January 3, 2022
William H. Janeway, University of Cambridge
In chapter 13 of Raising Keynes, Stephen Marglin addresses the theory of money that is haphazardly and, to a substantial extent, implicitly presented in Keynes General Theory (“GT”). Liquidity preference is the central innovation in Keynes’ sketch of an alternative to the established quantity theory. Keynes proposes that demand for money balances, effectively identified with cash, defines two separate but substitutable quantities, transactional (M₁) and speculative (M₂). The supply of money is taken as fixed, either because money itself is a commodity whose supply is exogenously given or because it is set by the central bank.
Demand for M₁ is a function of the flow of nominal output in the economy. Demand for M₂ reflects the calculations of investors balancing the return available from owning bonds with their own uncertainty about the future, both the potential for future adverse contingences that motivate self-insurance today and the future movement of the prices of bonds that carry the risk of capital loss. From those calculations the interest rate emerges at the level needed to compensate investors for those risks.
In turn, the level of interest rates that are generated from the financial markets informs the rate of investment in the construction of real assets and, thus, aggregate demand. Reflecting his deep engagement with the capital markets, Keynes asserts that it is transactions in the stock of existing assets that set interest rates, not the flow of savings and investment. As anyone who has participated in an initial public offering of equity or the issuance of a bond will testify, the secondary market in existing assets prices the primary market where new capital is raised. That is, the valuations of existing, traded assets set the valuation at which new assets are introduced to the market.
Marglin identifies two lacunae in the GT’s theory of money. These are its failure to take account of two observable characteristics of the financial system: interest-bearing “safe” short-term assets and fractional reserve banking, although it is evident that Keynes was thoroughly aware of both of them. It is a curious fact that explicit note is taken of each in his prior work, The Treatise on Money, Volume II, The Applied Theory of Money.
Marglin fills these two holes in the GT. The consequences complicate the analysis of money yet change the fundamental message of the GT only slightly. Marglin shows that Keynes’ division of the demand for and supply of money into M1 and M2 cannot hold, given that the liabilities generated by the working capital needed for business transactions can be held as assets by portfolio owner-managers. Thus, the same assets may be included in both categories.
Moreover, money as an asset competes with interest-bearing assets — typically known as “bills” — that are sufficiently short-term to minimize price risk and sufficiently diversified to minimize credit risk. Thus, liquidity preference becomes a theory of spreads, rather than a theory of the entire structure of interest rates when there is no universally accepted asset with an interest rate of zero. And so, to anchor the term and credit spreads that financial markets generate, an active central bank is required that can issue just such an asset. Without the central bank, the interest rate on bills is indeterminate. In turn, so is the hurdle rate that determines the volume of investment by business and, therefore, aggregate demand and the level of employment, and the nominal income which determines the volume of saving. “Capitalism left to itself,” to invoke Marglin’s terms, dissolves in the absence of a central bank.
In filling in these holes, Marglin liberates Keynes from the “Keynes Effect,” an equilibrating force that the GT sketches as a path by which a monetary production economy might return to full employment. If wages are reduced and prices follow, then nominal output and, therefore, the transaction demand for money will also decline. This will free up M₁ funds to be available to satisfy the speculative demand for money which will, in turn, induce a decline in interest rates and stimulate investment. But if liquidity preference acts to satisfy speculative demand with near-money assets — and so the structure of rates, including the hurdle rate, is already established — the reduction in transactions demand is irrelevant.
In fact, in chapter 7 of Raising Keynes, Marglin has already done the heavy lifting. By first reconstructing Keynes’ theory of aggregate demand in a dynamic setting and then marrying it with Irving Fisher’s debt-deflation argument, Marglin disposes of the variety of “real balance” arguments for attacking Keynes’ proposal of a macro-economic equilibrium with unemployed resources. The Keynes Effect potentially offers an alternative path to full employment, but any path that depends upon a fall in the current value of output and income while the value of nominal debt — a legal institution in most cases — remains fixed is similarly and summarily rendered null. For, unless the volume of debt is indexed to the price level, the negative consequences of default and bankruptcy on aggregate demand will overwhelm any possible positive effects from an increase in the real value of money holdings.
Marglin’s corrections to the GT’s theory of money deliver an additional benefit. I myself have long been allergic to the application of the word “natural” to any parameter of the economic and/or financial system. The economy and the financial system are both social constructs and have evolved through human agency, both intended and unintended, and operate in an institutional context subject to political, cultural and social forces. That any parameter of either should be endowed explicitly with an epithet purporting to establish its truth-value as akin to the law of gravity or the speed of light I find intellectually offensive. This very much applies to Wicksell’s “natural rate of interest.”
Marglin notes the persistence of the construct long after the notion had expired that economic forces would necessarily drive alignment of the central bank’s policy rate with the so-called natural rate. The modern-day version, however, still asserts the existence of a natural rate determined in the real economy by the forces of thrift and productivity. A central bank that deliberately sets interest rates out of line with the natural rate is, in this view, defining a rate of inflation or deflation to bring real rates into line with that natural rate. But all this derives from a theoretical construct where money is merely a “veil” and economic outcomes are exclusively a function of “real” variables. It was precisely Keynes’ innovative recognition of money as an alternative asset, available for market participants to hold, that can short-circuit the flow of real savings into real investment. Keynes’ liquidity preference explicitly recognizes that money plays an active role in the monetary production economy of the world in which we actually live.
When the natural rate is read as having fallen to negative levels, the central bank is indeed challenged to bring real rates into alignment by reducing nominal rates: hence the proliferation of negative policy rates and quantitative easing since the Global Financial Crisis, as well as the expressed concern for “secular stagnation.” Marglin recasts the analysis in line with Keynes. The rate of change in prices is a function of the state of aggregate demand and supply. The nominal level of interest rates then sets the real rate, about which there is nothing “natural.”
The macroeconomic policy problem, in this case, is not the inability to reduce nominal rates as far as needed below the “zero lower bound” to match the supposedly negative natural rate. The underlying problem was defined in the heart of the Great Depression by the great central banker Marriner Eccles. In congressional testimony in 1935, the then Governor of the Federal Reserve engaged in an exchange with Congressman T. Alan Goldsborough:
Congressman T. Alan Goldsborough: You mean you cannot push a string.
Governor Eccles: That is a good way to put it, one cannot push a string. We are in the depths of a depression and…beyond creating an easy money situation through reduction of discount rates and through the creation of excess reserves, there is very little, if anything that the reserve organization can do toward bringing about recovery.
A central bank that can only push on that string in response to excess real resources, whether the endogenous result of persistently inadequate aggregate demand or in response to an exogenous plague-like shock, may be expected to generate increasing stocks of excess reserves, as Jamie Galbraith points out in a companion paper. Moreover, when negative real rates of interest result, investors may be expected to search for real returns through an increasingly aggressive acceptance of risk. This, indeed, is the lesson of the past dozen years, the age of Quantitative Easing on both sides of the Atlantic, accompanied by bubbly excess in financial markets and stubbornly low rates of real investments. Only now have relevant degrees of fiscal stimulus been applied, finally motivated by the exogenous shock of Covid-19, creating conditions for increased economic growth and enough inflation to legitimize the normalization of monetary policy.
Eccles himself was an active proponent of stimulative fiscal policy in Depression conditions. He played a substantial role in the fiscal policy initiative of 1938 that belatedly reversed the “Roosevelt Recession,” induced by an untimely shift to budgetary austerity and tightening of monetary policy. Keynes, of course, was long identified with debt-financed public expenditure as a corrective to unemployment, although as Marglin points out in chapters 1 and 14, he barely touches on the subject in the GT, apart from his provocative invocation of the practical utility of filling old bottles with banknotes and burying them in garbage dumps.
The central message of the General Theory is that, left to itself, a capitalist economy cannot be expected to find and hold a general equilibrium with all resources fully employed. To the extent that this message is compromised by an inadequate theory of money, Marglin corrects the missing element. Yet, as noted and as exhaustively discussed by Marglin, the theoretical onslaught on Keynes and the General Theory was largely expressed through the “real balance” effect that Marglin effectively counters by marrying Keynes with Fisher. In practical terms, then, the corrective work of chapter 13 of Raising Keynes is of relatively limited significance. Marglin summarizes that significance in characteristically clear terms:
Once we drop the assumption of commodity money and replace it with the more realistic assumption of fractional-reserve banking and related forms of credit, we have to contend with both the endogeneity of money and the complementarity of speculative and transactional demands. What then is left of liquidity preference? Surprisingly, everything that was there before. 
In turn, the case for an active central bank remains untouched. And what also remains is that, whenever the central bank encounters the challenge of pushing an economy back towards full employment through the “string” of monetary ease, then the case for active fiscal stimulus also stands.
 Stephen A. Marglin, Raising Keynes: A Twenty-First Century General Theory (Cambridge: Harvard University Press, 2020), 514. BACK TO POST