(Text of a talk I delivered at the Watson Institute for International and Public Affairs at Brown University on June 17, 2024.)
There is an odd dual quality to the world around us.
Consider a building. It has one, two or many stories; it’s made of wood, brick or steel; heated with oil or gas; with doors, windows and so on. If you could disassemble the building you could make a precise quantitative description of it — so many bricks, so much length of wire and pipe, so many tiles and panes of glass.
A building also has a second set of characteristics, that are not visible to the senses. Every building has an owner, who has more or less exclusive rights to the use of it. It has a price, reflected in some past or prospective sale and recorded on a balance sheet. It generates a stream of money payments. To the owner from tenants to whom the owner delegated som of their rights. From the owner to mortgage lenders and tax authorities, and to the people whose labor keeps them operating — or to the businesses that command that labor. Like the bricks in the building’s walls or the water flowing through its pipes, these can be expressed as numbers. But unlike those physical quantities, all of these can be expressed in the same way, as dollars or other units of currency.
What is the relationship between these two sets of characteristics? Do the prices and payments simply describe the or reflect the physical qualities? Or do they have their own independent existence?
My starting point is that this is a problem — that the answer is not obvious.
The relationship between money-world and the concrete social and material world is long-standing, though not always explicit, question in the history of economic thought. A central strand in that history is the search for an answer that unifies these two worlds into one.
From the beginnings of economics down to today’s textbooks, you can find variations on the argument that money quantities and money payments are just shorthand for the characteristics and use of concrete material objects. They are neutral — mere descriptions, which can’t change the underlying things.
In 1752, we find David Hume writing that “Money is nothing but the representation of labour and commodities… Where coin is in greater plenty; as a greater quantity of it is required to represent the same quantity of goods; it can have no effect, either good or bad.”
And at the turn of the 21st century, we hear the same thing from FOMC member Lawrence Meyer: “Monetary policy cannot influence real variables–such as output and employment.” Money, he says, only affects “inflation in the long run. This immediately makes price stability … the direct, unequivocal, and singular long-term objective of monetary policy.”
We could add endless examples in between.
This view profoundly shapes most of our thinking about the economy.
We’ve all heard that money is neutral — that changes in the supply or availability of money only affect the price level while leaving relative prices and real activity unchanged. We’ve probably encountered the Coase Theorem, which says that the way goods are allocated to meet real human needs should be independent of who holds the associated property rights. We are used to talking about “real” output and “real “ interest rates without worrying too much about what they refer to.
There is, of course, also a long history of arguments on the other side — that money is autonomous, that money and credit are active forces shaping the concrete world of production and exchange, that there is no underlying value to which money-prices refer. But for the most part, these counter-perspectives occupy marginal or subterranean positions in economic theory, though they may have been influential in other domains.
The great exception is, of course, Keynes. Indeed, there is an argument that what was revolutionary about the Keynesian revolution was his break with orthodoxy on precisely this point. In the period leading up to the General Theory, he explained that the difference between the economic orthodoxy and the new theory he was seeking to develop was fundamentally the difference between the dominant vision of the economy in terms of what he called “real exchange,” and an alternative he vision he described as “monetary production.”
The orthodox theory (in our day as well as his) started from an economy in which commodities exchanged for other commodities, and then brought money in at a later stage, if at all, without changing the fundamental material tradeoffs on which exchange was based. His theory, by contrast, would describe an economy in which money is not neutral, and in which the organization of production cannot be understood in nonmonetary terms. Or in his words, it is the theory of “an economy in which money plays a part of its own and affects motives and decisions and is … so that the course of events cannot predicted, either in the long period or in the short, without a knowledge of the behavior of money.”
*
If you are fortunate enough to have been educated in the Keynesian tradition, then it’s easy enough to reject the idea that money is neutral. But figuring out how money world and concrete social reality do connect — that is not so straightforward.
I’m currently in the final stages of writing a book with Arjun Jayadev, Money and Things, that is about exactly this question — the interface of money world with the social and material world outside of it.
Starting from Keynes monetary-production vision, we explore question of how money matters in four settings.
First, the determination of the interest rate. There is, we argue, a basic incompatibility between a theory of the interest rate as price of saving or of time, and of the monetary interest rate we observe in the real world. And once we take seriously the idea of interest as the price of liquidity, we see why money cannot be neutral — why financial conditions invariably influence the composition as well as the level of expenditure.
Second, price indexes and “real” quantities. The ubiquitous “real” quantities constructed by economists are, we suggest, at best phantom images of monetary quantities. Human productive activity is not in itself describable in terms of aggregate quantities. Obviously particular physical quantities, like the materials in this building, do exist. But there is no way to make a quantitative comparison between these heterogeneous things except on the basis of money prices — prices are not measuring any preexisting value. Prices within an exchange community are objective, from the point of view of those within the community. But there is no logically consistent procedure for comparing “real” output once you leave boundaries of a given exchange community, whether across time or between countries
The third area we look at the interface of money world and social reality is corporate finance and governance. We see the corporation as a central site of tension between the distinct social logics of money and production. Corporations are the central institutions of monetary production, but they are not themselves organized on market principles. In effect, the pursuit of profit pushes wealth owners to accept a temporary suspension of the logic of market – but this can only be carried so far.
The fourth area is debt and capital. These two central aggregates of money-world are generally understood to reflect “real,” nonmonetary facts about the world — a mass of means of production in the case of capital, cumulated spending relative to income in the case of debt. But the actual historical evolution of these aggregates cannot, we show, be understood in this way in either case. The evolution of capital as we observe it, in the form of wealth, is driven by changes in the value of existing claims on production, rather than the accumulation of new capital goods. These valuation changes in turn reflect, first, social factors influencing division of income between workers and owners and, second, financial factors influencing valuations of future income streams. Debt is indeed related to borrowing, in a way that capital is not related to accumulation. But changes in indebtedness over time owe as much to interest, income and price-level changes that affect burden of existing debt stock as they do to new borrowing. And in any case borrowing mainly finances asset ownership, as opposed to the dissaving that the real-excahnge vision imagines it as.
Even with the generous time allotted to me, I can’t discuss all four of those areas. So in this talk I will focus on the interest rate.
*
Some of what I am going to say here may seem familiar, or obvious.
But I think it’s important to start here because it is so central to debates about money and macroeconomics. Axel Leijonhufvud long ago argued that the theory of the interest rate was at the heart of the confusion in modern macroeconomics. “The inconclusive quarrels … that drag on because the contending parties cannot agree what the issue is, largely stem from this source.” I think this is still largely true.
Orthodoxy thinks of the interest rate as the price of savings, or loanable funds, or alternatively, as the tradeoff between consumption in the future and consumption in the present.
Interest in this sense is a fundamentally non-monetary concept. It is a price of two commodities, based on the same balance of scarcity and human needs that are the basis of other prices. The tradeoff between a shirt today and a shirt next year, expressed in the interest rate, is no different between the tradeoff between a cotton shirt and a linen one, or one with short versus long sleeves. The commodities just happen to be distinguished by time, rather than some other quality.
Monetary loans, in this view, are just like a loan of a tangible object. I have a some sugar, let’s say. My neighbor knocks on the door, and asks to borrow it. If I lend it to them, I give up the use of it today. Tomorrow, the neighbor will return the same amount of sugar to me, plus something extra – perhaps one of the cookies they baked with it. Whatever income you receive from ownership of an asset — whether we call it interest, profit or cookies — is a reward for deferring your use of the concrete services that the asset provides.
This way of thinking about interest is ubiquitous in economics. In the early 19th century Nassau Senior described interest as the reward for abstinence, which gives it a nice air of Protestant morality. In a current textbook, in this case Gregory Mankiw’s, you can find the same idea expressed in more neutral language: “Saving and investment can be interpreted in terms of supply and demand … of loanable funds — households lend their savings to investors or deposit their savings in a bank that then loans the funds out.”
It’s a little ambiguous exactly how we are supposed to imagine these funds, but clearly they are something that already exists before the bank comes into the picture. Just as with the sugar, if their owner is not currently using them, they can lend them to someone else, and get a reward for doing so.
If you’ve studied macroeconomics at the graduate level, you probably spent much of the semester thinking about variations on this story of tradeoffs between stuff today and stuff in the future, in the form of an Euler equation equating marginal costs and benefits across time. It’s not much of an exaggeration to say that mathematically elaborated versions of this story are the contemporary macro curriculum.
Money and finance don’t come into this story. As Mankiw says, investors can borrow from the public directly or indirectly via banks – the economic logic is the same either way.
We might challenge this story from a couple of directions.
One criticism — first made by Piero Sraffa, in a famous debate with Friedrich Hayek about 100 years ago — is that in a non monetary world each commodity will have its own distinct rate of interest. Let’s say a pound of flour trades for 1.1 pounds (or kilograms) of flour a year from now. What will a pound or kilo of sugar today trade for? If, over the intervening year, the price of usage rises relative to the price of flour, then a given quantity of sugar today will trade for a smaller amount of sugar a year from now, than the same quantity of flour will. Unless the relative price of flour and sugar are fixed, their interest rates will be different. Flour today will trade at one rate for flour in the future, sugar at a different rate; the use of a car or a house, a kilowatt of electricity, and so on will each trade with the same thing in the future at their own rates, reflecting actual and expected conditions in the markets for each of these commodities. There’s no way to say that any one of these myriad own-rates is “the” rate of interest.
Careful discussions of the natural rate of interest will acknowledge that it is only defined under the assumption that relative prices never change.
Another problem is that the savings story assumes that the thing to be loaned — whether it is a specific commodity or generic funds — already exists. But in the monetary economy we live in, production is carried out for sale. Things that are not purchased, will not be produced. When you decide not to consume something, you don’t make that thing available for someone else. Rather, you reduce the output of it, and the income of the producers of it, by the same amount as you reduce your own consumption.
Saving, remember, is the difference between income and consumption. For you as an individual, you can take my income as given when deciding how much to consume. So consuming less means saving more. But at the level of the economy as a whole, income is not independent of consumption. A decision to consume less does not raise aggregate saving, it lowers aggregate income. This is the fallacy of consumption emphasized by Keynes: individual decisions about consumption and saving have no effect on aggregate saving.
So the question of how the interest rate is determined, is linked directly to the idea of demand constraints.
Alternatively, rather than criticizing the loanable-funds story, we can start from the other direction, from the monetary world we actually live in. Then we’ll see that credit transactions don’t involve the sort of tradeoff between present and future that orthodoxy focuses on.
Let’s say you are buying a home.
On the day that you settle , you visit the bank to finalize your mortgage. The bank manager puts in two ledger entries: One is a credit to your account, and a liability to the bank, which we call the deposit. The other, equal and offsetting entry is a credit to the bank’s own account, and a liability for you. This is what we call the loan. The first is an IOU from the bank to you, payable at any time. The second is an IOU from you to the bank, with specified payments every month, typically, in the US, for the next 30 years. Like ordinary IOUs, these ledger entries are created simply by recording them — in earlier times it was called “fountain pen” money.
The deposit is then immediately transferred to the seller, in return for the title to the house. For the bank, this simply means changing the name on the deposit — in effect, you communicate to the bank that their debt that was payable to you, is now payable to the seller. On your balance sheet, one asset has been swapped for another — the $250,000 deposit, in this case, for a house worth $250,000. The seller makes the opposite swap, of the title to a house for an equal value IOU from the bank.
As we can see, there is no saving or dissaving here. Everyone has just swapped assets of equal value.
This mortgage is not a loan of preexisting funds or of anything else. No one had to first make a deposit at the bank in order to allow them to make this loan. The deposit — the money — was created in the process of making the loan itself. Banking does not channel saving to borrowing as in the loanable-funds view, but allows a swap of promises.
One thing I always emphasize to my students: You should not talk about putting money in the bank. The bank’s record is the money.
On one level this is common knowledge. I am sure almost everyone in this room could explain how banks create money. But the larger implications are seldom thought through.
What did this transaction consist of? A set of promises. The bank made a promise to the borrowers, and the borrowers made a promise to the bank. And then the bank’s promise was transferred to the sellers, who can transfer it to some third party in turn.
The reason that the bank is needed here is because you cannot directly make a promise to the seller.
You are willing to make a promise of future payments whose present value is worth more than the value the seller puts on their house. Accepting that deal will make both sides better off. But you can’t close that deal, because your promise of payments over the next 30 years is not credible. They don’t know if you are good for it. They don’t have the ability to enforce it. And even they trust you, maybe because you’re related or have some other relationship, other people do not. So the seller can’t turn your promise of payment into an immediate claim on other things they might want.
Orthodox theory starts from assumption that everyone can freely contract over income and commodities at any date in the future. That familiar Euler equation is based on the idea that you can allocate your income from any future period to consumption in the present, or vice versa. That is the framework within which the interest rate looks like a tradeoff between present and future. But you can’t understand interest in a framework that abstracts away from precisely the function that money and credit play in real economies.
The fundamental role of a bank, as Hyman Minsky emphasized, is not intermediation but acceptance. Banks function as third parties who broaden the range of transactions that can take place on the basis of promises. You are willing to commit to a flow of money payments to gain legal rights to the house. But that is not enough to acquire the house. The bank, on the other hand, precisely because its own promises are widely trusted, is in a position to accept a promise from you.
Interest is not paid because consumption today is more desirable than consumption in the future. Interest is paid because credible promises about the future are hard to make.
*
The cost of the mortgage loan is not that anyone had to postpone their spending. The cost is that the balance sheets of both transactors have become less liquid.
We can think of liquidity in terms of flexibility — an asset or a balance sheet position is liquid insofar as it broadens your range of options. Less liquidity, means fewer options.
For you as a homebuyer, the result of the transaction is that you have committed yourself to a set of fixed money payments over the next 30 years, and acquired the legal rights associated with ownership of a home. These rights are presumably worth more to you than the rental housing you could acquire with a similar flow of money payments. But title to the house cannot easily be turned back into money and thereby to claims on other parts of the social product. Home ownership involves — for better or worse — a long-term commitment to live in a particular place. The tradeoff the homebuyer makes by borrowing is not more consumption today in exchange for less consumption tomorrow. It is a higher level of consumption today and tomorrow, in exchange for reduced flexibility in their budget and where they will live. Both the commitment to make the mortgage payments and the non-fungibility of home ownership leave less leeway to adapt to unexpected future developments.
On the other side, the bank has added a deposit liability, which requires payment at any time, and a mortgage asset which in itself promises payment only on a fixed schedule in the future. This likewise reduces the bank’s freedom of maneuver. They are exposed not only to the risk that the borrower will not make payments, but also to the risk of capital loss if interest rates rise during the period they hold the mortgage, and to the risk that the mortgage will not be saleable in an emergency, or only at an unexpectedly low price. As real world examples like, recently, Silicon Valley Bank show, these latter risks may in practice be much more serious than the default risk. The cost to the bank making the loan is that its balance sheet becomes more fragile.
Or as Keynes put it in a 1937 article, “The interest rate … can be regarded as being determined by the interplay of the terms on which the public desires to become more or less liquid and those on which the banking system is ready to become more or less unliquid.”
Of course in the real world things are more complicated. The bank does not need to wait for the mortgage payments to be made at the scheduled time. It can transfer the mortgage to a third party, trading off some of the income it expected for a more liquid position. The buyer might be some other financial institution looking for a position farther toward the income end of the liquidity-income tradeoff, perhaps with multiple layers of balance sheets in between. Or the buyer might be the professional liquidity-providers at the central bank.
Incidentally, this is an answer to a question that people don’t ask often enough: How is it that the central bank is able to set the interest rate at all? The central bank plays no part in the market for loanable funds. But central banks are very much in the liquidity business.
It is monetary policy, after all, not savings policy.
One thing this points to is that there is no fundamental difference between routine monetary policy and the central bank’s role as a lender of last resort and a regulator. All of these activities are about managing the level of liquidity within the financial system. How easy is it to meet your obligations. Too hard, and the web of obligations breaks. Too easy, and the web of money obligations loses its ability to shape our activity, and no longer serves as an effective coordination device.
As the price of money — the price for flexibility in making payments as opposed to fixed commitments — the interest rate is a central parameter of any monetary economy. The metaphor of “tight” or “loose” conditions for high or low interest rates captures an important truth about the connection between interest and the flexibility or rigidity of the financial system. High interest rates correspond to a situation in which promises of future payment are worth less in terms of command over resources today. When it’s harder to gain control over real resources with promises of future payment, the pattern of today’s payments is more tightly linked to yesterday’s income. Conversely, low interest rates mean that a promise of future payments goes a long way in securing resources today. That means that claims on real resources therefore depend less on incomes in the past, and more on beliefs about the future. And because interest rate changes always come in an environment of preexisting money commitments, interest also acts as a scaling variable, reweighting the claims of creditors against the income of debtors.
*
In addition to credit transactions, the other setting in which interest appears in the real world is in the price of existing assets.
A promise of money payments in the future becomes an object in its own right, distinct from those payments themselves. I started out by saying that all sorts of tangible objects have a shadowy double in money-world. But a flow of money payments can also acquire a phantom double. A promise of future payment creates a new property right, with its owner and market price.
When we focus on that fact, we see an important role for convention in the determination of interest. To some important extent, bond prices – and therefore interest rates – are what they are, because that is what market participants expect them to be.
A corporate bond promises a set of future payments. It’s easy in a theoretical world of certainty, to talk as if the bond just is those future payments. But it is not.
This is not just because it might default, which is easy to incorporate into the model. It’s not just because any real bond was issued in a certain jurisdiction, and conveys rights and obligations beyond payment of interest — though these other characteristics always exist and can sometimes be important. It’s because the bond can be traded, and has a price which can change independent of the stream of future payments.
If interest rates fall, your bond’s price will rise — and that possibility itself is a factor in the price of the bond.
This helps explain a widely acknowledged anomaly in financial markets. The expectation hypothesis says that the interest rate on a longer bond should be the same as the average of shorter rates over the same period, or at least that they should be related by a stable term premium. This seems like a straightforward arbitrage, but it fails completely, even in its weaker form.
The answer to this puzzle is an important part of Keynes’ argument in The General Theory. Market participants are not just interested in the two payment streams. They are interested in the price of the long bond itself.
Remember, the price of an asset always moves inversely with its yield. When rates on a given type of credit instrument go up, the price of that instrument falls. Now let’s say it’s widely believed that a 10 year bond is unlikely to trade below 2 percent for very long. Then you would be foolish to buy it at a yield much below 2 percent, because you are going to face a capital loss when yields return to their normal level. And if most people believe this, then the yield never will fall below 2 percent, no matter what happens with short rates.
In a real world where the future is uncertain and monetary commitments have their own independent existence, there is an important sense in which interest rates, especially longer ones, are what they are because that’s what people expect them to be.
One important implication of this is that we cannot think of various market interest rates as simply “the” interest rate, plus a risk premium. Different interest rates can move independently for reasons that have nothing to do with credit risk.
*
On the one hand, we have a body of theory built up on the idea of “the” interest rate as a tradeoff between present and future consumption. On the other, we have actual interest rates, set in the financial system in quite different ways.
People sometimes try to square the circle with the idea of a natural rate. Yes, they say, we know about liquidity and the term premium and the importance of different kinds of financial intermediaries and regulation and so on. But we still want to use the intertemporal model we were taught in graduate school. We reconcile this by treating the model as an analysis of what the interest rate ought to be. Yes, banks set interest rates in all kinds of ways, but there is only one interest rate consistent with stable prices and, more broadly, appropriate use of society’s resources. We call this the natural rate.
This idea was first formulated around the turn of the 20th century by Swedish economist Knut Wicksell. But the most influential modern statement comes from Milton Friedman. He introduces the natural rate of interest, along with its close cousin the natural rate of unemployment, in his 1968 Presidential Address to the American Economics Association, which has been described as the most influential paper in economics since World War II. The natural rates there correspond to the rates that would be “ground out by the Walrasian system of general equilibrium equations, provided there is imbedded in them the actual structural characteristics of the labor and commodity markets, including market imperfections, stochastic variability in demands and supplies, the cost of gathering information … and so on.”
The appeal of the concept is clear: It provides a bridge between the nonmonetary world of intertemporal exchange of economic theory, and the monetary world of credit contracts in which we actually live. In so doing, it turns the intertemporal story from a descriptive one to a prescriptive one — from an account of how interest rates are determined, to a story about how central banks should conduct monetary policy.
Fed Chair Jerome Powell gave a nice example of how central bankers think of the natural rate in a speech a few years ago. He introduces the natural interest rate R* with the statement that “In conventional models of the economy, major economic quantities … fluctuate around values that are considered ‘normal,’ or ‘natural,’ or ‘desired.’” R* reflects “views on the longer-run normal values for … the federal funds rate” which are based on “ fundamental structural features of the economy.”
Notice the confusion here between the terms normal, natural and desired, three words with quite different meanings. R* is apparently supposed to be the long-term average interest rate, and the interest rate that we would see in a world governed only fundamentals and the interest rate that delvers the best policy outcomes.
This conflation is a ubiquitous and essential feature of discussions of natural rate. Like the controlled slipping between the two disks of a clutch in a car, it allows systems moving in quite different ways to be joined up without either side fracturing from the stress. The ambiguity between these distinct meanings is itself normal, natural and desired.
The ECB gives perhaps an even nicer statement: “At its most basic level, the interest rate is the ‘price of time’ — the remuneration for postponing spending into the future.” R* corresponds to this. It is a rate of interest determined by purely non monetary factors, which should be unaffected by developments in the financial system. Unfortunately, the actual interest rate may depart from this. In that case, the natural rate, says the ECB, “while unobservable … provides a useful guidepost for monetary policy.”
I love the idea of an unobservable guidepost. It perfectly distills the contradiction embodied in the idea of R*.
As a description of what the interest rate is, a loanable-funds model is merely wrong. But when it’s turned into a model of the natural rate, it isn’t even wrong. It has no content at all. There is no way to connect any of the terms in the model with any observable fact in the world.
Go back to Friedman’s formulation, and you’ll see the problem: We don’t possess a model that embeds all the “actual structural characteristics” of the economy. For an economy whose structures evolve in historical time, it doesn’t make sense to even imagine such a thing.
In practice, the short-run natural rate is defined as the one that results in inflation being at target — which is to say, whatever interest rate the central bank prefers.
The long-run natural rate is commonly defined as the real interest rate where “all markets are in equilibrium and there is therefore no pressure for any resources to be redistributed or growth rates for any variables to change.” In this hypothetical steady state, the interest rate depends only on the same structural features that are supposed to determine long-term growth — the rate of technical progress, population growth, and households’ willingness to defer consumption.
But there is no way to get from the short run to the long run. The real world is never in a situation where all markets are in equilibrium. Yes, we can sometimes identify long-run trends. But there is no reason to think that the only variables that matter for those trends are the ones we have chosen to focus on in a particular class of models. All those “actual structural characteristics” continue to exist in the long run.
The most we can say is this: As long as there is some reasonably consistent relationship between the policy interest rate set by the central bank and inflation, or whatever its target is, then there will be some level of the policy rate that gets you to the target. But there’s no way to identify that with “the interest rate” of a theoretical model. The current level of aggregate spending in the economy depends on all sorts of contingent, institutional factors, on sentiment, on choices made in the past, on the whole range of government policies. If you ask, what policy interest rate is most likely to move inflation toward 2 percent, all that stuff matters just as much as the supposed fundamentals.
The best you can do is set the policy rate by whatever rule of thumb or process you prefer, and then after the fact say that there must be some model where that would be the optimal choice.
Michael Woodford is the author of Interest and Prices, one of the most influential efforts to incorporate monetary policy into a modern macroeconomic model. He pretty explicitly acknowledges that’s what he was doing — trying to backfill a theory to explain the choices that central banks were already making.
*
What are the implications of this?
First, with regard to monetary policy, let’s acknowledge that it involves political choices made to achieve a variety of often conflicting social goals. As Ben Braun and others have written about very insightfully.
Second, recognizing that interest is the price of liquidity, set in financial markets, is important for how we think about sovereign debt.
There’s a widespread story about fiscal crises that goes something like this. First, a government’s fiscal balance (surplus or deficit) over time determines its debt-GDP ratio. If a country has a high debt to GDP, that’s the result of overspending relative to tax revenues. Second, the debt ratio determines to market confidence; private investors do not want to buy the debt of a country that has already issued too much. Third, the state of market confidence determines the interest rate the government faces, or whether it can borrow at all. Fourth, there is a clear line where high debt and high interest rates make debt unsustainable; austerity is the unavoidable requirement once that line is passed. And finally, when austerity restores debt sustainability, that will contribute to economic growth.
Alberto Alesina was among the most vigorous promoters of this story, but it’s a very common one.
If you accept the premises, the conclusions follow logically. Even better, they offer the satisfying spectacle of public-sector hubris meeting its nemesis. But when we look at debt as a monetary phenomenon, we see that its dynamics don’t run along such well-oiled tracks.
First of all, as a historical matter, differences in growth, inflation and interest rates are at least as important as the fiscal position in determining the evolution of the debt ratio over time. Where debt is already high, moderately slower growth or higher interest rates can easily raise the debt ratio faster than even very large surpluses can reduce it – as many countries subject to austerity have discovered. Conversely, rapid economic growth and low interest rates can lead to very large reductions in the debt ratio without the government ever running surpluses, as in the US and UK after World War II. More recently, Ireland reduced its debt-GDP ratio by 20 points in just five years in the mid-1990s while continuing to run substantial deficits, thanks to very fast growth of the “Celtic tiger” period.
At the second step, market demand for government debt clearly is not an “objective” assessment of the fiscal position, but reflects broader liquidity conditions and the self-confirming conventional expectations of speculative markets. The claim that interest rates reflect the soundness or otherwise of public budgets runs up against a glaring problem: The financial markets that recoil from a country’s bonds one day were usually buying them eagerly the day before. The same markets that sent interest rates on Spanish, Portuguese and Greek bonds soaring in 2010 were the ones snapping up their public and private debt at rock-bottom rates in the mid-2000s. And they’re the same markets that returned to buying those countries debt at historically low levels today, even as their debt ratios, in many cases, remained very high.
People like Alesina got hopelessly tangled up on this point. They wanted to insist both that post-crisis interest rates reflected an objective assessment of the state of public finances, and that the low rates before the crisis were the result of a speculative bubble. But you can’t have it both ways.
This is not to say that financial markets are never a constraint on government budgets. For most of the world, which doesn’t enjoy the backstop of a Fed or ECB, they very much are. But we should never imagine that financial conditions are an objective reflection of a country’s fiscal position, or of the balance of savings and investment.
The third big takeaway, maybe the biggest one, is that money is never neutral.
If the interest rate is a price, what it is a price of is not “saving” or the willingness to wait. It is not “remuneration for deferring spending,” as the ECB has it. Rather, it is of the capacity to make and accept promises. And where this capacity really matters, is where finance is used not just to rearrange claims on existing assets and resources, but to organize the creation of new ones. The technical advantages of long lived means of production and specialized organizations can only be realized if people are in a position to make long-term commitments. And in a world where production is organized mainly through money payments, that in turn depends on the degree of liquidity.
There are, at any moment, an endless number of ways some part of society’s resources could be reorganized so as to generate greater incomes, and hopefully use values. You could open a restaurant, or build a house, or get a degree, or write a computer program, or put on a play. The physical resources for these activities are not scarce; the present value of the income they can generate exceeds their costs at any reasonable discount rate. What is scarce is trust. You, starting on a project, must exercise a claim on society’s resources now; society must accept your promise of benefits later. The hierarchy of money allows participants in various collective projects to substitute trust in a third party for trust in each other. But trust is still the scarce resource.
Within the economy, some activities are more trust-intensive, or liquidity-constrained, than others.
Liquidity is more of a problem when there is a larger separation between outlays and rewards, and when rewards are more uncertain.
Liquidity is more of the problem when the scale of the outlay required is larger.
Liquidity and trust are more important when decisions are irreversible.
Trust is more important when something new is being done.
Trust is more scarce when we are talking about coordination between people without any prior relationship.
These are the problems that money and credit help solve. Abundant money does not just lead people to pay more for the same goods. It shifts their spending toward things that require bigger upfront payments and longer-term commitments, and that are riskier.
I was listening to an interview with an executive from wind-power company on the Odd Lots podcast the other day. “We like to say that our fuel is free,” he said. “But really, our fuel is the cost of capital.” The interest rate matters more for wind power than for gas or coal, because the costs must be paid almost entirely up front, as opposed to when the power is produced.
When costs and returns are close together, credit is less important.
In settings where ongoing relationships exist, money is less important as a coordinating mechanism. Markets are for arms-length transactions between strangers.
Minsky’s version of the story emphasizes that we have to think about money in terms of two prices, current production and long-lived assets. Long-lived assets must be financed – acquiring one typically requires committing to a series of future payments . So their price is sensitive to the availability of money. An increase in the money supply — contra Hume, contra Meyer — does not raise all prices in unison. It disproportionately raises the price of long-lived assets, encouraging production of them. And it is long-lived assets that are the basis of modern industrial production.
The relative value of capital goods, and the choice between more and less capital-intensive production techniques, depends on the rate of interest. Capital goods – and the corporations and other long-lived entities that make use of them – are by their nature illiquid. The willingness of wealth owners to commit their wealth to these forms depends, therefore, on the availability of liquidity. We cannot analyze conditions of production in non-monetary terms first and then afterward add money and interest to the story. Conditions of production themselves depend fundamentally on the network of money payments and commitments that structure them, and how flexible that network is.
*
Taking money seriously requires us to reconceptualize the real economy.
The idea of the interest rate as the price of saving assumes, as I mentioned before, that output already exists to be either consumed or saved. Similarly, the idea of interest as an intertemporal price — the price of time, as the ECB has it — implies that future output is already determined, at least probabilistically. We can’t trade off current consumption against future consumption unless future consumption already exists for us to trade.
Wicksell, who did as much as anyone to create the natural-rate framework of today’s central banks, captured this aspect of it perfectly when he compared economic growth to wine barrels aging in the cellar. The wine is already there. The problem is just deciding when to open the barrels — you would like to have some wine now, but you know the wine will get better if you wait.
In policy contexts, this corresponds to the idea of a level of potential output (or full employment) that is given from the supply side. The productive capacity of the economy is already there; the most that money, or demand, can accomplish is managing aggregate spending so that production stays close to that capacity.
This is the perspective from which someone like Lawrence Meyer, or Paul Krugman for that matter, says that monetary policy can only affect prices in the long run. They assume that potential output is already given.
But one of the big lessons we have learned from the past 15 years of macroeconomic instability is that the economy’s productive potential is much more unstable, and much less certain, than economists used to think. We’ve seen that the labor force grows and shrinks in response to labor market conditions. We’ve seen that investment and productivity growth are highly sensitive to demand. If a lack of spending causes output to fall short of potential today, potential will be lower tomorrow. And if the economy runs hot for a while, potential output will rise.
We can see the same thing at the level of individual industries. One of the most striking, and encouraging developments of recent years has been the rapid fall in costs for renewable energy generation. It is clear that this fall in costs is the result, as much as the cause, of the rapid growth in spending on these technologies. And that in turn is largely due to successful policies to direct credit to those areas.
A perspective that sees money as epiphenomenal to the “real economy” of production would have ruled out that possibility.
This sort of learning by doing is ubiquitous in the real world. Economists prefer to assume decreasing returns only because that’s an easy way to get a unique market equilibrium.
This is one area where formal economics and everyday intuition diverge sharply. Ask someone whether they think that buying more or something, or making more of something, will cause the unit price to go up or down. If you reserve a block of hotel rooms, will the rooms be cheaper or more expensive than if you reserve just one? And then think about what this implies about the slope of the supply curve.
There’s a wonderful story by the great German-Mexican writer B. Traven called “Assembly Line.” The story gets its subversive humor from a confrontation between an American businessman, who takes it for granted that costs should decline with output, and a village artisan who insists on actually behaving like the textbook producer in a world of decreasing returns.
In modern economies, if not in the village, the businessman’s intuition is correct. Increasing returns are very much the normal case. This means that multiple equilibria and path dependence are the rule. And — bringing us back to money — that means that what can be produced, and at what cost, is a function of how spending has been directed in the past.
Taking money seriously, as its own autonomous social domain, means recognizing that social and material reality is not like money. We cannot think of it in terms of a set of existing objects to be allocated, between uses or over time. Production is not a quantity of capital and a quantity of labor being combined in a production function. It is organized human activity, coordinated in a variety of ways, aimed at open-ended transformation of the world whose results are not knowable in advance.
On a negative side, this means we should be skeptical about any economic concept described as “natural” or “real”. These are very often an attempt to smuggle in a vision of a non monetary economy fundamentally different from our own, or to disguise a normative claim as a positive one, or both.
For example, we should be cautious about “real” interest rates. This term is ubiquitous, but it implicitly suggests that the underlying transaction is a swap of goods today for goods tomorrow, which just happens to take monetary form. But in fact it’s a swap of IOUs — one set of money payments for another. There’s no reason that the relative price of money versus commodities would come into it.
And in fact, when we look historically, before the era of inflation-targeting central banks there was no particular relationship between inflation and interest rates.
We should also be skeptical of the idea of real GDP, or the price level. That’s another big theme of the book, but it’s beyond the scope of today’s talk.
On the positive side, this perspective is, I think, essential preparation to explore when and in what contexts finance matters for production. Obviously, in reality, most production coordinated in non-market ways, both within firms — which are planned economies internally — and through various forms of economy-wide planning. But there are also cases where the distribution of monetary claims through the financial system is very important. Understanding which specific activities are credit-constrained, and in what circumstances, seems like an important research area to me, especially in the context of climate change.
*
Let me mention one more direction in which I think this perspective points us.
As I suggested, the idea of the interest rate as the price of time, and the larger real-exchange vision of which it is part, treats money flows and aggregates as stand-ins for an underlying nonmonetary real economy. People who take this view tend not be especially concerned with exactly how the monetary values are constructed. Which rate, out of the complex of interest rates, is “the” interest rate? Which f the various possible inflation rates, and over what period, do we subtract to get the “real” interest rate? What payments exactly are included in GDP, and what do we do if that changes, or if it’s different in different countries?
If we think of the monetary values as just proxies for some underlying “real” value, the answers to these questions don’t really matter.
I was reading a paper recently that used the intensity of nighttime illumination across the Earth’s surface as an alternative measure of real output. It’s an interesting exercise. But obviously, if that’s the spirit you are approaching GDP in, you don’t worry about how the value of financial services is calculated, or on what basis we are imputing the services of owner-occupied housing. The number produced by the BEA is just another proxy for the true value of real output, that you can approximate in all kinds of other ways.
On the other hand, if you think that the money values are what is actually real — if you don’t think they are proxies for any underlying material quantity — then you have to be very concerned with the way they are calculated. If the interest rate really does mean the payments on a loan contract, and not some hypothetical exchange rate between the past and the future, then you have to be clear about which loan contract you have in mind.
Along the same lines, most economists treat the object of inquiry as the underlying causal relationships in the economy, those “fundamental structural characteristics” that are supposed to be stable over time. Recall that the natural rate of interest is explicitly defined with respect to a long run equilibrium where all macroeconomic variables are constant, or growing at a constant rate. If that’s how you think of what you are doing, then specific historical developments are interesting at most as case studies, or as motivations for the real work, which consists of timeless formal models.
But if we take money seriously, then we don’t need to postulate this kind of underlying deep structure. If we don’t think of interest in terms of a tradeoff between the present and the future, then we don’t need to think of future income and output as being in any sense already determined. And if money matters for the activity of production, both as financing for investment and as demand, then there is no reason to think the actual evolution of the economy can be understood in terms of a long-run trend determined by fundamentals.
The only sensible object of inquiry in this case is particular events that have happened, or might happen.
Approaching our subject this way means working in terms of the variables we actually observe and measure. If we study GDP, it is GDP as the national accountants actually define it and measure it, not “output” in the abstract. These variables are generally monetary.
It means focusing on explanations for specific historical developments, rather than modeling the behavior of “the economy” in the abstract.
It means elevating descriptive work over the kinds of causal questions that economists usually ask. Which means broadening our empirical toolkit away from econometrics.
These methodological suggestions might seem far removed from alternative accounts of the interest rate. But as Arjun and I have worked on this book, we’ve become convinced that the two are closely related. Taking money seriously, and rejecting conventional ideas of the real economy, have far-reaching implications for how we do economics.
Recognizing that money is its own domain allows us to see productive activity as an open-ended historical process, rather than a static problem of allocation. By focusing on money, we can get a clearer view of the non-monetary world — and, hopefully, be in a better position to change it.
Also, in passing, I appreciate the fact that they coded the terms themselves and didn’t outsource the job to ChatGPT. I’ve seen a couple papers doing quantitative analysis of text, that use chatbots to classify it. I really hope that does not become the norm!
