In this piece at Reason.com, economist Robert Murphy writes the following:

But the U.S. economy has stayed in this holding pattern, where people expect low consumer price inflation and so commercial banks keep their excess reserves earning 25 basis points parked at the Fed rather than make new loans. Thus the process I described above has been thwarted; the quantity of money held by the public right now is much lower than it would be, if the banks decided they would rather make loans and earn a higher interest rate than the 25 basis points currently paid by the Fed.

Unfortunately, this statement implies something about how banking works that is simply untrue, which leads Murphy to some incorrect conclusions about QE’s effect on the economy and what to think about inflation. I don’t like seeing this because I like Murphy and the school of economics he espouses. I also see this same mistake made quite often by other economists, which only confuses everyone. So here’s my attempt at explaining this error in a constructive and uncritical manner:

There is no trade-off between loans and reserves except to the extent that new loans results in higher demand for banknotes.

This is seen by the fact of how the central bank’s balance sheet looks:

ΔAssets (A) = ΔReserves (R) + ΔBanknotes in circulation (BK) + ΔGovernment deposits (GD)

Rearranged, this equation looks like:

ΔR = ΔA – ΔBK – ΔGD

The implication here is that levels of reserves can change in only three ways:

1. The central bank increases or decreases its assets (Fed action).
2. The public increases or decreases the amount of banknotes (cash) it wants to hold (public’s actions).
3. The government increases or decreases its deposits by making or receiving net transfers to or from the private sector (government action).

Now, Murphy’s point regards the actions of commercial banks. He says that are choosing to keep their reserves “parked” at the Fed rather than make new loans.

But as I explained above, banks cannot change the amount of reserves in the system apart from actions by the Fed, the public or the government. They cannot, by making new loans, shed reserves or somehow turn reserves into an investment that earns a higher interest rate.

But one might ask: Doesn’t a new loan involve turning reserves (low-interest into loaned funds (higher-interest)?

The answer is no. When banks make a loan, they simply credit the borrowers account at the Fed (corrected 12/28/14) with new funds. They don’t take this from their reserves. They are, however, limited in the amount of funds they can loan by mandated reserve-requirement ratios. Every new loan moves them close to becoming reserve-constrained. But when they make a loan, they move closer to this limit by increasing the size of their balance sheet all around—not by moving funds around. Their total level of reserves stays the same and the total level of loaned funds increases.

In short, a new loan’s effect on a bank, ceteris paribus, is to increase assets by the amount of the loan. A loan has no effect on reserves.

Now, what could affect reserves is how borrowers use those loaned funds (public’s actions). If the loan causes demand for banknotes (cash) to increase, then reserves will fall as banks redeem such demands with banknotes. Outflow of banknotes, as explained above, reduces reserves for the bank in question, as well as for the system as a whole (assuming those banknotes aren’t withdrawn and then deposited in another bank, which merely transfers reserves from one bank to another).

Another way reserves could fall is if the borrower writes a check against their loaned funds account to someone who uses a different bank. This would result in a transfer of reserves out of the borrower’s bank and into the bank of the person to whom the check was written. This wouldn’t result in net loss of reserves for the system as a whole, though. It just transfers reserves from one bank to another.

So when Murphy says that banks might decide to make loans “rather” than keep reserves parked at the Fed, he’s mistaken. Banks might decide to increase lending, but not at the expense of losing interest on reserves at the Fed. In fact, banks would rather earn interest on both new loans and reserves at the Fed (which is possible because new loans don’t require an outflow of reserves). Ideally, Bob would write a check against his loaned funds account that is addressed to another customer of that bank. Then the bank sees no loss in reserves (and so earns the same interest on the reserves as before) plus an increase in loaned funds which, of course, earns interest.

This is a very subtle point, but has huge implications for predicting inflation and gauging the effects of QE and growth in the monetary base. For example, there is no threat of sky-high levels of reserves “turning into” loans funds and thereby launching us into hyperinflation. Sure, a higher level of reserves pushes banks further from being constrained by their reserve-requirement ratio, which means they can increase lending. But banks are normally not reserve-constrained, so the relationship between reserves and loans is not direct, and might be hardly related at all.

This is not to say that inflation won’t happen at all, or that QE doesn’t fund malinvestment to levels that will ultimately prove unsustainable. In fact, Murphy is right about the general, distortive effects of QE. He’s just wrong about why QE induces more lending, which has implications for his predictions regarding inflation (and, I think, might explain why his original predictions were wrong).

Now, if you’re following this closely, you might have a question: If reserves can’t be lent out and don’t have a substantial effect on how much banks lend, what’s the point of increasing them with QE?

The answer is that QE changes the aggregate portfolio composition held by the private sector by buying government debt and other securities with reserves and bank deposits—something economist Paul Sheard aptly calls “portfolio rebalancing effects.” In short, this induces more confidence among borrowers and encourages banks to lend by “liquefying” their balance sheets.

In short, banks cannot lend out reserves. We shouldn’t expect to see any net outflow of reserves apart from action on the part of the Fed and the government unless the public increases it’s demand to hold cash—an event over which individual banks have virtually no control.

So Murphy is wrong to say that banks might choose to lend “rather than” keep reserves parked at the Fed. When they lend, their reserves stay parked at the Fed no matter what.

And finally, note the implication here that banks wouldn’t want to see diminished reserves as long as interest paid on those reserves is positive. Even if interest rates on loans and other investments is higher, they’d rather earn interest both on their reserves at the Fed plus on these higher-earning investments. This is possible, again, because loans do not “come from” excess reserves. As Sheard explains:

…banks do not need excess reserves to be able to lend. They need willing borrowers and enough capital – the central bank will always supply the necessary amount of reserves, given its monetary stance (policy rate and reserve requirements).

For further reading, check out this short, excellent paper by Paul Sheard (here’s a one-page summary if you really don’t have time). I’m channeling him, really, as well as Forbes’ Frances Coppola and Nathan Lewis.

Finally, while I think I understand this issue quite well, I’m no expert on banking. Let me know if I’ve made a mistake and I’ll make updates as appropriate.