Banks Can Lend Out Excess Reserves
In the comment section of a post by Scott Sumner on EconLog, a commenter made the claim that banks cannot lend out excess reserves and cited this paper. While most of the details discussed in the paper are correct, the conclusion that banks can’t lend excess reserves is not. I will try to explain the confusion.
First, the author makes the important point that “A key distinction to bear in mind (hinted at in the last previous paragraph) is between individual banks and banks in aggregate.” But then he goes on to “prove” his claim that banks can’t lend out excess reserves by showing that they can’t change the quantity of aggregate reserves. This is true but an individual bank can lend out excess reserves.
Consider the bank’s balance sheet as represented in the paper.
Reserves (R) + Loans (L) + Bond holdings (B) = Deposits (D) + Equity (E)
When a bank makes a loan, they create additional deposits and additional loans (initially keeping their reserves unchanged). However, people don’t just take out loans to keep the money in a deposit account at the bank who gave them the loan. Usually they spend these deposits. When they do this, they become deposits in someone else’s account, possibly at another bank (some may become cash held by the public). When this happens the other bank settles with the first bank and reserves are transferred. This decreases reserves and deposits at the first bank, keeping the above equation intact. It’s true that this doesn’t change the total quantity of reserves in the economy but it does change the quantity of reserves held by that individual bank. They have been traded for loans (notice that this still keeps the above equation intact, it has just changed the composition of the left-hand side.
If a bank does this, it will increase the money multiplier because while the total reserves in the economy remain unchanged (they have just changed banks) the total amount of loans and deposits has increased (those deposits ending up on the balance sheet of the other bank which got the reserves which left the first bank).
Note that even if you assume that when a bank makes loans, it increases its own deposits only and does not decrease its reserves at all, the minimum reserve requirement will still be a constraint on lending because as it increases lending, L and D will increase but R will remain the same. Obviously, at some point, this will cause the ratio L/R to hit the minimum.
Now the claim that banks are not constrained by reserve requirements in normal times is mystifying. Again, this is a case of making an argument about aggregates and extrapolating (erroneously) to the behavior of individual banks. Here is the claim.
The money multiplier view of the world envisages the central bank creating reserves and the reserves multiplying into new lending. That is, reserves constrain bank lending. That would seem compelling. If banks are subject to minimum reserve requirements (requiring them to hold reserves in a certain proportion to their deposits, and deposits are the balance-sheet counterpart to loans at the point of credit creation), then, by restricting the amount of reserves that the central bank supplies, it should be able to control the amount of credit.
But modern central banking doesn’t work this way. Central banks don’t constrain the amount of bank reserves they supply. Rather they supply whatever amount of reserves that the banking system demands given the reserve requirements and the amount of deposits that have been created.
This seems to indicate a deep confusion about what it means to be “constrained.” It’s true (or at least I will accept the premise) that the central bank sets an interest rate target and provides whatever amount of reserves are necessary to hit that target. This does not mean that banks are not constrained by reserves!
This fact I think is indisputable: a legal minimum requirement for reserves exists. This means that if a bank has a ratio of loans to reserves which is equal to that minimum level, they cannot make more loans. When banks make loans in the way described above (which is the same way described in the paper), they increase aggregate loans and deposits without changing the aggregate quantity of reserves. This means that for a given quantity of reserves, there is a maximum amount of loans and deposits that the banking system is capable of creating and this maximum is determined by the minimum reserve requirement.
The author seems to claim that this technical constraint is meaningless because if a bank runs into it, the Fed just creates more reserves. This is not true for an individual bank. A bank cannot hit the minimum level of reserves, want to keep making loans, and just call up the Fed and say “give me more reserves, I want to keep making loans.” They have to trade some other asset for more reserves (this may include attracting deposits) or else they cannot make more loans. This means that the reserve requirement is a binding constraint if the bank is in that situation.
Furthermore, and more importantly, this claim that Fed just supplies whatever reserves the banking sector “demands” given the reserve requirements and deposits being created is complete nonsense. He is just saying that, instead of the Fed determining the amount of reserves (money base) and this leading to some amount of deposits which is determined by the reserve requirement (which is a constraint on the creation of deposits), there is some exogenously given quantity of deposits and the Fed just mindlessly fills in the reserves which this amount of deposits requires. That’s completely upside down. The amount of deposits depends on the amount of loans. The amount of loans and interest rates are determined in the market. Yes the Fed tries to target an interest rate but how do they do this?
A bank has the ability to trade assets of one kind for assets of another kind at certain prices which are determined in the market. Different assets have different rates of return. Barring any other constraints on their behavior, they will choose the quantities of these assets (and the market prices will adjust) in such a way that the rate of return (net of risk premiums and other such considerations) is equal. Reserves usually have a low or zero rate of return. This is fixed by the rate of interest on reserves.
The rate of return on loans is determined by the supply and demand. The demand is beyond the control of the banks. The aggregate supply of loans is essentially flat (maybe not perfectly flat but at least pretty flat) at (or near) the rate of IOR (abstracting from risk premiums) for quantities up to the maximum quantity of loans which is possible for the given quantity of total reserves and the minimum reserve requirement. At that point, it becomes perfectly inelastic (vertical) because there is no way that the banking system can make more loans without the Fed either increasing the total reserves or lowering the reserve requirement. (In other words it is constrained by these two things!)
In normal times, interest rates on loans are high enough that the return on them dominates the return on holding reserves so banks try to make more loans and they do this until they hit the reserve requirement constraint and the market interest rate is determined by where this quantity of loans hits the demand for loans.
Market For Loans
Here i* is the market interest rate R is total reserves and r is the minimum reserve requirement and I am assuming there is no currency held by the public just for simplicity. (I don’t know why it’s blurry I’m not very good at blogging…) Now if the Fed were to increase the total amount of reserves in the economy, this supply curve would shift to the right which would lower interest rates (of course there could be some effect on demand from changing expectations about the stance of monetary policy but that isn’t important for our purposes). The Fed can increase or decrease the quantity of total reserves to try to hit its interest rate target but this doesn’t mean that banks aren’t constrained by reserves. On the contrary, it is because banks are constrained that this works. By increasing the total quantity of reserves (money base) they are relaxing the constraint and this allows banks to make more loans until the constraint becomes binding again.
The quantity of reserves (money base) is a tool the Fed uses to manipulate interest rates to the supposed target. The reason this works is that the expansion of credit is constrained by this quantity. This means that the quantity of base money determines the quantity of credit created and the interest rate. You can say that the quantity of reserves the Fed has to create is determined by the interest rate target and the demand for loans and the reserve requirement but this is not saying something different from the “fractional-reserve banking theory of credit creation,” it is just saying it in a different order. It is still a fact that the Fed controls the size of the monetary base and that this puts a constraint on the total amount of loans and deposits banks can create.
“Zero” Lower Bound and Interest On Reserves
If for some reason the quantity of base money becomes high enough relative to the demand for loans that the demand hits the supply to the left of the maximum quantity of loans which can be created, then this constraint will cease to be binding and instead bank behavior (on an individual level) will be equating the marginal return on loans to the marginal return on reserves (recall that while banks as a whole cannot reduce their reserves, individual banks can). The reserve-requirement constraint still exists but it is not binding. When this happens banks could make additional loans but they are choosing not to because the rates at which they would have to make those loans are not worth it. This is not a criticism of banks. So while the imagery of reserves being parked at the Federal Reserve instead of being loaned out may seem misleading, this is only because if banks made more loans, the reserves in the aggregate would still be parked there. This does not mean that the presence of those reserves, however they are described, does mean that banks have the ability to make more loans.
In this environment, increasing the money base does not relax the constraint on lending because it is not binding and therefore, it does not have the usual effect on interest rates and borrowing/lending. (This is not to say it has no effect since it still affects inflation expectations which affect the demand). On the other hand, lowering the IOR would lower the rate of return on reserves (obviously) which would make loans relatively more attractive. In other words, it would shift supply downward, which would increase the quantity of loans and further reduce interest rates.
Market for Loans
Now if demand suddenly increased in this environment, banks would start making more loans and this could cause inflation if the Fed did nothing to restrain it. However, I agree with the author that the Fed would have no difficulty reigning this in by reducing the base if such a thing occurred. Indeed, I believe the Fed would love to have that problem instead of the current one. However, just because the Fed can change the constraint by changing the total quantity of reserves does mean that it isn’t a constraint.
The claim that “banks can’t lend out reserves” seems to be purely a semantic argument. He doesn’t want to call it “lending out reserves” but his reasoning for this–that an individual bank’s reserves don’t decrease when they lend more–is not correct since, even though they may trade deposits for loans, those deposits are usually quickly transferred, along with reserves, to other banks. This means that individual banks can’t change the total quantity of reserves but they most certainly can change the quantity of their own reserves.