March 2, 2022
Perry Mehrling, Boston University
In his recent book Raising Keynes, Stephen Marglin presents himself as an outsider attacking the citadel of mainstream economics. The biggest thing in the book is his emphasis on dynamics, which supports an argument that unemployment is a possible equilibrium position even without any frictions (Ch. 6). That’s certainly an attack on the citadel. Another big thing is the addition of capital accumulation and population growth, which supports an argument that unemployment is a possible equilibrium even in the long run (Ch. 18). That’s also certainly an attack on the citadel.
But my interest here is with the money chapters (11-13) where Marglin treats Keynes’ theory of liquidity preference, a key building block of the Keynesian corpus since it was meant as an alternative to the classical view that the rate of interest equilibrates desired investment and desired saving. As we shall see, Marglin’s recasting of Keynes in these chapters is quite explicitly in the spirit of Tobin’s 1958 “Liquidity Preference as Behavior Toward Risk,” which, though perhaps not the citadel today, was certainly the citadel in the heyday of Keynesian orthodoxy. On the matter of money, Marglin can thus be read more as a defense of that (older) citadel than as an attack.
To be clear, Marglin’s goal in the money chapters is to shore up Keynes’ theory of liquidity preference, which he finds wanting, as part of his larger agenda to write “A Twenty-First-Century General Theory.” That’s what Tobin was trying to do also. For present purposes, however, it is important to put those good intentions aside and to look at the proposed theory on its merits. As will be seen, I count myself a sympathetic critic, but a critic nonetheless.
***
When Chris Desan asked me to take part in this Symposium, I felt an obligation to say yes, not only because I have spent the last forty years of my life trying to construct an alternative approach to monetary theory that I call the “money view,”[1] but also because Marglin was instrumental in setting me on that path. In spring 1981 I was a senior in college and, having decided to become an economist, asked him to supervise an independent study on the foundational texts of the field (such was my naivete about the mores of the field!). That’s when I first read Keynes’ General Theory, and I wrote a 24-page paper “Keynes on Uncertainty and Expectations,” which emphasized liquidity preference as the rational response to living in a world of uncertainty, not just risk. Two years later, after a Masters in Mathematical Economics and Econometrics at the LSE, I returned to Harvard for the PhD and immediately set to work on what would be my dissertation, “Studies in the Credit Theory of Money,” for which Marglin was one of the committee members, along with Ben Friedman and Jim Duesenberry. I note in his acknowledgments that Marglin ran chapters 11-12 past Friedman, so I suppose those chapters must have passed muster with my second Doktorvater as well, and accordingly, I must tread carefully!
Marglin’s Chapter 11 recasts what Keynes says in The General Theory, where he treats a world with two financial assets, money and bonds, and Chapter 12 then considers how to extend that theory for a world more like our own, where the safe liquid asset is an interest-bearing Treasury bill. Chapter 12 is thus intended to be the “Twenty-First-Century” version of liquidity preference theory, what Keynes should have written had he had the necessary mathematics, and had he been considering our world rather than the rather special conditions of the Great Depression. It’s the longest chapter in the book, including long mathematical and empirical appendices, as well as a centrally important one for Marglin’s project. In his reading, the monetary theory Keynes develops in The General Theory was really just “an afterthought” to the rest of the system,[2] and when we look at it closely, it turns out to be nothing more than a theory of interest rate spreads which leaves the level of the interest rate indeterminate. “There is no theory of money in The General Theory. Or rather there is no coherent theory of money.”[3] It is this lacuna that Marglin sets out to fill.
To do this, he follows Tobin by interpreting the concept of liquidity preference as behavior toward risk, which implies that the demand for interest-bearing liquid assets is fundamentally a matter of portfolio choice. Concretely, he posits a utility function U(E(A),M), with E(A) representing the expected value of total wealth a year from now, and Mrepresenting holdings of the interest-bearing liquid asset. In effect, he puts money in the utility function, and then imposes restrictions on the indifference map as a way of capturing “liquidity preference.”[4] In his model, wealthholders maximize utility subject to a budget constraint, given the price of bonds and the discount rate on bills. In portfolio equilibrium, demand for bills (“money”) must equal supply of bills, and the price that equilibrates is the price of bonds.
The problem, Marglin notes, is that this theory only determines the price of bonds for any given discount rate on bills, and says nothing about the discount rate itself. Once we translate Keynes into math, it turns out that it is a theory only of the spread between the long-term bond rate of interest and the fixed zero rate of interest on cash. (In this regard, Marglin explicitly endorses the prescient criticism first offered by Dennis Robertson, who viewed the monetary theory of the General Theory as inferior to what Keynes had earlier offered in his Treatise on Money (1930).) As a consequence, once we introduce interest-bearing money as a third financial asset, the theory tells us about the term premium, the difference between the long-term rate and the short-term rate, but says nothing at all about the overall level of rates.
Since the whole idea behind the theory of liquidity preference was to propose an alternative to the classical theory of the interest rate, this is a big problem. Marglin’s suggested fix is to treat the short-term rate of interest as a policy variable fixed by the central bank, hence analogous to the exogenously fixed zero rate of interest on cash. The result is certainly a coherent theory, on its own terms, and also one that is arguably minimally distant from what Keynes himself wrote in the General Theory. It is, as Marglin says, “one way” to close the model,[5] but is it the best way? I shall have more to say on that point shortly.
With this apparatus in hand, Marglin can, first, rationalize a downward-sloping demand for money curve, and also a “liquidity trap” floor on bond yields when bill rates hit the zero lower bound. (That’s the relevant “building block” for Keynes’ model.) Mathematically speaking, to get those results it turns out that we need to assume both short-term risk aversion and a specification of “normal reversion” for the bill rate, the latter meaning that wealthholders view deviations from the “normal” bill rate as temporary, reverting to normal at a specified rate of change. That’s the job of the Mathematical Appendix.[6] Technical stuff, but the important thing is to appreciate that the game is about rationalizing the relevant building block.
Second, Marglin can take this model to the actual data on short rates and long rates. That’s the job of the Empirical Appendix.[7] As everyone knows, the standard “expectations hypothesis” of the term structure reliably fails to fit the data. (Marglin himself treats John Campbell, his Harvard colleague, as the authority on that matter.) Marglin does not go so far as to claim that his empirical exercise is a test of his alternative theory The point seems merely to show that there is a specification of the model that is not rejected by the data, and that therefore his recasting of the liquidity preference theory needs to be taken seriously a possible alternative. “The data, I argue, actually confirm Keynes’s view of the sources of liquidity preference and in this respect validate the theory.”[8]
So far, we have been talking about the demand for money as a store of value, what Marglin (and Keynes) calls M2, not the demand for money as a medium of exchange, what Marglin (and Keynes) calls M1. Chapter 13 then brings transactions demand into the story with the idea that whereas M2 is demand for Treasury bills (and other risk-free short-term paper), M1 is demand for bank deposits, along with the further idea that whereas the supply of Treasury bills is given, the supply of bank deposits is not. Responding to demand, banks can simply expand their balance sheets on both sides, and also contract them, creating and destroying transactions deposits as needed. The result is that transactions deposits play no role in determining macroeconomic equilibrium; rather it is macroeconomic equilibrium that determines the demand for transactions deposits, with supply adjusting elastically. (On this point, Marglin cites Marc Lavoie and J. Laurence Laughlin as fellow travelers.[9]) The important point is that Marglin’s central bank theory of interest survives extension of the model to fractional reserve banks that offer transactions accounts. “Indeed, if central banks did not exist, we would have to invent them just for the sake of economic theory.”[10]
What to make of all this? Me, I have since 1989 been following Hicks Market Theory of Money (1989) on these matters, where, like Robertson, he urges building on Keynes’ Treatise (1930) rather than his General Theory (1936). (That is to say not the Hicks of “Mr Keynes and the Classics” but rather the Hicks of “A Suggestion for Simplifying the Theory of Money,” see Mehrling 2019.[11]) Hicks urges us to remember that Keynes was writing in an unusual period, as Marglin recognizes,[12] a time when short-term interest rates were basically zero. For Hicks, that is the reason Keynes limited himself to a two-asset world:
“During the years 1932-8 (just when Keynes was writing his General Theory and defending it against its first critics) the market rate of discount on bills, in London, was hardly more than one-half per cent. So bills were standing at a discount which was practically negligible; to treat them as being money, as Keynes implicitly did, was very natural.”[13]
Natural for 1932-8, perhaps, but no longer. Like Keynes, Hicks thinks we need a monetary theory of the rate of interest. And like Marglin, Hicks is after a theory of the short-term rate, not the long-term rate of the General Theory. But Hicks’ alternative is different, a market theory of interest—not a central bank theory of interest.
To my way of thinking, the most important contribution of Marglin’s book is its emphasis on the dynamic “price mechanism,” which concerns how prices emerge from actual trading in disequilibrium. “Once dynamics become the focus of the model, it becomes essential to investigate the institutional basis of dynamic systems instead of requiring merely that the dynamics exhibit a surface plausibility.”[14] Here and there in the book, we get hints of a market-making operation in the background, with dealers building up and drawing down inventories to absorb temporary imbalances between supply and demand, and shifting prices at the same time. For example, in a primer chapter on standard dynamic market adjustment stories: “Inventories and order books play a central role in the story: as order books shorten and inventories pile up, Marshallian producers respond by cutting prices, while, conversely, increasing backlogs and inventory depletion lead them to increase prices.”[15]
The problem, to my mind, is that this kind of analytical frame is used only for commodity prices and macroeconomic aggregates, and is nowhere to be found in the discussion of asset prices. In money and financial markets, the relevant “institutional basis” is arguably the dealer function, which is exactly what Hicks puts at the center of his Market Theory of Money. Just so, Hicks suggests thinking of the short-term rate of interest as a kind of exchange rate between a mercantile sector which uses bills of exchange as means of payment and the outside sector which uses cash.[16] Such a theory, it seems to me, would be completely in the spirit of Marglin’s core emphasis on dynamics, and so might be considered a friendly amendment to his larger project, a different and perhaps better way to close the model.
But is my amendment really so friendly? In Marglin’s reading, the fact that The General Theory has no theory of the rate of interest amounts to a deep critique of capitalism. If interest rates are indeterminate, that means that output and employment are also indeterminate, and Marglin draws the conclusion that capitalism requires the guiding hand of the central bank if it is to have a determinate outcome. From this point of view, Hicks’ market theory of interest might be interpreted as an argument in favor of capitalism?
I would urge not. A market theory of money is very much an attack on the classical citadel, and it is also very much compatible with the conclusion that capitalism requires the guiding hand of the central bank. Indeed, I read Hicks as very much following the line of British central banking theory that goes back at least to Bagehot, who famously urged that “Money will not manage itself, and Lombard Street has a great deal of money to manage.” The problem is not indeterminacy but instability, indeed “the inherent instability of credit” as Hawtrey famously put it in his essay on “The Art of Central Banking.”[17] It is because of instability, not because of economic theory, that we have had to invent central banking!
Finally, for today’s problems, it is important to remember that Keynes was writing in an unusual period in another respect as well. Because world markets had broken down, it was natural for Keynes to treat the case of a closed economy operating on its own. Just so, his building blocks were consumption and investment, not exports or imports, much less capital flows either short term or long term, and the exchange rate enters nowhere. Fine for 1932-8 perhaps, but note well that these are the same building blocks that Marglin uses for his “Twenty-First-Century General Theory,” even though today the only closed economy is the world as a whole, and arbitrage links interest rates and asset prices across borders.
Can a central bank theory of interest survive in such a financially globalized world? Raising Keynes does not ask the question, but all of its modern-day readers will. Indeed, the Bagehot problem of managing money is today a problem not for any individual central bank on its own but rather for all of them collectively and cooperatively, which is to say that it is a political problem as much as an economic one. Were Keynes alive today, I dare say he would be putting the problem of managing global money at the center of his attention. That’s the Keynes we need for the 21st century.
Forty years ago, when I was Marglin’s student, he was just finishing his great book Growth, Distribution, and Prices, which I found to be an inspiration even though there was no money in it anywhere. Today, he has written another inspiring book, though what he says about money I find not completely satisfactory. Indeed, so far as I can see, the citadel I have been attacking all these years, with the help of Hicks, is the citadel he is now defending, in the name of Keynes. With all due respect, I beg to suggest that no such defense is necessary. The model closure that Hicks proposes is, so far as I can see, quite consistent with Keynes’ project and, even more, quite consistent also with Marglin’s own core project of rooting a modern Keynes in the dynamics of the price mechanism.
[1] For an introduction to this approach, meant for a general audience, see “Financialization and its Discontents.” For serious students, the place to start is the MOOC on Coursera, filmed in Fall 2012, and then, depending on interest, the papers and videos collected on my website. BACK TO POST
[2] Stephen A. Marglin, Raising Keynes: A Twenty-First Century General Theory (Cambridge: Harvard University Press, 2020), 376. BACK TO POST
[3] Ibid, 522. BACK TO POST
[4] Ibid, 433. BACK TO POST
[5] Ibid, 784. BACK TO POST
[6] Ibid, 432-56. BACK TO POST
[7] Ibid, 457-96. BACK TO POST
[8] Ibid, 11. BACK TO POST
[9] Ibid, 507, 519. BACK TO POST
[10] Ibid, 514. BACK TO POST
[11] Perry Mehrling, “The monetary education of John Hicks,” in Money, Finance and Crises in Economic History, eds. Annalisa Rosselli, Nerio Naldi, & Eleonora Sanfilippo (New York: Routledge, 2019). BACK TO POST
[12] Marglin, Raising Keynes, ch. 12, fn 6. BACK TO POST
[13] John Hicks, A Market Theory of Money (Oxford: Claredon Press, 1989), 63. BACK TO POST
[14] Marglin, Raising Keynes, 214. BACK TO POST
[15] Ibid, 204. BACK TO POST
[16] Hicks, A Market Theory of Money, 51. BACK TO POST
[17] Ralph Hawtrey, The Art of Central Banking (Longson: Longmans, Green Co., 1932), 166. BACK TO POST