Programming Note
This is a re-post of a piece from August 2024. I’m reanimating it here for two reasons: first, because there are a ton of new subscribers who were never emailed this post; second, because I have long intended this to be the first in a three-part series, but the other posts got displaced by the crazy current events of late 2024 and early 2025. I’m therefore sending this again now to kick off that serial. “On Banks” will be followed by “On Risk” and brought to a conclusion with “On Equity” – all to appear in the northern hemisphere summer of 2025.
No one ever talks about banks. Well, we talk about them briefly when they fail. But that news cycle doesn’t last for very long, and it is dominated by recriminations and moral condemnations. In spring 2023, all the major news outlets ran stories about the collapse of Silvergate, Signature, and Silicon Valley banks, but they didn’t insult their readers’ intelligence by explaining the nature of money or the practice of banking. Everyone is presumed to already know those things.
Yet it’s worth asking, where would we have learned those things? Not in secondary school civics or social studies classes; not in university economics, politics, or even business courses; not in the business or financial press. The “theory of the bank” is not a thing in any of those domains. I identify this fact as a real problem, because within the approach taken in this newsletter, banks are in some way the whole ballgame. That is, in terms of money/power:
The quintessence of money is bank money. Banks thereby provide the primary laboratory site for understanding both the nature of money and the way money works.
Banks are the primary institutional site for money/power: they are quite literally a place where power and money meet (here “power” names a broad array of phenomena including state regulatory power, legal rules and requirements, and social and cultural norms around money practices).
This means that banks – including shadow banks!1 – will and must play a central role in any and every capitalist social order. All of which seem like prima facia reasons why anyone who studies such social orders – or really anyone who tries to make sense of how the world works – would seek to gain a fundamentally sound understanding of banks. Hence my succinct and seemingly boring topic: what is a bank?
I would argue that we simply do not have an extant, fully developed theory of the bank, and it is obviously beyond my powers to draw one up here, particularly given my ongoing efforts to complete a post before substack starts yelling at me it’s too long to send over email. What I can do is quickly outline the answers that have been given to the basic question “what is a bank”, and then use the strengths and limitations of those answers to map the contours of a working theory.
Orthodox Econ
I spend a lot of my time in writing this newsletter suppressing any desires or inclinations to lambast neoclassical economics and the teaching of introductory econ across the globe. But every now and then the textbooks get something so wrong that I simply must take a moment to discuss it. Here’s Gregory Mankiw’s Macroeconomics, which finally brings up banks on page 547 (in the nineteenth and final chapter):
We begin by imagining a world without banks. In such a world, all money takes the form of currency, and the quantity of money is simply the amount of currency that the public holds. For this discussion, suppose that there is $1,000 of currency in the economy.
Now introduce banks. At first, suppose that banks accept deposits but do not make loans. The only purpose of the banks is to provide a safe place for depositors to keep their money. (Mankiw, Macroeconomics, 7th edition, 548)
Putting aside the utter inanity of the assumptions,2 their raison d’être is to ground the very concept of banking on a prior, if still tacit, idea of money with positive value (just like a commodity). We start here with a given “money,” which we put in the bank the same way we would put family heirlooms and expensive jewelry in a safety deposit box – namely, for storage and safekeeping. This model makes money exogenous in the sense that it is there from the start; only at a second step do we bring money into the economic system, which could just as well function without it (and for Manikiw, the economic system has been doing so for the previous 546 pages).
From this beginning, Mankiw can then tell the long-disproved story of “fractional reserve banking,” in which the bank chooses to “loan out” a portion of this positive money. These actions do create more money in the economic system, but such money remains secondary, mere credit, still dependent upon the primary “currency” (real money) that was given and their from the start. Thus can Mankiw tidily conclude with the very theory of banking that Keynes invalidated almost a century ago: banks are intermediaries that “transfer[] funds from savers to borrowers” (Mankiw, 550).
All of this is wrong, and everyone already realizes it: from Keynes’s original critique to a long line of heterodox economists and non-orthodox money theorists, we’ve known for quite some time that this is not how money and banking work. And when I say “everyone knows” I do not just mean radical money theorists: a decade ago none other than the Bank of England put out two pamphlets carefully showing that fractional reserves, money multipliers, and all the rest is absolutely not a description of real-world banking practices. But Mankiw has almost no choice but to trot out this false narrative, because the real story would undermine the entire model of neoclassical economics that the multi-million-dollar textbook industry teaches.
To hold together the mythical models, Mankiw engages in some wonderfully devious sleight of hand: he uses balance sheets to map “fractional reserve banking” but he does so by skipping (literally not showing) the key step wherein loans create deposits. Mankiw thus approprates the language of heterodox economists and bankers (T accounts, balance sheets) but inverts the story by always putting money of positive value first.
Heterodox Econ
The heterodox (1B) account starts not with a bankless economy, not with a bank that makes no loans, but instead with banks that make loans. And the core of the heterodox account of banks can be expressed in a phrase: loans create deposits. In response to the idea that bankers are intermediaries standing between savers and borrowers, the heterodox account emphasizes that the banker does not need to start with a prior money of positive value, nor even to wait for depositors to show up. The banker makes the loan first and in so doing creates the deposit itself. Here’s the balance sheet representation of the essence of banking – the swap of IOUs that occurs when a banker makes a loan.
The above represents the engine of money creation: money is created not by central banks through high-level policy; money is created on a daily basis, and in a much more banal sense, by commercial banks, every time they make loans.3
The heterodox argument is absolutely correct, but it can also be quite limited and limiting in its vision of money/power. The problem lies in what the orthodox and heterodox accounts share as explanations of banking articulated from within the framework of neoclassical economics. For both, banks seem to be naturally occurring economic entities. Beginning with a formal “model” of “the economy” the question “what is a bank?” gets relegated if not displaced; instead, the bank is described in terms of the functions it performs within the larger system, which itself then tends toward equilibrium, disequilibrium, etc.
Don’t get me wrong, the key heterodox claim is always trying to burst free from the constraints of that very neoclassical model. It fails to do so in Keynes, whose later work tries to find a way to hold onto a standard economic model at the expense of the earlier, much more radical account of money and banking. Arguably, it succeeds in Minsky’s work.4 Regardless of how one assesses various heterodox economic thinkers, the fact remains that a good theory of the bank today must leave the terrain of economics and draw from the rich resources of both banking and finance.
The Business of Banking
The most important move is to break from the confines of economic models to an analysis of banks as real-world institutions, and this leads to a completely different starting point, one that is both painfully simple yet somehow (vis-à-vis neoclassical econ) also quite radical – banks are capitalist firms.
Banks do not come into being in order to solve a functional problem within an economic model. They emerge for the same reason, and with the same goal, as any other capitalist firm: to make money. Since Marx, but probably well before, we have been capable of understanding bankers as the “purest” form of capitalist. The goal of the capitalist is to turn money into more money (M—>M’) but for most capitalists – and certainly for capitalism as a whole – that’s only possible through commodity production, (M—>C…P…C’—>M’). Bankers get to do it directly (though only, of course, parasitically upon global capitalist production).5
To treat bankers as what they are (capitalists) means that banks: first, seek to maximize profit, and, second, will go bankrupt and cease to exist if they fail to do so. When we properly identify a bank as a capitalist firm we both link it to other capitalist firms and also contrast it with them – that is, draw out its distinctiveness. My local banker and my local dry cleaner are both capitalists, trying to earn profit on which they are existentially dependent. However, the societal costs of a bank going under look quite different than that of your typical capitalist firm. This makes banks a unique type of capitalist firm. If my dry cleaner declares bankruptcy I might be without clean shirts for a few days; if my bank goes bankrupt I might “lose my shirt.”
The Finance View
How can we understand the mechanics of the business of banking? To start, a helpful reminder: cutting against the grain of a great many commonsense understandings (as encouraged by textbooks like Mankiw’s6), deposits are a bank’s liabilities, what they owe to customers. When you deposit money at a bank you are not asking them to hold a positive asset for you; you are loaning them money. In turn, a bank’s actual assets are the loans that it makes. Given these basic facts, there are two different ways to describe the basic mechanism of banking.
Viewed from the worldview of modern finance, banks are doing something quite elementary: borrowing short to lend long (Answer 2 in the table above). “Short” and “long” refer here to time duration, so the bank is borrowing money short term and lending it long term. This accurately describes banking because bank deposits are not just loans from customers, they are “callable” loans. That is, unlike a standard loan that has a set repayment schedule, the bank customer (the depositor) can “call” their loan in at a moment’s notice. To “withdraw” their money is to demand repayment from the bank for the loan the depositor previously made.
These short-term customer loans then provide the bank with funding to make their own, longer term loans – in the form of mortgages to individuals and corporate loans to businesses. Profit is generated in the obvious way: on the interest rate spread between these two loans. Other things being equal,7 longer term loans pay higher rates than shorter term loans. The bank borrows money more cheaply than it loans it out. M—>M’
The finance view of banking was mobilized fully after, and in order to analyze, the collapse of SVB, a case that brought home with force the short-term nature of bank funding. I previously described what happened on the asset side of SVB’s balance sheet: in brief, the value of their portfolio of long-dated bonds plummeted because of rising interest rates. Now we see what happened (and can always happen to a bank) on the liabilities side: the loans were all called in, which is perhaps a more accurate way of describing a run on a bank. In this context, one might be led to wonder: how can the business of banking ever really work if one’s funding can always disappear, literally overnight?
The Bankers’ View
That brings us to the third answer, the one provided by bankers themselves. The traditional business of banking has not, in fact, understood customer deposits as short term loans. Indeed, that’s a rather recent and novel perspective, which emerges by looking at banking through the lenses usually worn by money markets traders (by thinking of deposits as callable loans). But by tradition, and historically as a matter of fact, bank customers don’t call their loans.
This fundamentally changes our understanding of the business of banking. Here’s Matt Levine:
Depositors can take their money out, but they reliably don’t, because they have long-term relationships with their banks; the banks invest a lot in building those relationships to get this cheap long-term funding, and they can use it to make long-term investments.
Bankers do not conceive of deposits as short-term loans, because they have a different understanding of depositors – namely, as customers. Traditional bankers build relations of trust and support and long-term commitment with their customers, designed, among other things, to make sure those deposits stay with the bank, long-term. These standard practices create what has been called the “deposit franchise,” a name that captures the unique nature of the deposit: a callable (hence short-term) low or no-interest loan to the bank, that the creditor (the depositor) thinks of as neither short-term, nor a loan.8
While institutional actors in money markets make short term loans according to a fundamental financial logic (e.g., they seek the best rates possible when they loan out their money), bank customers just put their money in the bank. They have checking and savings accounts at their local branch, a fact they just don’t normally ever question. All of this effectively makes deposits function like long-term loans (Answer 3). The bank earns profit on the spread between the zero or very low interest they pay on deposits, and the substantial interest rates they charge on loans.
What happened with SVB is, on the one hand, quite typical and easily understandable: typical because there have been runs on banks throughout the history of banking; easily understandable, because anyone could see (at least post hoc!) the interest-rate risk SVB had built up by putting so much of their total assets in long-term bonds.
On the other hand, what happened with SVB was quite unprecedented and can be summarized succinctly as follows: the funding that SVB treated as long-term, based on the deposit franchise, turned out to be short-term, when SVB’s customers looked at their deposits the way modern finance tells them to: as callable loans.9 Never in history has that much money been withdrawn that quickly (and successfully!) from a bank.
We might say then that the finance view and the banker’s view are both right, and both wrong, so any robust theory of the bank today would need to plot a course through these tensions – to understand how both views can be viable, despite being utterly opposed.
To achieve this end, I think we need to start with the heterodox insight (1B, loans create deposits) and the basic balance sheet model; add to it the powerful views of both finance and banking (2 and 3); and finally, massively supplement those views with a much richer understanding of institutional power and politics. I would start with the following.
The hierarchy of money and the need for reserves
Money is always plural, never singular. There is never money but rather an always shifting hierarchy of monies. This is Perry Mehrling’s (quite profound) general thesis. We see a particular version of it in the fundamental, but I think poorly understood, fact that banks cannot pay one another with their own deposits (and banks have to pay one another all the time). If I write you a check for $47, then in order to complete the payment at least 3 more things have to happen after you have “deposited” the check:
Your bank has to credit your deposit account, increasing your balance by $47.
My bank has to debit my deposit account, reducing my balance by $47.
My bank has to pay your bank $47.
For the banking system to work, banks need a higher form of money in order to pay one another. Because money is itself a claim on a debtor, a higher form of money is nothing other than a valid claim on a higher bank. You and I have deposit accounts at our respective banks, but our banks have their own deposit accounts at the central bank. In this American example, that means they have “Fed reserves” – i.e., deposit accounts held within the US Federal Reserve banking system. For my bank to pay your bank they instruct the Fed to debit their deposit account by $47 and credit that same $47 to your bank’s deposit account.
This illuminates one of the potential weaknesses of the endogenous money creation thesis (1B). One may be tempted to take the valid thesis that loans create deposits and allow it to underwrite a vision of banks as all-powerful, capable of creating money out of “thin air” and doing with it what they wish.10 But the immediate money that banks create, the deposit account – which really can come into existence through a few keystrokes – is the bank’s debt to the depositor. And when the depositor then spends that money, the bank has to pay in something other than their own debt.
All of this means that although the fractional reserve story is rubbish, and although loans really do create deposits, banks still need “reserves.” And “reserves” are exactly that – a higher form of money, a central bank deposit account. “Reserves” are not the “fraction” of real/positive money that the bank holds onto when they make loans. Reserves are funding for the bank’s business of making loans. When you move your money from bank A to bank B, it feels to you like you are just shifting your deposits (moving your money), but bank B agrees to let you deposit your cash previously held at bank A because bank A will have to pay them in central bank reserves.
Reserves matter – first, because of the hierarchy of money, and, second, because banks only exist in a regulatory environment. And in the regulatory environment, reserves obviously play a crucial role because they are the first thing the regulator scrutinizes.11 This means that any good theory of the bank must include a constitutive role for both central banks and regulation. Banks do not exist in a state of nature: to be a bank you need a charter; to be a bank you must have a regulator.
The Normalization of Money and the Magic of Banking
Banks are the most crucial site for the production of both the stability and precarity of money. Here’s the key: a bank must be both sites at once – the place where money congeals as stable money and the place where money is most at risk.12 We might call this the dialectics of the bank. This argument of mine emerges from my many years of reading Levine: it reworks and brings together two theses that he has articulated separately.
Thesis 1
bank money just is money
These are my words, but they draw out a central claim that Levine makes over and over: a banking system only works if deposits remain usable money – that’s the point. A dollar in the bank is a dollar – the very definition of what a dollar is.
At a first cut, we can say that bank money should serve as the safest money, the money with the most moneyness. But that doesn’t quite capture the full significance of bank money. Banks should provide the site where the very question of money’s moneyness disappears. That is, bank money is so taken for granted as being money, that we need not think of our deposits as claims on our bank (though that is very much what they are). Rather, in our head but also in our actions, bank dollars are just our dollars. In a properly functioning banking system, within a healthy monetary order, bank money ought not even be a type of money. It should be/become money simpliciter.
The false theory of money – money as thing of positive value – that vision is a productive effect of a functional banking system. In other words, it is only through the alchemy of banking that we might come to think of money as a simple thing of positive value.
We can therefore draw out the conclusive force of this thesis by redescribing the banking system as a machine that normalizes deposits as money. This rereading of Levine’s Thesis 1 anticipates his Thesis 2.
Thesis 2
the bank is a machine for turning safe deposits into risky investments*
This statement expresses the obvious truth of banking as a capitalist enterprise: regardless of whether deposit funding is understood as short-term money-market loans (finance view) or a long duration deposit franchise (bankers’ view), the bank is an engine that takes customer deposits as an input and generates loans as an output. The business of a bank is to do risky things with safe deposits. If banks do not seek risk they will not find the “yield” necessary to generate the “spread” (between the interest they pay on deposits and the interest they receive on loans) that keeps them liquid and solvent.
A bank that did not take risks would go bankrupt. The paradoxical logic is palpable: the only way to keep deposits safe is to risk losing them. This is one way to tell the story of SVB: the originary cause of their downfall was that they had too much money. By late 2020 they were “flush with deposits.” But a bank cannot just “hold on” to deposits. They have to do something with them. And by definition, they have to do something riskier with them than the depositors are doing by loaning their money to the bank.
In hindsight everyone has accurately described SVB’s mistake as taking on too much interest-rate risk, but we can just as easily say that their mistake was taking on too little credit risk. SVB tried to play it safe and ended up failing spectacularly. Their depositors’ money was rendered less stable, less moneylike, precisely because SVB did not do risky enough things with it.
Banking and Capitalism
This is the nature of banks not because of spineless regulators or greedy bankers. It’s the nature of banks because it’s the nature of money and capitalism, a point Marx draws out powerfully with his dual images of, first, the “crazy” miser who loses his money by holding onto it, and, second the “rational” capitalist who retains his money only by throwing it into circulation. Under capitalism, the relative “stability” of money is not a binary alternative to “risk.” Rather, under capitalism, only through practices of risk-taking can the semblance of something like “stable money” be established in the first place.
This is the alchemical magic of banking: “issu[ing] risk-free liabilities in order to finance risky businesses.” It is also what makes banks a central institutional engine of any capitalist social order. What, then, could be more important than the theory of the bank?
Shadow banks are just deposit-taking and loan-making entities that lack a bank charter. They range from relatively highly regulated, stable and respected money market funds, to the wild west of crypto – where everything is a shadow bank. Here’s the key: you cannot eliminate shadow banks. That is, you cannot regulate all shadow banks out of existence because bank regulation creates both real banks and the possibility of shadow banks. Of course, the harsher you make the penalties for banking without a license, the more you may be able to limit the number of shadow banks. But those same harsh penalties will increase the incentives to engage in shadow banking (by reducing the competition and increasing the potential profits).
That is, bracketing questions such as the following. Where does “currency” come from if we live in a world with no banks? (Historically “currency” means banknotes, and today “currency” means central bank banknotes.) Why would people need a separate bank to hold onto their banknotes?
I sketch this process out carefully, and then develop it into an explanation of “repo,” in section 2 of chapter 6 of my money book.
Anush Kapadia’s book makes an absolutely crucial contribution to the recent literature. It is arguably the best theory of the bank to be found within heterodox economics (the strongest example of a 1B-type answer). Neoclassical econ has always worked with a formal model of the economy that makes banks so secondary as to be easily ignored. Kapadia responds directly to this blindness by himself sketching a formal model of the economy that makes banks primary. This is, without doubt, a superior answer to anything offered by orthodox economics, yet I still see the project as hamstrung by the fact that it answers neoclassical econ on its own terms – namely, by beginning with that formal economic model. Kapadia borrows an implicit model of “the economy” from neoclassical economics, and then he adds banks and banking into it, with the goal of making them central, fundamental. But his methodology (his m.o. of building out a model) undermines his very efforts to do so. By baking “credit” into commodity exchange (and then having banks come along to formalize what was already happening) Kapadia reifies both the idea of “an economy” and the notion that it could exist without money (through barter). Ironically**, Kapadia here misses Mehrling’s main point: dealers provide liquidity and there are no markets without it. Mehrling’s key claim applies not just to money markets but to commodity markets as well, yet Kapadia’s framework begins with a concept of commodity exchange that forgets all this – and takes us back to neoclassical barter. **The irony is that Kapadia is a student of Mehrling’s.
Not everyone living under capitalism can be a capitalist and not every capitalist can be a banker. Both are true by definition: capitalism could not survive without workers and without non-financial capitalists.
There is an obvious tension in Mankiw’s textbook account. His narrative conceptualizes money as a given currency that the bank just holds onto. Therefore his first balance sheet (548) has to register that currency as both an asset (the reserves) and a liability (the deposits). The bank here isn’t actually doing any banking: it’s just guarding the money, which would really be better understood (within this silly model) as a liability of the central bank that issued it and as an asset of the individual that rightfully holds that liability as a claim on the central bank (and has placed it in the local bank for safekeeping).
Sometimes, of course, they are not at all equal. Sometimes, the yield “curve” (constructed by plotting duration on the x axis and interest rates on the y axis) inverts, with long-term rates lower than short-term rates. In the US, this has been the case for almost two years now. Why this happens and what it means…cannot be answered in a brief footnote, and will have to wait for another day.
In this paragraph I echo and elaborate Levine’s primary term: “relationships.” The deposit franchise depends on the bank customer thinking about their banker a little bit like James Stewart in It’s a Wonderful Life. This also sheds light on the cliche of banks giving away things like toasters when you sign up for an account: the idea is to see your banker as anything but a capitalist looking to earn profit at your expense (they are just welcoming you to the community). I tend to read almost every sentence Levine writes with a subtle undertone of irony, and this very much applies to his emphasis on relationships. Levine knows that the banker really does want to build those relationships, not, however, for their own sake but for the sake of the very valuable (to the banker) deposit franchise.
This logic – customers seeing their deposits through the finance lens – has played out in other ways over the past few years, especially in the massive outflows of cash from low-interest savings accounts into high-yielding savings accounts and money market funds.
This brings up a cautionary reminder that I hope to write more about in the future: the phrase “printing money” evokes the idea of the printer (almost always a government or a bank) creating some quantum of positive value that they then use. The vision is deeply skewed because in reality the “printer” is always and only creating their own debt. Money creation is nothing like counterfeiting. In counterfeiting I create tokens of debt on some other debtor entity (usually the central bank), and then I try to spend them as if I were a valid creditor. This is fraud. But when a government or bank “prints money” they create their own tokens of debt. It’s quite true that the money printer and the counterfeiter can both spend the tokens they create, but the former still has outstanding debt claims while the latter does not.
One way to deprecate the significance of reserves is to point out that (in the US) commercial banks can borrow federal reserves on demand. Hence the reserve constraint is not binding in an economic sense: a bank can make loans all day long, never bring in any deposits, and then borrow reserves overnight from the Fed. That’s all true, and yet…no commercial bank ever wants to go to the Fed discount window and borrow directly from the central bank: doing so makes them look weak and it draws the scrutiny of regulators. This is just another version of Bagehot 101: “Every banker knows that if he has to prove that he is worthy of credit, however good may be his arguments, in fact his credit is gone.” Hence the reserve requirement does constrain, just not on strictly economic grounds. This makes perfect sense, because the bank is not a purely economic entity, but an utterly hybrid creature.
A “narrow bank” is not a real thing. Even posited in our philosophical imagination, a narrow bank is not a bank.
This isn't correct. Banks can't create money out of thin air. They are intermediaries. Please read the analysis here: https://www.worldsocialism.org/spgb/wp-content/uploads/2020/10/Banking-text.pdf