Money is Memory
A professor at Texas A&M put me onto a 1923 book called Monetary Theory Before Adam Smith. It was a Harvard Ph.D. dissertation on the history of thought concerning money. And its author, Arthur Eli Monroe, asks, did Aristotle have this absence-of-double-coincidence notion? Probably not. But he finds someone named Paulus, a Roman jurist in the second or third century A.D., who said something like, when two people meet, it’s often the case that one has something that the other person wants, but not vice versa. And without money, nothing can happen. ...
Now, this story is incomplete. That’s what I tell students: It’s incomplete. Why is it incomplete? Well, think about some isolated pioneer family. At times, someone is going to not feel so well, so he or she isn’t going to be able to chop down wood for a fire. There’ll be lots of absence of double coincidences that arise in that family situation, many cases where one family member is called on to do a favor for another family member. ... Now think about Robinson Crusoe, after he meets Friday. They don’t need money, but again, there might be plenty of absence of double coincidences. Now think further. Here we are in the middle of Pennsylvania. There are lots of Amish communities around here. When they’re isolated, the usual story about an Amish community—or an isolated Israeli kibbutz—is that they didn’t use money. ... Think about this Amish community. The vision is, if my barn burns down, then everybody will come and help me rebuild it. In economics, we try to rationalize behavior without altruism, if we’re able to; so what makes that work without altruism? Everybody notices who shows up to help rebuild it. ... And the guy who doesn’t show up, if he does that repeatedly, will get kicked out eventually. This can work without money because people remember what people have done in the past. ... Yes, “money is memory” is a casual way to state that. Now, that’s a hugely powerful idea that I and other people have been working with. ...
“Money is memory” is a better idea. It leads you to think about various kinds of payment instruments in terms of the kind of informational structure that supports them. The money that is the best current counterpart to the “money is memory” idea is currency. You don’t need much of an informational network for currency; in fact, you probably don’t need any, except for worrying about counterfeiting. When you use a credit card, you’re issued a loan. Why are you able to receive one? Because there’s an informational network behind your card. Your bank is actually guaranteeing your credit payment up to probably some large amount, as large as you mostly use. And they’re doing that because they know something about you.
In general, you want to describe banking illiquidity as a balance sheet which is unbalanced in terms of maturities: short-term liabilities and, on average, longer-term assets. Now, economists have weighed in on this for a long time. Some have said this is a natural thing. This is what banks are for. Others have said this is dangerous, and we ought to regulate it out of existence. Henry Simons, for example, wrote a book called Economic Policy for a Free Society. And Friedman  often credits him [with this idea]. ... So on one side is a bunch of people who are saying banking system illiquidity—and maybe illiquidity more generally—is harmful; we ought to regulate it out of existence. And on the other side were people who vaguely said, “It’s natural, that’s the function of banks.”
In 1983, Doug Diamond and Phil Dybvig published what to me was an eye-opening paper, a very simple, stripped down model, but one whose elements all seem quite reasonable. One element is that people can’t fully plan the pattern of their future expenditures, so they want something like a demand deposit to be able to spend at any time. A spending opportunity might arise that they hadn’t anticipated, so they want the flexibility of being able to spend at any time.
But giving them that flexibility in, say, the form of a demand deposit, allowing them to withdraw whenever they want, means they might also withdraw not just when they want to spend but because they’re worried about the [safety of that financial] institution. ...[I]n the model, the realization of the spending desire is private information. So, as an example, when you go up to the bank window to make a withdrawal, it’s not written on your forehead whether you genuinely want to make a payment or whether you’re worried about the solvency of the bank. That information is private. Then a second element in the model is that the technology is such that longer-term investments have bigger payoffs than short-term investments. ... That’s why it’s socially a good idea for those deposits to be used to finance this long-term investment. It’s like wine, if you leave it in, it’s going to turn good. If you withdraw it quickly, it’s going to be just the grape juice that you started out with.
How Banking Doesn't Fit into a Basic Model of Market Competition
The literature on banking has always been—like that on money—a troublesome literature. This goes back to economists’ feelings that the general competitive model, often labeled the Arrow-Debreu model, is the main model in economics. It’s very general. We don’t need to have a special theory of production for bookcases and a special theory for bottled water.
But when people try to shove banking into this model, it’s hugely unsuccessful. Why? Because anything that banks might be viewed as doing is redundant in that model. According to the Arrow-Debreu model, you face prices at which you can costlessly trade anything for anything. More generally, no activity that we see in the economy that has to do with transacting fits comfortably within that model. In particular, nothing in the GDP accounts that falls under the FIRE heading—finance, insurance, real estate—fits into that model.