This is, in my limited view, the classic introduction to what central banks do. Bagehot writes just after the Franco-Prussian War, when London has become the world’s sole financial center after Paris lost that role (and lost the war). Hence English banks are now not only responsible for keeping enough money on hand to meet domestic depositors’ needs; they’re responsible for converting other currencies to gold when foreign bankers demand it. British banking has just turned a corner, and Bagehot is trying to understand how the Bank of England should respond.

Lombard Street is maybe one-third a basic explanation of how banks work. Depositors put their money in the bank; the bank then turns around and lends out some fraction of the deposit. If all depositors then turned around and demanded all their money at the same time, banks would be unable to provide it. Normally this isn’t a problem, since normally all depositors aren’t demanding their money. But “normally” isn’t the adjective you want to be using with a banking system; you need to know that banks will be able to survive during a panic.

Panics might start when someone starts to question a major financial institution’s ability to honor its obligations; let’s call this institution “A” (or we could call it “Lehman”). Suppose bank B does a lot of business with A, and C does a lot of business with B. C knows that B gets its money from A, and C fears that A is on the verge of collapse. Hence C demands that B repay the debts that B owes to C. Now B has to find a way to come up with cash. In order to come up with cash, B calls in the debts that A owes to B. This happens on a wide enough scale; now everyone is demanding cash from everyone else.

Where does this cash come from? Let’s look, for instance, at what happens when B repays C. Maybe C owns some bonds that B has issued. C now demands the most liquid security there is (cash) in exchange for a somewhat less liquid security (bonds). Under normal circumstances, those bonds would be safe beyond question. In a panic, no one will buy those bonds, because everyone is simultaneously looking for liquid securities; today we call this the flight to liquidity. “Highly liquid securities” is a fancy term that just means “something which can always be sold, no matter what.” Cash will always be exchangeable for goods and services, even when bonds are not.

Central banks “create liquidity.” As we mentioned above, (non-central) banks never have as much cash on hand as their books say they do; so if everyone is panicking, and rushing to get their hands on cash, it’s quite likely that their banks won’t be able to help them. It’s in cases like this that central banks are “banks of last resort”: when no one else will provide cash, they will. A smaller bank (think Citibank, for instance) sells the Federal Reserve some security which under ordinary circumstances could be turned into cash without hesitation; the Fed hands bank cash. As Bagehot put it in the 1870s, quoting a Mr. Harman:

We lent it … by every possible means and in modes we had never adopted before; we took in stock on security, we purchased Exchequer bills, we made advances on Exchequer bills, we not only discounted outright, but we made advances on the deposit of bills of exchange to an immense amount, in short, by every possible means consistent with the safety of the Bank, and we were not on some occasions over-nice. Seeing the dreadful state in which the public were, we rendered every assistance in our power.

Bagehot adds, “After a day or two of this treatment, the entire panic subsided, and the ‘City’ was quite calm.”

As Bagehot describes it, a central bank should buy during a panic any assets that it would buy under ordinary circumstances. If the central bank refuses to buy some normally-safe asset, the rumor will spread rapidly that the market is illiquid, and then the panic will go into overdrive. A central bank, unlike a commercial enterprise, is supposed to be immortal; hence a central bank should be able to take a longer view — the central bank sees that a panic will subside in a few months, and that those assets can be sold back at non-fire-sale prices when the panic is over. (For a recent illustration, see TARP.)

This is still, if I understand correctly, the basic outline of how central banks work: during panics, central banks provide liquidity when no one else will. But are there limits to that liquidity? That is, does the central bank have unlimited power to create liquidity, or are its hands tied? During Bagehot’s time, the Bank of England’s power was strictly limited: one desk, called the Issue Department, issued currency, while another, called the Banking Department, held reserves. When a panic happened, banks would draw on the central bank’s reserves; those reserves could, in principle, run out, at which point liquidity would be at an end. When reserves ran out, the Issue Department could not print new currency to meet the demand for liquidity. This structure came out of the Bank Charter Act of 1844. As J.K. Galbraith put it in his magnificent Money: Whence It Came, Where It Went — which really ought to be viewed as a 20th-century updating of Bagehot’s book —

In 1844, after an intense discussion of the respective roles of currency and banking in monetary management, Sir Robert Peel put the Bank firmly in a straitjacket — what Walter Bagehot, thirty years later, was to call the “cast-iron” system. The Bank Charter Act of that year fixed the note issue of the Bank of England at £14 million. This amount was to be secured by government bonds. Beyond that, more notes could be issued only as there was gold and silver (no more than one-fourth the latter) in the vault. The cast-iron system was much too rigorous for another of the previously mentioned functions that the Bank was by now acquiring — that of supplying funds when people came in distressing numbers for their deposits in the lesser banks. This fault was remedied by suspending the law whenever it proved unduly inconvenient.

My naïve take is a simple supply-and-demand story: there’s a demand for cash which spikes during panics, and there’s a supply of cash in the world that’s available to meet that demand. If the demand exceeds the supply, there are two ways to handle that problem: either increase the supply, or reduce the demand. The way to reduce the demand is to increase interest rates; that is, if I come to the Federal Reserve pleading with them to give me cash in exchange for my bonds, they can agree to give me cash, but only via a short-term loan for which they charge me a high rate of interest. This is meant, Bagehot says, to ensure that the central bank truly is a lender of last resort — that borrowers come to the central bank only during periods of real panic.

Alternatively, they could increase the supply of cash by printing money. But as Bagehot and Galbraith explain, increasing supply was not one of the Bank’s powers: there was a fixed amount of money, and going beyond that limit was forbidden.

There really is nothing magical about how central banks work. When they need to “create money,” they do it by buying securities (usually highly safe assets like corporate or government bonds, though — again, see TARP — not always). It’s a simple sale: bonds (or mortgage-backed securities, or whatever) disappear from the market, and cash goes into the market. During ordinary times, people trust one another, so they don’t demand liquidity; hence you’d expect that during ordinary times, society is relatively more indebted than during panics — party A is more willing to loan money to party B, and therefore A is more willing to hold security from party B than A would be during times of panic. When the amount of trust in the economy is higher, the central bank can withdraw cash from the market because, again, demand for cash is lower than the supply. The central bank’s job is to match demand for liquidity with the supply of cash. Modern banks, unlike in Bagehot’s time, can “print money” to achieve this goal.

One alternative that really leapt out at me when reading Bagehot was simply: why not forbid banks from loaning out depositors’ money? I put $100 in the bank, and then … there’s $100 in the bank. As opposed to today, when some fraction of that $100 will immediately be lent out to someone else. I’ve not thought through all the consequences here, so maybe this is an absurd idea; but my understanding is that this was part of the Chicago Plan during the Great Depression. If panics come about because people aren’t sure that their money will be in the bank when they come to ask for it; and if, in response, we set up a “lender of last resort” infrastructure; well then, why don’t we just make it a certainty that people’s money will be in the bank when they come to ask for it?

This gets to another principle of banking systems, which in the United States is made flesh in the form of the FDIC: if everyone knows that their money will be in the bank, then the panic will never even get started. We no longer have bank runs, because if your bank goes out of business, the government will make your account whole (up to $250,000).

But the FDIC has its problems, as does any system of insurance on anything. It’s called “moral hazard”, and it’s a quite general principle: if you know you’re insured, you’re more likely to do the thing that insurance protects you from. The way the system works today is, in short, that the government will backstop your bank, in exchange for which the bank submits to regulation. The regulation ensures that deposit insurance will never need to be used. The bank gets a guarantee that bank runs will never happen; if they do happen, the government will make them whole; and in exchange, the bank agrees to behave responsibly. That’s the theory, but read about the FDIC’s role in the savings-and-loan crisis.

So again … wouldn’t it be easier just to prevent bank runs by preventing banks from lending? This would require a massive reorientation of American society, away from credit and away from an emphasis on growth, and the topic is probably far too large to consider here. But it does seem like a natural option to consider, and it surprises me that Bagehot didn’t even mention a word about it. It’s particularly surprising, given that Bagehot didn’t even have deposit insurance in his time: banking panics happened every few years. At least today, we can shrug and reject the Chicago Plan because we have the FDIC; what was Bagehot’s excuse? Panics were, I suppose, just considered an unavoidable part of nature. Bagehot also seemed to take it as given that the Bank of England would be a profit-making enterprise, and that it was required to return a dividend to its stockholders. A Bank of England which never had to lend during panics would be forced to charge lower rates of interest, and hence would return less to its owners. Seems wise that the Bank eventually was nationalized.

There are questions here which, to me, quite difficult. If the central bank can satisfy an unlimited demand for liquidity by “printing money” (i.e., buying bonds in unlimited quantities), and if modern central banks are — by design — somewhat insulated from the political process, then isn’t there a risk that the bank will print too much money? Again, that’s a much larger topic that I can’t get into here (and about which I have no expertise). I could gesture vaguely in the direction of “the central bank’s interest in securing its reputation,” and that’s probably the truth. But it requires more discussion.

Today we’re in the era of “shadow banking”. That is, the problem now isn’t panics in ordinary depository institutions; the problem is panics in hedge funds, investment banks, and so forth. We have the FDIC to protect our deposits, but there’s still a risk of systemic collapse when trust between investment banks disappears. Even if you were to separate, let’s say, Bank of America’s investment functions from its depository functions — the 21st-Century Glass-Steagall Act that Bernie Sanders, Elizabeth Warren, and others support — it does seem like you’d still have the problem of shadow banks collapsing. I need to understand this problem more deeply; in particular, I need to understand why shadow banks have suddenly become such a problem. Weren’t there always non-depository financial institutions?

It’s good to read Bagehot to understand that the more things change, the more they stay the same. At the same time, I’d like to see an update to Bagehot’s book, to explain how certain things really have changed, and what to do about it.