On Debt: A Reply to Josh Mason

(pictures of money / flickr)

Josh Mason and I have been debating since my critique of David Graeber’s Debt went online a few weeks ago. We started at Crooked Timber — completists can read the original comments box exchange here.

At that point we switched to email, with several lengthy back-and-forths — the total wordcount a multiple of the original Jacobin piece. His post here at Jacobin sets out his position in the email exchange, and here I’ll set out mine. Throughout it’s been a friendly and productive discussion, and I’d like to echo upfront Josh’s kind words — I don’t think there are many economists of our generation I feel more in tune with, and if you don’t already read The Slack Wire, you should. As he said in the post yesterday, we’ve found a lot of common ground, but he intended to focus on the published piece as pose his disagreements as sharply as possible, even though that “doesn’t imply any larger divide between us.” This reply is in the same spirit.

The debate between Josh and I centers on the question of whether or not the distinction between a “commodity/fiat money economy” and a “credit money economy” is a useful one in understanding our present economic system and its history. He thinks it is so useful as to be the central dividing line in economics; I think it is liable to mislead. The rest of the disagreement comes, I think, because Josh conflates the commodity/fiat-credit money economy divide with other divides in economic thinking. So he seems to that if I challenge that distinction, I must be a quantity theorist, must believe that money is simply a veil, neutral in its economic effects, and must misunderstand how banking works. In fact we are on the same side in all those other dichotomies, but Josh for some reason continues to maintain that if I disagree over the core distinction, I must be standing on the other side of all the others. In what follows, I’ll establish (1) why I think the “commodity/fiat money economy” — “credit money economy” divide is a problem; and (2) that the rest of his criticisms rest on the conflations with other theoretical dichotomies.

I think our discussion has moved away from the book that started it. I’ve said what I want to say about Debt. The parts in my piece that praised the book and its author may have seemed disingenuous amidst the forthright criticism, but I meant them. I don’t mean to dissuade anybody from reading the book — it is a valuable contribution, and if many people have found it an entry-point into thinking politically about money, that’s great. As I said in my article, I strongly disagreed with some of the conclusions, and especially the way it frames its questions, but of course that doesn’t make it a bad book at all — a book that sets out a strong position is very useful even to those who think it is wrong, because it helps them to reflect on and improve their own position. If I thought Debt was a terrible book, I wouldn’t have bothered writing a critique. I didn’t feel the need to be overly generous, given the reception the book has received elsewhere, but in case there is any doubt, I have a lot of respect for David Graeber’s work. However, Debt doesn’t have much to say about what is at issue between Josh and me.

Types of Money and Types of System

I don’t at all disagree with the claim that there are different kinds of money that can usefully be described as “commodity-money,” “fiat-money,” and “credit-money.” I also don’t have a problem with building abstract models of ideal-type systems and calling them “credit-money economies” or “commodity-money economies.” What I disagree with is the view that it is useful to characterize actual, historically-existing monetary systems of the last few centuries in those terms. The ideal types might be helpful in setting out certain concepts with which we go on to look at how the real systems actually operate — but they leave a lot of work to do because they don’t say much about what happens between the ideal-types, so might not in fact be the best starting point. It is wrong, I think, to locate the essence of a system in one of its interdependent parts.

Commodity, credit and fiat aspects to money have existed together throughout capitalist history. Credit money and state token money already made up the bulk of actually-circulating money for much of the period of metallic standards. Even where coins circulated supposedly with face value equal to the value of the metal they consisted of, this was not easy for everyday users to ascertain and they relied in practice on marks from the mints, and of course there were always divergences between face value and value-as-bullion due to clipping, debasement etc. Thus state and commodity features were intertwined. Further, credit-money emerged as promises to pay in currency, promises which came themselves to circulate as money.

The key factor that made the metallic standards was convertibility – that there was some rate at which the mint or central bank was committed to exchange between bullion and coin, with notes denominated in terms of the coins. Bank credit-money also rested on convertibility, in that the backer was committed to redeeming it for coins or central banknotes at face value. Particular credit-monies became devalued as uncertainty emerged about the ability of the backer to redeem at face value.

It might be argued that the distinction between a “commodity-money economy” and a “fiat-money economy” lies in whether or not the state commits to convertibility with a commodity. But this was something that lasted, with interruptions, right up until 1971, well into the period of “modern money” For most countries, a metallic standard was not something that linked the currency directly to a commodity produced domestically. It was first and foremost a peg to foreign currencies. Thus the policy problematic of a commodity-standard has much in common with that of a fixed exchange rate. The value of gold was determined internationally, with production mostly outside Europe — it had little link to domestic costs in most countries. The value of circulating money was anchored to the value of a commodity, but quite loosely, at several removes, with the anchor ultimately making itself felt in deflationary crises usually associated with drains abroad out of the central bank — in other words, what would come to be known as balance-of-payments crises, a hazard that threatens any country with a fixed exchange rate.

It seems much harder to find a historical dividing line between the predominance of commodity-money and the predominance of credit-money. In Europe and elsewhere by the nineteenth century there was already a pretty wild ecosystem of financial institutions and credit monies, which had to be tamed by the state in the interest of stability. Credit-money was far from being antithetical to the gold standard, because the gold standard did not depend on gold actually circulating. I make reference in my article on Debt to the gold standard as a tool by which governments disciplined their domestic banks in order to foster credit money — to save it from its own tendency towards overextension and crisis. (I get this idea from a great paper by Samuel Knafo in the Review of International Political Economy, 13:1 (2006): “The gold standard and the origins of the modern international monetary system.”) I see the whole classical tradition of monetary theory as oriented around this question, mostly with reference to the British situation.

The formal gold standard helped build trust in private monies. Eventually that part of the scaffolding could be dismantled. But the disciplines on private banking remained, and continued to develop. And it is no accident that as the gold anchor fell away, governments made keeping a lid on inflation an overriding policy priority. An anchor to one commodity was replaced by an anchor to a basket of commodities — a flexible, moving anchor, to be sure, but one that was nevertheless designed to retain trust in the value of a national currency when most actually circulating monies denominated in that currency are private liabilities.

Crucially, it is still an anchor that continues to constrain policy. Josh accuses me of channeling technocratic conventional opinion with my talk of monetary policy using unemployment to discipline money-wages in pursuit of price stability. But obviously I’m not endorsing the policy — only trying to explain why such a framework has become dominant. This is why I keep using the line, paraphrased from Keynes, about states printing the money but not the price lists. So long as the price lists are out of their hands, policymakers have to act strategically to maintain the value of the currency in terms of commodities — not by fiat, but as actors with particular powers in a monetary system that is largely private. Such strategy pulls policy instruments in some directions rather than others. It seems to me the critical aspect of the relationship between money and the state in our time — and it is not one that shows up in a simplistic distinction between “commodity-” and “credit-money economies.”

So that is where I am coming from: our monetary regime now involves state management, a largely private banking system, and even a commodity anchor of a sort. All three aspects were also present under commodity standards, though related in different ways. Commodity-money, fiat-money and credit-money are symbiotic elements within the same system, and to reduce the system to one of its elements for the purposes of a crude pseudo-historical periodization misses what I take to be the most important elements of the actual historical evolution of capitalist money.

Capitalism as a Monetary Production Economy

I hope it is clear that none of the above depends on a commitment to the quantity theory, to the neutrality of money, to seeing the economy as essentially a barter system, or any of the other evils I am accused of. If you are satisfied on that, feel free to stop reading here — my positive argument is above; what follows is a long discussion of niggly little theoretical points.

Josh sets out a line of the “leading lights of critical economics” running through Marx, Keynes, Minsky and Mehrling. What these thinkers share is a vision of the capitalist economy as a “monetary exchange economy,” as contrasted with an orthodox vision of a “real exchange economy.” This is a genuine and long-established divide in economic theory — it is the line laid out by Schumpeter between Real Analysis, which “proceeds from the principle that all the essential phenomena of economic life are capable of being described in terms of goods and services, and of relations between them,” and Monetary Analysis, which “introduces the element of money on the very ground floor of our analytic structure and abandons the idea that all essential features of economic life can be represented by a barter-economy model.” (History of Economic Analysis, 1954, pp. 277–78)

(It may be worth mentioning that Schumpeter himself thought the mainstream had come to recognize that “essential features of the capitalist process may depend upon the ‘veil’ and that the ‘face behind it’ is incomplete without it,” so that “this is almost universally recognized by modern economists, at least in principle, and that, taken in this sense, Monetary Analysis has established itself.” (p. 278))

But this is uncontroversial — Josh and I are both on the side of the Monetary Analysts. It just doesn’t have much to do with the question of whether it is helpful to fit the economy into “credit-money” or “commodity-money” boxes. That Josh apparently thinks this is the same question helps pinpoint where we differ.

The slide in meanings begins because Josh associates Real Analysis with “beginning from exchange” and Monetary Analysis with “beginning from obligations”:

To clarify the point a little, let’s say there are two sequences. One (the “economics textbook” one) goes from exchange of goods, to exchange of tokens representing goods, to exchange of commitments to give tokens representing goods. The other (the “Debt” one, though as I’ll show in a moment it’s found in economics too) goes from a variety of incommensurable obligations of people to each other, to a single standard to measure different obligations, to tokens representing units of that standard, to the use of those tokens to allocate goods.

But then he goes on to talk about the distinction between C-M-C’ (commodity to different commodity via money) and M-C-M’ (money to commodity to more money) developed by Marx and used by Keynes (in a draft). But this distinction is about quite a different point — it has nothing to do with “beginning from obligations.” Marx and Keynes are distinguishing there between two approaches to monetary exchange: in the first, people sell what they have (often their labor-power) in order to buy a different commodity of the same value; in the second, capitalists use money to finance a production process to sell the output for more money.

The point is that the capitalist strategy both depends on and reshapes the sphere of exchange. Because the production of commodities has come to be organized by capitalist firms pursuing profit, exchange itself is misunderstood if it is seen as barter between people who simply happen to have different endowments of goods to trade. Therefore, both the price structure and what is produced reflect not just present allocations of goods and desires, but also expectations about future profitability. Again, this has nothing to do with starting from debt rather than exchange. Both Marx and Keynes do talk about credit and debt, but neither could be said to start with it. Marx is especially explicit about beginning from commodity exchange and bringing debt in later.

Josh and I are both in agreement that there is an important distinction between tracing the historical origins of money and explaining its present character and position within a broader social system. As Josh puts it, “The real debate is mostly about the logical order, with historical claims acting as stand-ins on both sides.” I would add (and I doubt Josh would disagree) that history is vital in discussing a dynamic system — but an evolutionary history, i.e. one which deals with the development of a system, rather than the origins of particular components.

When it comes to building a map of the system, I’m actually agnostic about starting points. Any vision of the capitalist monetary system is going to have to deal with exchange, and with debt. Josh believes it is important to start with the latter. As I see it, the big obstacle to starting your explanation of money with debt is explaining the quantity in which debt is denominated in. It seems hard to build up a theory of debt in a capitalist economy before you have established why people want to borrow and lend particular sums, how the lender comes by a sum to lend, and the relationships by which the borrower expects to be able to pay it back over time.

Josh’s answer seems to be that we start from “a variety of incommensurable obligations of people to each other, to a single standard to measure different obligations, to tokens representing units of that standard, to the use of those tokens to allocate goods.” But I can’t think of any reason why starting from “incommensurable obligations” is superior to starting from exchange of incommensurable goods — in fact you could argue that they amount to more or less the same thing, the former focusing on exchange of services and the latter focusing on goods. (Unless, of course, we are thinking of some hypothetical pre-capitalist society with blood and marriage obligations — in which case we seem to be falling back into the trap of looking for the origins of money  instead of building a model of capitalist money.)

In any case, neither Marx, Keynes, Minsky or Mehrling progresses from debt to money in the way Josh describes. Marx emphatically not — he is the most explicit of them all about the logic of presentation, and derives money from commodity exchange, though explicitly criticizing barter derivations along the way. Keynes begins the General Theory not only with exchange in place, but also already with entrepreneurs and workers, factors of production with pre-existing real factor costs, expectations of selling on a market, and so on. Where he delves most abstractly into “the essential properties of interest and money” in Chapter 17, he derives money from a situation in which many commodities have spot and futures markets. The Treatise on Money starts with chartalist, not credit, foundations, and sets up the value of money in terms of price lists. Minsky and Mehrling begin from more or less the same starting point as Keynes: every unit is a bank, but most units are plugged into flows from their activities in production for sale. I’m completely on board with Perry Mehrling’s “money view” — but it’s an answer to a different question. Mehrling is making a case for paying attention to liquidity, to the survival constraint imposed by needing to make contractual payments, against the financial view that liquidity will always be available, given complete markets, and therefore can be abstracted away from. It’s not an argument about money flows being logically prior to exchange — in fact it presupposes a developed system of production for exchange from which units ultimately derive their flows.

My argument, then, is that Josh has conflated two quite different distinctions: that between Real and Monetary analysis, and that between thinking about money from starting points in exchange and debt. To start with exchange is not necessarily to adopt Real analysis, to treat money as a neutral veil, etc. Our mutual heroes certainly didn’t.

Commodity-Money and the Quantity Theory

Josh also wrongly conflates his credit-money — commodity-money economy distinction with another divide in economic thought — that between quantity and anti-quantity theories of money’s value. This latter distinction is also well-defined and traced in Schumpeter’s History of Economic Analysis. There it is recognized as relating to quite a separate question to the Real-Monetary Analysis divide, though often we find the same authors on the same side of each divide. The quantity theory, remember, holds that the quantity of money explains the price level (and thus the value of money), given exogenous velocity and real output. Anti-quantity theories challenge the direction of causation between quantity of money and price level, and/or the exogeneity of one or both of the other two variables.

Josh writes that “a big part of our difference… comes from the different ways we group” commodity, fiat and credit money. He says that I group fiat and credit together, while he groups fiat with commodity money. In fact, as should now be clear, I think they are all separate categories. But one of  the reasons Josh groups fiat- and commodity-money is telling:

…both a commodity like gold and distinct tokens minted by the state exist in a distinct quantity at any given moment, independent of people’s decisions to hold them or spend them, while credit is created and destroyed in the course of the transactions its used for…

For Josh, the quantity theory evidently holds in a commodity- or fiat-money system, but not in a credit-money system. The former two systems are, he thinks, characterized by an exogenous money supply — i.e., one determined outside the private sphere of exchange — while the last is characterized by endogenous money, where private actors can expand the money supply to meet their needs. My own view is that the money supply in endogenous even where money is convertible into a money-commodity as described above. That is, the quantity theory does not hold in any case. How, then, does Josh get the impression that I keep making “statements of Friedman’s quantity theory of money,” despite my repeated explanations on this point?

Part of the answer is that Josh is convinced that pre-Keynesian classical monetary theory held that the gold standard established the price level because it kept the money supply constant. This is just not the case. It was certainly quite common to see the quantity theory as explaining the price level in the short run. But many theorists saw the quantity of money as endogenous at least in the long run. In Ricardian theory, the quantity theory holds in the short run, but the industry that produced the money-commodity will react to divergences between its cost of production and the value of money. If the value of money falls, the least profitable mines will begin to close; when it rises, more marginal mines are brought back into operation. Alternatively, in a world with many countries on the commodity standard, it need not be produced within a particular country. In that case, the money-commodity will flow in and out in response to price-level differentials — Hume’s old story of the price-specie-flow mechanism. In either case, the quantity of money is not exogenous.

When you also consider the fact that under the gold and silver standards there was already a sophisticated banking system capable of producing private credit-money, it is even more clear that there is no clear tie between a commodity standard of value and an exogenous money supply.

It is hard to see how Josh’s framework can incorporate the likes of Marx, who was both insistent that money must be based on a commodity standard, and a fierce opponent of the quantity theory. Josh recognizes Marx as a great theorist of credit-money, but has to twist his argument to shoehorn it into his framework. He quotes Marx’s famous line that “the monetary system is essentially a Catholic institution, the credit system essentially Protestant,” but relegates the other half of the remark to a footnote: “But the credit system does not emancipate itself from the basis of the monetary system any more than Protestantism has emancipated itself from the foundations of Catholicism.” Yet Marx’s main point in the whole passage is to assert the anchoring of “the credit system” in a “monetary system” — the second part is the thrust of the argument, not the first, as you can judge for yourself.

The confusion arises from the fact that Marx simply doesn’t mean what Josh means by “credit system” and “monetary system.” For Josh these are ideal types of different systems. That is, he interprets Marx as doing something similar to Wicksell in his descriptions of the “pure credit” and “pure cash” economies. For Marx they are interrelated sub-systems sharing a social space: the credit system is something that emerges from the monetary system but remains tied to it. The credit system can expand the money supply and raise the velocity of existing money. (In fact it can be hard to tell these processes apart.) But it remains anchored in the money system in the sense that these expansions are still limited by the perceived ability of units to make payments as they come due.

This is exactly the point I was making in the passages Josh finds so contentious: “however far credit may stretch money, it still depends on a monetary base: people ultimately expect to get paid in some form or other.” And:

What circulates need not be a physical thing, but it is a thing in the sense that it cannot be in two places at once: when a payment is made, a quantity is deleted from one account and added to another. That the thing that is accepted in payment may be a third party’s liability does not change this fundamental point.

For Josh these are “statements of Friedman’s quantity theory, refuted by generations of Post Keynesian economists but still carrying on its zombie existence in the textbooks.” This is just not the case — the quantity theory posits a stable relationship between an exogenous quantity of money and the price level. (Friedman’s version is a little more complex, but that’s beside the point.) My statements do nothing of the sort. It is just that my argument, like Marx’s, does not compute within Josh’s frame of reference, conflating as he does several questions within the single credit-money — fiat/commodity-money binary.

Josh has either been reading the wrong Post Keynesians, or misinterpreting the right ones. My argument fits squarely in the structuralist tradition of Minsky, Victoria Chick and others, who emphasize both the elasticity of credit-money and its constraints. Josh objects to my talking about money as a thing, or perhaps jumps from my statement that a payment involves a “quantity” being deleted from one account and added to another, to a conclusion that I am talking about an exogenously determined quantity of money. This is to confuse two quite different frames: to jump from a discussion of what a payment is, to a macroeconomic theory!

There is a slippage in Josh’s terminology between “credit” and “credit-money.” He writes that while a money-commodity and state monetary tokens “exist in a distinct quantity at any given moment, independent of people’s decisions to hold them or spend them, while credit is created and destroyed in the course of the transactions it’s used for.” (Emphasis added.) It may be true that credit is created and destroyed by loans and repayments, but credit-money takes on a life of its own, circulating among parties who were not party to the original debt relationship.

The distinction is subtle, but of huge importance to understanding our monetary system. When a bank lends a customer money, it creates both credit and credit-money. The credit is something the borrower owes the bank, and remains until that particular borrower repays it. The credit-money is the increment to the borrower’s bank account created at the same time. The borrower spends this money. If the receiver happens to be a customer of the same bank, the amount paid is deleted from the borrower’s account and added to the receiver’s. But if the receiver is a customer of a different bank, the bank must settle that payment with the other bank. Eventually, of course, even if the original receiver is a customer of the lending bank, they themselves make payments, to people who also make payments, and it is highly unlikely that all these payments will be circulating among customers of the same bank.

There are thus two hierarchical levels of payments happening here, both involving deletions and additions to accounts. There is the level of transactions between customers, and the level of settlement between banks, involving their accounts with the central bank. Of course, there are many payments back and forth between customers of different banks every day, and they don’t all require flows of bank reserves: banks settle those that don’t cancel out on a daily basis. But a bank’s decision to extend credit depends partly on a liquidity constraint — the ability to meet reserve drains to other banks. There is nothing technically stopping a bank from doubling its balance sheet in a day — it can create all the deposits for its customers that it likes — but it would quickly get into trouble as those customers started to spend their money with customers of other banks.

In practice, banks don’t perceive a hard limit on reserves. It is routine to borrow reserves from surplus banks or from the central bank, and they think in terms of the cost of borrowing such reserves rather than the absolute availability. It is normal practice for the central bank to target an interest rate on this market and let quantities adjust, rather than maintain a fixed quantity of reserves. But the fact remains that every unit has to make payments and manage its liquidity in terms of some form of money it does not create itself — except for the central bank, and the central bank is constrained by its strategic aims. This is the point I was making, and not that there is some fixed quantity of money at a macroeconomic level.

Let’s take a look at the example Josh uses in making his objection to my treating a payment as a movement of a thing:

Open your wallet again: Yes, you see things called dollars, but most likely you also see a piece of plastic labeled Visa or Mastercard. This is money too — you can buy almost anything with it that you can buy with the bills. When you do so, new monetary liabilities are created on the spot, linking you to your bank and your bank to the vendor. Nothing is deleted from anywhere. (Emphasis added.)

This is only true if you and the vendor share the same bank, and the vendor does not use the addition to their funds to make payments to customers of another bank. The bank pays the vendor immediately, and you pay the bank later. The credit card changes nothing fundamental that was not made possible by, say, a bill of exchange or a check under the gold standard.

Josh refers here and elsewhere to Wicksell’s “pure credit economy.” There are some problems with his interpretation. The “pure credit economy” was for Wicksell an ideal type and not a description of the actual monetary system he saw around him, which he thought was somewhere between his ideal types (pure cash and pure credit). As Leijonhufvud himself argues in the paper Josh refers to, this makes it difficult to use to draw conclusions about the real world.

In creating his ideal type “pure credit economy” Wicksell makes the crucial assumption that “the whole monetary system is in the hands of a single credit institution” (Interest and Prices, p. 71). Of course this gets rid of the major liquidity constraint on individual banks in the real world — that they need to be able to settle net outflows to other banks. Even so, even in Wicksell’s pure credit economy the one big bank still has to maintain a gold reserve to meet drains abroad. This was a serious real-world constraint on national banking systems under a fixed exchange rate, at least for smaller countries.

Josh argues that we have over time evolved in the direction of Wicksell’s ideal type, and would have reached it long ago if, in Leijonhufvud’s words, “Wicksell’s “Day of Judgment” when the real demand for the reserve medium would shrink to epsilon” had not been “greatly postponed” by government decisions to “monopolize the note issue and to impose reserve requirements on banks.” But the critical limit here is not reserve requirements — i.e. the stipulation that banks must actually hold a certain ratio of reserves to deposits. This went out many years ago in a number of countries. The critical limit is the need for banks to make interbank payments in reserves. This limit will not pass until there is One Big Bank, as Wicksell knew.

The irony here is that Josh is thinking in terms of a quantity on bank balance sheets, whereas I believe it is important to think of liquidity in terms of flows, as I have written elsewhere. Liquidity is about the need for a unit to make payment commitments. Banks no longer manage their liquidity simply by holding reserves — it’s a rather expensive way to do it. Instead, they work with a range of liquidity strategies, often, as Minsky highlighted, on the liability side of the balance sheet, by borrowing when necessary. Nevertheless, however flexible liquidity has become, the need to make payments with some external thing remains a constraint on every private unit. The vision that Josh is hinting at, of banks freed from liquidity constraints, is the “finance view” that Perry Mehrling’s “money view” takes aim at.


This post is now almost as long as my original Jacobin piece on Debt. There is more I could quibble with in Josh’s reply — such as his interpretation that when I said that the state prints the money but not the price lists, I meant “interest rates.” (Again, I mean that the state does not control by fiat the value of money because that depends on the prices sellers place on their commodities. Policy may engage in strategic action to influence the price level, but doing so makes demands on policy instruments which may conflict with other goals, and which may in any case be unsuccessful.) But the main points have been made:

  1. I think theoretical distinctions between a “credit-money economy,” “commodity-money economy,” and “fiat-money economy” hide more than they illuminate about how money works in actual capitalist economies, now and historically. Credit, commodity and fiat aspects to money have always been interwoven.
  2. Josh’s other arguments depend on conflating the above view with other views that I do not hold, such as the quantity theory and monetary neutrality.