Modern Money Theory Misunderstands the “Magic” of Money

I had already queued this post up for delivery as part of my ongoing series responding to reader questions (see below), but then Stephanie Kelton weighed in on Musk’s “magic money computers” quote. For Kelton, Musk was basically right:

The point is, there’s no “bombshell” in what Musk told Senator Cruz. As Mosler likes to say, “the government spends by changing numbers (up) in bank accounts.” The money doesn’t “come from” anywhere. It materializes out of nothing, just like the points that are awarded by the scorekeeper in a football match (my emphasis)

I’ve already had a go at the sports points metaphor, but here I want to attack the larger claim: Kelton is agreeing with Musk that governments have some magic power that creates money out of nothing.

In the Substack notes app I have, over the past few months, repeatedly linked to great pieces by Kelton. And here again, she’s right about one thing: there’s a certain magic involved in modern money. Nonetheless, she is utterly wrong about the nature of that magic, and this misunderstanding has serious implications.

“Magic Beans”

I will unpack this argument by attending to a reader question that could itself be posed as a direct attack on MMT. After all, if the government has magic computers that create money out of nothing, then (as a reader wrote to me):

why does the US state issue new debt at all?

The best way to grapple with this question is to go back to the start and build out three different conceptual models of money, “the economy,” and the state.

1. Orthodoxy

The first move of the traditional account directly analogizes the state budget to the budget of an individual household. At the level of household finance, you cannot spend more than you earn – or, at least, you cannot do so indefinitely. Running a budget deficit can only be temporary and it will always “cost you,” in that you will have to pay back in the future whatever you borrow today. Hence the constant evocations of our grandchildren and the terrible woes we are wrecking upon them.

This foundational position entails, in turn, that if the state has an overall debt, it should run annual budget surpluses until it “pays back” that debt. There are a bunch of related arguments (e.g., the “crowding out” thesis) and theoretical assumptions (e.g., a commodity theory of money) that appear under this heading, but we will ignore all that, because the major questions raised today about state debt and state borrowing, about money and money markets, center more on the critique of orthodoxy.

2. MMT

The first move of orthodoxy is wrong. Keynes understood this; Keynes proved it conceptually; Keynesian economics and state policy has proved it in practice for almost one hundred years.¹ These facts are the starting point for the recent rise of MMT as (fundamentally, and most tenably) an alternative approach to budget deficits (and secondarily, less tenably, an alternative theory of money). MMT clearly draws out the key point: household “debt” is absolutely nothing like state “debt,” for numerous reasons:

1. What the state owes, it owes to its own citizens.

2. A society’s “money” is circulating state debt; with no state debt there would be no money.

3. The state’s sovereign powers extend to its capacity to issue tokens of its own debt as money, and this means that in order to pay back debt it owes, the state can always issue more money. In other words, unlike a household, the state cannot go bankrupt.

Adding it all up leads to the annihilation of the orthodox position: it’s not just that the state can run a deficit – they need to maintain debt in order to maintain the monetary system.

Now, MMT is often accused of naively declaring that “deficits don’t matter,” but that’s unfair: given the political context, the weight of the traditional view, and especially its intuitive appeal (if I have to balance my budget, why shouldn’t Washington) MMTers have spent many years taking the “run a deficit” side of the political argument. But from the beginning they gave an account that explained when it made more sense to increase deficits (when there is slack in the economy) and when it made more sense not to (when there is no such slack). And they rightly argued that inflation is not caused directly by deficits or by “printing money”; it’s caused when an economy running at full capacity continues to expand credit.

Nonetheless, our reader’s line of thinking (in the quote above) clearly channels key MMT logic. MMT insists that the US state never needs to borrow first:² if they want to pay for something they just issue dollars. And they can do this, even to pay for, or pay off, their own debt: just issue dollars.

In a previous piece on MMT, I described the flatness (and flattening force) of this approach:

MMT advances this equation: Debt = Money = Cash. That is, the government is a “money issuer”; it issues debt (in the form of “currency”) for the benefit of “money users,” who use the currency as their cash (for buying things). This equation explains the quasi-ledger you will see over and over again in the film [Finding the Money]: currency issuer on the left (–$10) and currency user on the right (+$10), symbolized with green circles to connote the cash.

To MMT, US sovereign bonds are a distraction whose significance authors like Stephanie Kelton are always trying to downplay. Kelton argues that when the Bank of Japan (BOJ) buys trillions of yen worth of Japanese government bonds it actually eliminates that debt:

while Japan is often described as the most indebted developed country in the world, half of its debt has already been essentially retired (i.e., paid off) by its central bank. (Kelton, The Deficit Myth)

We are now in a position to reformulate our reader’s question in MMT terms: why sell bonds at all if you are always capable of: (a) paying for things without bonds – just issue money; and (b) paying off all the debt in a stroke – just issue money? And from within the MMT paradigm the questions start to sound rhetorical: the argument is that there’s no need to deal bonds/debt at all – just issue money?

3. Money/Power

But what does it mean to issue money? Kelton and MMT consistently position the sovereign nation-state as a “currency-issuing government” and they constantly suggest that the issuance of currency is (almost aways) a very good thing (a true public good performed by the state). MMT dramatically sweeps the spotlight away from debt – and therefore credit as the correlative of debt, and therefore money as a relation of credit/debt – and onto “currency,” which is created by the state (the “currency issuer”) and utilized by citizens and firms (“currency users”).

MMT is right to challenge the orthodoxy on so many fronts. But they replace a false theory of money with a naive and distorting alternative. Money is not a thing; it’s not a device or tool that gets issued and used (like a floor sander you could rent from your local home improvement store). Money is always a relation of credit/debt. MMT likes to talk about “currency” that the government issues, but that’s not actually a thing.

Here are some actual things, that is, some actual “government-related” (see below) money-credits:

1. Federal Reserve physical banknotes.

2. Federal Reserve digital reserves.

3. US Treasury Bonds.³

We tend to think of all of these as “money” and also as “government issued” and often as “government debt.” Much of the time those assumptions work out fine and the technical differences don’t matter much, but this is not one of those times. At the end of the day, (1) and (2) are liabilities of a private network of national banks, while (3) is the direct debt of the US state.

It turns out, contrary to both MMT and our everyday intuitions, the US government does not pay for stuff just by “issuing dollars.” Rather, to a certain extent, the US state pays for stuff the same way we all do: by holding their money in a bank and ordering the bank to pay. The US Treasury has a deposit account at the Fed. My money is my claim on Ally Bank. The United States’ money is a claim on the Fed. When I get paid by my employer, Ally credits my account, and when I pay for stuff they debit it.

Indeed, let’s assume that when I do my taxes next month I owe $1,000. To pay them, I will instruct Ally to make a digital transfer to the IRS, and after I do, 3 things will happen:

1. Ally will reduce my deposit account by $1,000.

2. The Fed will reduce Ally’s fed reserves (deposit) account by $1,000.

3. The Fed will increase the US Treasury’s fed reserves (deposit) account by $1,000.

Obviously, then, when the US government pays for things, their fed account diminishes. And what happens if they start to run out of money? They sell bonds!⁴ Selling a bond to me increases the Treasury’s account at the Fed in exactly the same way as my paying taxes.

So why does the Treasury sell bonds? Why do they borrow at all?

First, the Federal Reserve and the US Treasury have separate balance sheets. This is built into the architecture of the US financial system. When I say “system” here I mean both political and economic; this is a money/power system. So, if the government decides to build a road, or to pay an Elon Musk company billions of dollars in the form of Starlink contracts for rural broadband or SpaceX contracts to send up government satellites – those are not liabilities of the Fed. When the Fed earns interest on loans it makes to commercial banks through the discount window, that is not an asset of the US government.

Now, of course we know that the Fed (like other government agencies) is in some sense “backed” by the US government. But only a separate entity can serve as a “backstop.” Having separate balance sheets provides a crucial precondition that makes it possible for the US government (the Treasury) to run a unique balance sheet. Put differently, the fact that the Fed’s balance sheet balances makes it possible for the US government to run deficits and always be in debt.

---

Let me unpack this crucial point, one that, I believe, MMT often overlooks and almost always to its peril. First, the Fed is a liquid bank, and also a solvent bank. That is, first, as MMT likes to point out, because the Fed itself can always issue more of the highest form of “currency” in the US, they can never become illiquid (i.e., run out of “cash”). But more than this, second, the Fed always remains solvent: its assets exceed its liabilities. The US state (through the Treasury) remains subject to a certain financial discipline in the form of their own bank, the Fed. They cannot authorize the Fed to pay for things unless and until they put money in their Fed account.

On this basis, the Treasury sells bonds and racks up a massive number in the liabilities column of the separate government balance sheet – currently $36.5 trillion. To repeat, that number is not a liability of the Fed. It doesn’t show up directly on the Fed balance sheet, although, importantly, it does show up indirectly, in that part of the Fed’s assets are its holding of Treasuries. But the mutuality between the Fed and the Treasury underscores the importance of the separation of their balance sheets.

MMT conflates the power of a bank to “create money” with the power of a government to “raise money.” The MMT distinction between money-issuer and money-user turns money into an object of positive value that, once created, simply gets passed from hand to hand. Just think of the constant refrain of the government “printing money.” But this is not what banks actually do. And it’s not even what governments do.

The magic of money is not the power to create value out of the ether. No one has that power. Rather, money magic is banking magic, and the magic of banking is the swap of IOUs; this swap is what creates money.

But when a commercial bank conducts this magic, they do not create positive value out of nothing; they create money (and with it, purchasing power) out of mutually constituting credit and debt relations. You have money now in your deposit account because the bank owes you, but the bank only agrees to owe you because you also owe it in the form of the loan you took out.

The stability and legitimacy of the financial system depends upon the fact that the government never actually “prints money.” They let a technically private, central bank create money the same way that commercial banks do. The government raises money through bond sales to the public. If the fed then goes out and buys those bonds on the secondary market, this does not make the debt go away, it just changes who the government owes. But it matters very much that the government still owes the Fed, because it is on the basis of the Fed owing the government (in the form of the Treasury deposit account at the Fed) that the government pays for things in the first place.

It is on the basis of the confusion between commercial bank money-creation and sovereign state debt-issuance that Kelton can make the ridiculous claim that when a central bank buys the bonds of its own government it “retires the debt.” To think of the US government debt as retired when the Fed buys bonds is to conceive of the Fed as insolvent: if those bonds are no good (because the US state doesn’t really owe them) then the Fed’s balance sheet is immediately massively underwater. And it is not at all clear to me that a nation’s sovereign monetary system can survive an insolvent central bank.

---

Why does the US state issue debt?

Ironically, neither the orthodox nor the MMT view can provide a robust answer to that question. Orthodoxy assumes debt must be, at best, temporary, at worst, immoral and fiscally dangerous. It imagines budget surpluses and no debt at all. Orthodoxy gets it wrong from the start, because it cannot imagine a state that continues to owe. MMT nicely refutes the fiscal discipline of orthodoxy, but indirectly agrees with them. Orthodoxy says the state should not borrow (too much, for too long), while MMT redescribes state actions as not really even borrowing at all. The state doesn’t actually borrow money; they just issue currency. MMT pretends that the state never really owes and certainly never needs to pay back.

Viewed through the money/power lens we see that the state really does issue debt, and they must do so in order to sustain a monetary system with a central bank at its pinnacle. Further, the central bank really is a bank, and must therefore be solvent; its assets exceed its liabilities. The soundness of the central bank’s balance sheet enables and sustains the unique balance sheet presented by sovereign states: their money liabilities are never matched by money assets (they are always technically insolvent, because they always have debt) but rather by non-money assets that only they possess.

The “balance sheet” of a government is a strange beast, because to truly understand it we need to place a lot of non-money-stuff on the assets side. This includes not only the power to police and the power to tax, but also the unique power to project a future in which those capacities can be used to honor any debt. If the entire monetary system depends on the existence of the government, then we can project that existence into the future as a guarantee of future repayment of debt.⁵ This leads us to describe the asset side of the state’s balance sheet as the place where money and power meet. Therefore the shared mutuality between the central bank and the state – a relation that depends on the state issuing debt – must be grasped as a, perhaps the key site of money/power.

We might say that sovereign monetary power is a lot like the House of Lannister: they want and need for people to say of them, “they always pay their debts,” yet their real motto is hear me roar!

1. None of this is to suggest that no one prior to Keynes understood that state budgets were different from household budgets in significant and important ways. But in order to explore the basic question before us, we need neither to render Keynes a scientific discoverer nor to do the genealogical work of locating his precursors.

2. Much hinges on where the emphasis lies in this phrase. I would describe “never needs to borrow first” as correct, and at times MMT limits itself to this claim, but often they veer into “never needs to borrow first.” As I am trying to show here, the latter claim fails to describe contemporary money practices, and proves conceptually untenable. A separate, but related issue concerns what the state was doing in, say, the early 17th century, prior to the invention of the central bank. At this point in time there was only one balance sheet, that of the state, and unlike the technical reality today, it really did pay for things in its own money (minted coins and printed notes). MMT would describe such payment as “issuing money” and “paying” simultaneously, and thus absolutely not “borrowing first” - because they were merely using money they created. I would describe such payment as “borrowing” and “paying” simultaneously – because their money is their debt.

3. Standard nomenclature here might refer to these as “bills, notes, and bonds” i.e. the formal names of these money instruments are “US Treasury Bills,” “US Treasury Notes” and “US Treasury Bonds.” Historically the Treasury refers to their very short term debt as “bills,” the medium term as “notes” and the longer term as “bonds.” I just say “bonds” in the text above because that accurately describes the nature of these financial instruments. A note is exactly like a bond, and a bill is effectively like a bond. (A bill lacks a coupon – bills sell below par at issuance – but so would a bond purchased in the secondary market less than 6 months before maturity.)

4. Because of a very weird and stupid political history, sometimes they have to ask Congress for approval first, before they can even issue the bonds – but that’s another story entirely.

5. The “consol” might be the perfect embodiment/metaphor here: this is a government bond that never matures. You loan the government money it is never required to pay back: instead you just get a stream of coupon payments, forever. Consols are not hypothetical: they funded the British Empire at its peak.