liquid assets to satisfy redemption of demand deposits.