Anyone who has taken ECON1000 has probably seen the simple model of how banks create money in a fractional-reserve banking system, and how an increase in reserves creates a multiple expansion of loans and the money supply. An alternative approach, cogently argued in comments here by JKH, says that it is bank capital, not reserves, that plays the crucial role. I think there may be some truth in what JKH says, especially at present, but it is not the whole truth. I'm going to lay out what i believe. Others can either learn from what I say, or try to help me learn where I might be wrong (or maybe even both).
Let's start with the simplest textbook story.
Bank deposits are money, by assumption. Each bank desires to keep (say) 10% reserves against deposits, either to cover liquidity risks, or because it is required to by law, or a bit of both. Bank capital is irrelevant. Start in equilibrium, where reserves are 10% of deposits at every bank. Now assume the central bank does something that causes each bank's reserves to increase by $10. Each bank now has $10 excess (undesired) reserves. In the first round, each bank increases the supply of loans and deposits by $10. It does not increase loans and deposits by $100 immediately, because it anticipates that when the deposit is spent, it will be re-deposited in another bank, so it will lose $10 in reserves. (The textbook story implicitly assumes that each bank is small relative to the whole banking system, and is looking for the Nash equilibrium.) But in aggregate, of course, there is no loss of reserves. If all banks are doing the same thing, each bank finds it gains as many reserves and deposits as it loses (absent a currency drain, of course). So with deposits and reserves both $10 higher than in the original equilibrium, each bank now has $9 excess desired reserves, so it increases loans and deposits again...
In the new equilibrium, deposits (the money supply) expands by 10 times (1/10%) the increase in reserves. That's the simplest textbook story. (OK, I've told it slightly differently from the textbook, by assuming all banks get the extra $10 reserves, rather than just one bank. That helps me think about the symmetric Nash equilibrium.)
Now let me give a totally different theory. It's one I just thought up this morning. Initially it was just a thought-experiment to help get my head clear. But then I wondered if there might be some truth to it after all. I call it the "Loan Officer Theory of Money Supply".
Forget reserves. Banks don't need reserves to make loans; they need loan officers to manage those loans. The desired reserve ratio is probably zero anyway, and doesn't matter. What matters is the ratio of loans to the loan officers who are needed to manage those loans. Assume, given an average turnover and complexity of loans, that one loan officer can manage a $10 million loan portfolio.
Start in equilibrium, with the desired ratio of loans to loan officers. If the central bank increases the supply of reserves, that does nothing to the money supply. The extra reserves just sit there. Banks won't increase loans with the same number of loan officers. But an increase in the number of loan officers, one per bank, would increase loans by $10 million per bank, and would also increase the money supply by $10 million per bank.
It's the supply of loan officers, and the desired ratio of loans to loan officers, that determine the supply of loans and money.
What's wrong with the loan officer theory? Absolutely nothing, provided we make explicit some assumptions. The first assumption is that the banking technology has fixed proportions between loans (the output good) and loan officers (one of the inputs). There is zero substitution between loan officers and other inputs. This means that banks' demand curves for composite other inputs, holding the quantity of loan officers fixed, is perfectly inelastic when loan officers are fully employed. The second assumption is that the market supply curve of loan officers is perfectly inelastic. Given these two assumptions, and change in the price or availability of any other input (like reserves, or capital) will have no effect on the quantity of loans, and so no effect on the money supply.
But if we relax either of those two assumptions, the supply of loan officers to the industry will no longer be the sole determinant of the supply of loans and money. A fall in the price (or increased availability) of other inputs will cause banks to expand loans by using more other inputs per loan officer, or hire more loan officers (pushing up wages along their supply curve) to make more loans.
You can see where I'm going with this. Here's the Bank Capital Theory of Money Supply.
Forget reserves and loan officers. What matters is the ratio of capital to loans. Assume banks desire (or are required by law, or both) to have capital equal to 10% of their loans. Then the money supply is 10 times bank capital. A fall in the price, or increased availability, of reserves (or loan officers) will have no effect on the money supply. But an increase in banks' capital will cause a tenfold increase in loans and the money supply.
Again, this assumes that there are fixed proportions between loans and capital. And it assumes the supply curve of bank capital is perfectly inelastic. Relax either of those two assumptions, and a fall in price or increased availability of other inputs will cause an increase in the supply of loans and money. If banks can vary the loan/capital ratio, by varying the average riskiness of their loan portfolio (at the expense of lower returns or greater loan management costs of course) then the model fails. Or, if banks can all raise more capital, perhaps at a higher price, the model fails.
The Loan Officer and Bank Capital models fail except under extreme assumptions. But that's not surprising. All simple models fail. That doesn't mean they contain no truth. The supply of bank capital, and the supply of loan officers, will affect the supply of loans and the supply of money, other things equal. And perhaps their effect is more important in the current recession than normally. Bank capital is certainly important now, but has been discussed by others. But maybe, just maybe, my Loan Officer model contains more truth than normal as well. If there have been large structural changes in the demand for loans, so that loan management is now much harder to do, and in greater demand than normal, then perhaps the supply of experiences loan officers does matter much more than normal. (Sound plausible, bankers?)
But, but, but. Why all the emphasis on the supply of reserves, if reserves are just one of many inputs? And more importantly, are reserves really an input?
Let me tackle the second question first. Are reserves really an input in the production by banks of loans and money?
Yes, and no. At the level of the individual bank, reserves are certainly an input at the margin; and rational individuals and banks make choices at the margin. At the level of the banking system as a whole, reserves aren't an input (or, are only r% of an input with an r% desired reserve ratio, and I am quite happy to let r go to zero).
Suppose the desired reserve ratio is zero, for simplicity. An individual bank that makes a new $100 loan, by crediting the borrower's chequing account $100, knows that the borrower will spend the loan, and if his cheque is cashed at another bank, the first bank will lose $100 reserves. If it doesn't have $100, it will need to borrow $100 reserves. That's a required input, and that input has a cost. The cost is the interest rate at which it could borrow reserves, or, in an opportunity cost sense, the interest rate at which it could have lent its own reserves. So the interest rate on reserves is a marginal cost of an input to the individual bank, and affects its supply of loans in exactly the same way that the marginal cost of capital and the marginal cost of loan officers affects its supply of loans.
It simply does not matter to the individual bank's decision, in Nash equilibrium, where it chooses its own quantity of loans taking other banks' quantities of loans as given, that there is no loss of reserves to the banking system as a whole. It's maximising its own profits, not those of the whole banking system. It does not internalise the externality of the fact that its reserve loss is another bank's reserve gain.
So the price and availability of reserves matters, at the margin, for an individual bank's decision, in exactly the same way that the cost of loan officers and bank capital matters.
So why do economists concentrate so much on reserves, and downplay or ignore other inputs in the money supply process?
Because reserves can be influenced by policymakers. Other things equal, the price and availability of reserves, capital, loan officers, etc., all influence the money supply and loan supply process. But a central bank's job, when it determines the price and availability of reserves, is to make sure those other things aren't equal. The slope and position of the market supply curves of bank capital and loan officers are what they are. The slope and position of the market supply curve of reserves is whatever the central bank wants it to be. It can make it horizontal, or vertical, or anything in between. It can make it shift left, right, up, down, back and forth, to try to attain whatever objective it wants to attain.