What Banks Do

why I am not a monetarist

I think of banks as performing two functions: providing access to the payment system; and liquidity transformation.

Demand deposits (i.e., checking accounts) are involved in both payment system access and liquidity transformation. A lot of monetary theory either explicitly or accidentally links these two functions, and I think that causes confusion.

It may seem odd to speak of “access to the payment system,” because there is no single payment system. We don’t enter all of our transactions into one big computer—although we could (it’s an interesting thought experiment). We make payments in a decentralized way, in the market. You may, if you wish, think of “access to the payment system” as the ability to obtain goods and services in the market economy.

You go to a grocery store or a restaurant, and you want food. Unless you’re thinking of stealing it, you need to use the payment system.

I am old enough to remember that when you went to the grocery store or a restaurant, in order to buy food you needed to either pay cash or write a check. Currency and checking accounts gave you access to the payment system. Access to the payment system was synonymous with what economists call M1, which is the sum of currency plus demand deposits.

Monetarists back then would have you believe that the quantity of M1 determined the amount that people could spend. If the Fed stimulated banks to provide more M1, people could spend more, and prices would go up.

But when was the last time you were in a grocery store or a restaurant and used M1 to pay? Most people these days pay using plastic. Some people use their phones. I expect that when my grandchildren are adults, the idea that you access the payment system using currency and demand deposits will seem as quaint as the notion that you access the telecommunications system using twisted-pair copper wire and a landline phone.

So at what point will we no longer think in terms of a causal link between the amount of M1 in the economy and spending and prices? I actually think we passed that point a long time ago.

Instead, I think we should be paying attention to the other function of banks, which is liability transformation. In fact, the entire financial sector is involved with this function, not just banks.

Consumers and businesses issue risky, long-term liabilities to banks, and in return banks issue riskless, short-term liabilities to consumers and businesses. Risky, long-term liabilities include mortgage loans, auto loans, business loans, and long-term government bonds. The riskless liabilities of banks are demand deposits that can be redeemed at par at a moment’s notice.

Suppose I tried to do liability transformation myself. I borrow from Sally and I lend to Joe to start a business. Joe’s business loan is a risky, long-term liability. But my liability to Sally is riskless and short term. If tomorrow she comes in and asks for her funds, I am supposed to give them to her. Obviously, my personal attempt at liability transformation is bound to fail.

There are three mechanisms that enable a bank to perform liability transformation that I cannot perform myself. First, there is expertise. The bank knows much more than I do about how to select loans and how to manage loans once they have been selected. With this expertise, the bank is able to lend at much lower risk than I can myself.

The second mechanism is diversification. A diversified loan portfolio is less risky than a single loan. Also, a diversified set of deposits is less likely to be withdrawn all at once than is a single deposit. With diversification, deposits that are individually short-term become collectively long-term.

The final mechanism is hocus-pocus. The bank puts on a convincing act of being able to do liability transformation successfully. The more that people believe the act, the more likely it is that the bank will be able to pull it off.

I think that a lot of the influence of the financial sector in the economy comes from the hocus-pocus factor. Think of something like the Minsky cycle. At a low point, the financial sector is too conservative, it does very little liability transformation, and economic activity is weak. At a high point, the financial sector is too aggressive, it attempts too much liability transformation, and we have an unsustainable boom. In between, we have the “right” level of economic activity.

What does this mean for the role of the Fed? Because I am inclined to downplay the role of M1 in providing access to the payment system, I do not see the Fed as affecting the economy through that channel, which is the way that the textbooks depicted things when I was first taking econ courses.

But I do see a role for the Fed, and for other regulators, in affecting liability transformation. For one thing, since 2008, the Fed has done increasing amounts of liability transformation itself. It does this using interest on reserves and using reverse repos. Together, these mechanisms, known as “quantitative easing,” allowed the Fed to convert risky, long-term liabilities issued by the Treasury, Freddie Mac, and Fannie Mae into riskless, short-term liabilities owed to banks and Wall Street.

QE was initially marketed by policy makers as a way to expand the economy in what Paul Krugman called a “liquidity trap” and what came to be known as “the zero bound,” meaning that it was difficult or impossible for interest rates to fall below zero. The idea was that in a liquidity trap, the Fed had to expand its balance sheet and buy long-term instruments, like government bonds and mortgages.

Today, short-term interest rates are way above zero. But I haven’t seen Paul Krugman declare the end of the liquidity trap. I haven’t seen officials declare that there is no need for the Fed to hold long-term instruments any more. This confirms my suspicion that “quantitative easing” would turn out to be nothing but an excuse for the Fed to expand its role in financial markets and also facilitate more deficit spending.

But we can set QE aside. In fact, the main way Washington affects liability transformation is through regulatory policy. Deposit insurance and “too big to fail” act as subsidies to the eligible financial actors. Capital requirements and other regulations act as taxes on their activities. Small changes in tax rates can have big effects.

For example, in the early 2000s, regulators reduced the tax rate, i.e., the capital requirement, for highly-rated mortgage securities. This set off the boom in mortgage security issuance, a boom which collapsed in 2008.

More recently the low tax rate, i.e. capital requirement, on long-term government bonds helped steer Silicon Valley Bank and many other banks into loading up on these long-term liabilities of the government. The result is that many banks have large mark-to-market losses on these bonds, as does the Fed itself.

[Incidentally, at this event, the last question from the audience came from a woman who had located SVB’s terms of service. These read, in part

We may require that you give us advance notice no later than 11:00 a.m. Pacific Time seven (7) Business Days preceding the day of withdrawal of $10,000 or more in cash.

She asked how the bank could have let more than $30 billion walk out the door in a matter of hours.

1. One possibility is that the higher-ups at the bank did not know their own terms of service, and also nobody bothered to code the $10,000 limit into the bank software, so SVB accidentally waived its right to limit withdrawals.

2. A more conspiratorial possibility is that they knew about the limit but decided that they ought to waive the limit, because enforcing it would have made other depositors even more eager to run. It would have been the opposite of how Jimmy Stewart handled the situation in It’s a Wonderful Life.

3. An even more conspiratorial possibility is that regulators told SVB to waive the limit, because they knew that enforcing the limit would cause depositors at other banks to want to run on those banks.

I think that the probabilities of these explanations are 70 percent, 10 percent, and 15 percent, respectively, with the remaining 5 percent reserved for “other possibilities.”

While I am talking about the AEI event, the author of the book I was discussing insisted that when a bank makes a loan it creates a deposit at that bank. I thought about this afterword, and the author’s claim is bull hockey. It might be true for business loans. It might be true for a personal loan. But if a bank makes an auto loan to me as a consumer, the money goes to the auto dealer, not to me as a deposit with the lending bank. When it makes a mortgage loan to me, the money goes to a settlement attorney, and then to the seller of the house, not to me as a deposit.]

I emphasize regulatory policy as the channel by which the Fed, along with other Washington entities, affects the financial sector. I downplay any role for the Fed in increasing or decreasing the “money supply,” because I think that most people have access to the payment system regardless of the amount of M1 is in the system. So I would describe myself as a non-monetarist, or an anti-monetarist.

My inclination is to attribute the government’s role in fostering inflation to all of government debt, rather than narrow money. I know that many of my readers are also fans of Scott Sumner. Scott’s argument for treating interest-bearing obligations separately from money is that money pays no interest. True enough, but if interest-bearing obligations are not money, then the Fed these days does not fund its operations primarily by issuing money.

Nowadays, under QE, the Fed is paying interest on most of its liabilities. It pays interest on bank reserves and it pays interest to non-banks via the “reverse repurchase facility.” With QE, “monetary policy” is debt management, not money creation. The Fed is refinancing the government’s long-term debt with short-term debt.

And Sumner himself does not pay attention to M1, either (does anyone?). Instead, he looks at the behavior of nominal GDP or wages as an indicator of the effect of monetary policy. To gauge the current stance of monetary policy, he recommends looking at the forecasts of such variables.

Where I differ is that I doubt that the Fed can exercise the exquisite control over nominal magnitudes that is implied by monetarism. Instead, I see the Fed as having to operate through the liability transformation channel. The cyclical nature of liability transformation makes it difficult to tame. And I see the main impact of the Fed on liability transformation as operating through regulatory policy changes, which are blunt instruments, not frequently adjusted, and fraught with unexpected consequences. Also through crisis responses, as in 2008 and during the pandemic.

This essay is part of a series on human interdependence.