Book review: Marglin, Stephen A. (2021): Raising Keynes: A Twenty-First Century General Theory, Cambridge, MA, USA (896 pages, Harvard University Press, hardcover, ISBN 978-0-674-97102-8)
Marc Lavoie Professor Emeritus, University of Ottawa, Canada, University of Sorbonne Paris Nord, France and FMM Fellow

Search for other papers by Marc Lavoie in
Current site
Google Scholar
Open access

Published Open Access on a CC-BY license. This license does not apply to any third party materials included in the article for which permission should be sought from the original rights holder before reuse.

This is a special book. It has nearly 900 pages, and it is printed on high-quality paper, with coloured figures and tables, at a reasonable price if one compares it to other hardcover publications. There are 60 pages of endnotes, 24 pages of references, and thankfully most of the algebra is found in its appendices. It deals with several issues, as one would guess in such a large book: the consumption and investment functions, liquidity preference, various meanings of money, fiscal policy and functional finance, long-run growth models, actual financial and real data regarding the Great Depression and the Great Recession, and the empirics of wage and price inflation.

Marglin's main goal, however, which occupies most of the first half of the book, is to convince readers that Keynes was right to argue that there could be an unemployment equilibrium in an economy with perfect competition, flexible prices, flexible wages and possibly flexible interest rates. The main intent is to refute the orthodox belief of a self-regulating economy, despite excluding imperfections, frictions or rigidities. As the title of the epilogue says, the purpose of the book, as in Keynes's General Theory, is to attack mainstream economics in its citadel, by maintaining these assumptions of neoclassical economics, including the assumption of diminishing returns. Ultimately, Marglin wishes to show that Keynes was right, even though he did not express his views in a clear manner, and that aggregate demand plays an active role in determining output in the short run, as New Keynesians or Marxists would concede, but also in the long run, as post-Keynesians claim: ‘Aggregate demand matters in the long run as well as in the short run’ (Marglin 2021: 771). There is some similarity to the objective pursued by Marglin and that of Duncan Foley (2014), who also argued that wage rigidity was not at issue when trying to explain unemployment equilibria, relying instead on informational problems classified as externalities, thus seemingly embracing David Colander's (1996) post-Walrasian economics based on the lack of coordination. But this is not the route taken by Marglin.

It is well known that Keynes's involuntary unemployment was questioned by Keynes's own effect, whereby a fall in the money wage or an increase in the money supply could generate a fall in the interest rate and hence an increase in investment, output and employment. Denying a return to full employment after a negative shock, Keynes's critics said, thus required rigid wages, a liquidity trap or an inelastic investment function. Involuntary unemployment was also questioned through the real-wealth effect or the real-balance effect, which would raise consumption and hence output and employment as prices would fall. Once again, denial would require rigid prices. Marglin concedes that critics of unemployment equilibria, basing themselves on the Keynes effect or the real-wealth effect, are correct when making use of comparative statics, that is, the comparison of two economies, one with high wages and prices and the other with low wages and prices, or what Joan Robinson (1980) called logical time. Marglin contends, however, that when using Robinson's historical time, which he calls real time, based on dynamic models with disequilibria, which some refer to as sequential analysis, where one observes the consequences of falling wages and prices, then it can be shown that Keynes is right, or at least that Keynes is relying on an adjustment mechanism that occurs during the traverse to the new equilibrium. In historical time, indeed, the new equilibrium depends on the path taken during the transition.

Marglin presents progressively more complex macroeconomic models. He starts out with Pigou's theory of employment, which can also be found as the view of contemporary textbooks regarding the determination of output and employment. In the output and real-price plane, the real price being the ratio of the price level to the wage rate (P/W), the supply curve of goods is given by the rising marginal cost curve; in that plane, there is a downward-sloping curve, which corresponds to the upward-sloping labour supply curve that could be found relative to the real-wage rate (W/P). Aggregate demand plays absolutely no role; all is determined by the supply conditions of firms and the disutility of labour. There cannot be unemployment, unless the real wage is rigid.

Marglin then introduces what he calls Keynes's first pass model, where both the interest rate and the nominal wage rate are given. In that model, the aggregate demand schedule is vertical (with output on the x axis and P/W on the y axis), as it depends on investment, which itself depends on the interest rate, the marginal efficiency of capital schedule and the propensity to consume. The intersection of this vertical aggregate demand curve with the rising marginal-cost curve generates the equilibrium real-wage rate and an equilibrium that need not coincide with the third curve, given by the disutility of labour. It will thus imply unemployment (or conceivably, more than full employment). For full employment to be achieved, investment has to increase, thanks to a lower interest rate, shifting the vertical aggregate demand curve towards the right, thus generating a lower real wage and a higher P/W ratio, so as to induce firms to produce more. This first pass model is reminiscent of Paul Davidson's (1998) and Pasinetti's (1974: ch. 2) representations of the principle of effective demand, which both have a one-way decomposable system, with Davidson's arguing that the level of activity determines the real wage and not the other way around.

In what Marglin calls Keynes's second pass model, the interest rate becomes endogenous because of an exogenous money supply. The one-way system is lost, because the change in the price level has an impact on the demand for money for transactions, hence an impact on the interest rate, hence on investment, hence on aggregate demand, which is now downward-sloping (instead of vertical). Hence economic activity is determined by both aggregate demand and the marginal cost curve of firms. Marglin is back to a version of the IS–LM model, and he shows that within a comparative static analysis, the analyses of Modigliani and of Patinkin are vindicated, that is, a sufficiently low combination of wages, prices and interest rates will be compatible with full employment.

Marglin then considers dynamics in real time, starting with the first pass model (at a fixed interest rate). He considers two possible adjustment mechanisms, with both real prices and quantities getting modified, as Bhaduri (2008) had done within Kaleckian growth models. With the Marshallian mechanism, based on flexprice dynamics (a difference between expenditure and income, or equivalently between desired investment and desired saving, triggers a change in prices), the equilibrium always remains on the profit-maximizing marginal cost curve (and in general off the aggregate demand curve); with the Walrasian mechanism, based on fixprice dynamics (a difference between expenditure and income triggers a change in quantities), the economy always comes back to the vertical aggregate demand curve (and in general off its marginal cost curve), at either higher or lower real wages. Marglin then provides his own definition of a dynamic equilibrium: it is a situation where wages and prices are falling at the same rate, so that real wages, and the P/W ratio, become constant. In his concluding chapter, Marglin (2021: 782) justifies it in the following way:

We can take account of all three schedules if we are willing to drop the identification of equilibrium with the equality of supply and demand. Instead, a stationary real price is achieved by nominal prices and wages falling (or rising) at the same percentage rate. One way this can happen is that employers reduce prices in order to promote sales, while workers reduce wages in order to increase employment. If the percentage rates of price and wage reduction are the same, the real price will not change. Both workers and capitalists are frustrated: they are trying to change real wages and real prices, but their actions are offset by the actions on the other side. Everybody may be frustrated, but the economy is in equilibrium.

Except in the case of an ultra-rapid wage adjustment, there is an unemployment equilibrium. But what kind of equilibrium does he offer in the case of the Marshallian mechanism based on the so-called flexprice dynamics? With a negative shock on aggregate demand, prices and wages may end up falling at the same speed, but production keeps exceeding sales, and inventories just keep accumulating. In the limiting case of a very fast price adjustment, production gets reduced to the level of sales, and inventories stop growing, but they are now much higher than what was presumably their initial target level. Marglin has nothing to say about the inconsistency of this peculiar equilibrium concept and its impact on inventories, presumably because he deals with short-term models, but only a few sentences, buried in an appendix, appear later when dealing with the long term (Marglin 2021: 673–674). There are authors out there – Wynne Godley, Duménil and Lévy, Peter Flaschel and his numerous acolytes – who have dealt in a consistent way with inventories.
Marglin next considers the second pass model, with endogenous interest rates and a fixed money supply. He grants, on the basis of the assumptions deployed so far in the argument, along the lines of Modigliani, that in the flexprice case, the economy always returns to full employment. However, in the fixprice case, the Keynes effect may not work out as expected: if wages fall quickly and if the labour supply elasticity is low, then the model explodes in an ever-widening orbit. Furthermore, as we shall see below, Marglin (2021: 783) points out that the Keynes effect does not work in real time as it is supposed to work in a comparative-statics framework because

a fall in prices leads to a reduction in transactions demand, but this does not lead to greater amounts of money pouring into wealth portfolios. ‘Money’ turns out not to be homogeneous, and reductions in the demand for money to transact business are reflected in reductions in bank lending, not in an influx of money into wealth portfolios.

As a consequence, the fall in prices has no bearing on interest rates.

As to the wealth effect, Marglin finds that it will hold, thanks to the real balances held for speculative purposes (but not for transaction purposes, as these will endogenously fall), as long as households with businesses don't react too much to the rise in the real value of their debt. The possibility of bankruptcy, due to the fall in business income relative to interest payments on the value of debt (which remains constant in nominal terms), with its likely negative impact on economic activity, is mentioned but not modelled. Thus, on this, Marglin has little to say beyond Kalecki (1944). One would need a stock–flow consistent model à la Godley.

In later chapters, these models are extended to a situation where saving depends on the Cambridge equation, thus creating a U-shaped aggregate demand curve in the output and P/W plane, intersected at its minimum point by the marginal cost curve. Marglin could have reminded readers that Bertram Schefold (1983), among others, had already noted this feature in the employment and W/P space. Marglin adds a distinction between two sorts of investment: capital-widening investment, which expands capacity and is output-enhancing, and which is negatively related to the level of the real wage, as it was in the famous Bhaduri/Marglin (1990) paper; and capital-deepening investment, which substitutes capital for labour and which is positively related to the real wage, because the higher the real wage, the more profitable capital substitution will be. Marglin (2021: 352) notes that capital widening is likely to be important in prosperous times, while capital deepening will be very much dominant in a recession, thus implying that wage-led demand is likely to arise in a stagnant economy as higher real wages will stimulate both consumption and investment.

These models, including the complications just mentioned, are then also extended to long-run models of growth and distribution, within the same (P/W) framework, also found in Marglin (2017), but they are difficult to follow. Indeed, I recall Marglin looking at his own curves with some puzzlement when presenting these long-run models at a plenary session of the FMM conference in Berlin a few years ago. The fault can be attributed to the adoption of decreasing returns within these models, which create non-linearities. One may wonder whether this is a judicious choice when there is so much evidence that we live in a world of constant marginal costs or constant direct unit costs, as emphasized by post-Keynesian authors. In the long run, the disutility of labour is replaced by the growth rate of the labour force. Marglin assumes that this is an endogenous variable, which responds to employment pressures, as most post-Keynesians would argue, giving however the impression that only himself, Joan Robinson, John Cornwall and Donald Harris have ever thought about it.

The other major theme of Marglin's book is that Keynes's analysis of liquidity preference and money in The General Theory was a mess, on two grounds.1 Nearly 150 pages are devoted to this theme. First, as noted above, Keynes assumes, at least most of the time, that the supply of money is exogenous, as if he were living in a world of commodity money. Marglin insists that this is not the case, since modern economies are based on fractional-reserve banking, an expression that appears no fewer than 56 times in the book. For Marglin, such a system, with given reserves and banks being fully loaned-up, is little different from a world of commodity money, where money is scarce. He seems to adopt, instead, a world where money is (partially) endogenous, either because banks are not always fully loaned-up or thanks to Minsky-like innovations of the financial system, where banks can get around reserve requirements by getting loans off their books by creating collateralized debt obligations (CDOs). Marglin argues that US banks, when Keynes wrote The General Theory in the 1930s, had large excess reserves and hence were not fully loaned up; but such a situation, instead of being attributed to the behaviour of banks, could have resulted at the time from the same kind of quantitative operations that the Federal Reserve has pursued since 2008, meaning that banks could not avoid being stuck with excess reserves. While Marglin very clearly rejects the first mainstream view of banking, the one based on loanable funds and banks as financial intermediaries, he keeps making references to the second mainstream view – fractional-reserve banking, usually associated with the money-multiplier theory of banking – thus neglecting the credit-creation view of money and the financing-through-money-creation view of banking of post-Keynesian economics, which have been clearly endorsed by central bankers over the past ten years, and which ultimately he seems, reluctantly, to endorse himself.2

Marglin also believes that Keynes's liquidity preference theory was rather unsuccessful and that Keynes failed to deliver a theory that would be clear enough to replace the mainstream interest-rate theory based on productivity and thrift. The reason for this is that ‘liquidity preference could never be a theory of interest … liquidity preference can never be more than a theory of interest-rate differentials’ (Marglin 2021: 514, emphasis in original). Marglin gives the impression that this is a revelation, a new and important discovery, and that at long last somebody has cleared up the concept of liquidity preference by adding short-term assets that carry interest to the main two assets considered by Keynes in his General Theory: long-term assets and cash. But we have known this for a long time, as I argued 30 years ago (Lavoie 1992: 195–196), and as did others before me.

The good news, however, is that, thanks to this view of liquidity preference, we are back to what Marglin called the first pass at Keynes's model, where the long-term rate of interest is fixed independently of economic activity or the size of monetary transactions, depending only on the target short-term rate of the central bank and on the spreads arising from liquidity preference between interest-carrying short-term and long-term assets. This representation is even better than Pasinetti's (1974), because it excludes the impact of the supply of money, since the size of the flow of transactions (nominal GDP) determines the transaction demand for money, which itself causally generates the needed supply of money. Furthermore, with this understanding of liquidity preference and with the recognition of the role played by the target rate of interest set by the central bank, Marglin (2021: 520) can allege, in agreement with many post-Keynesians, that ‘the rate of interest turns out to be a conventional phenomenon rather than a reflection of market equilibrium …. There is no “natural rate of interest” that emerges from the free play of market forces and regulates the accumulation of capital’. As John Smithin (1996: 47) once put it, ‘the real economy must adjust to the policy-determined interest rate, rather than vice-versa. This is therefore the precise opposite to the natural rate doctrine’.

Readers should be aware by now that Marglin's book is quite rich and full of insights, especially for those who wish to make sense of Keynes's General Theory. It also contains interesting statistical and empirical material, dealing both with the Great Depression and contemporary periods. However, it generates some frustrations when Marglin keeps referencing a fractional-reserve banking system, normally associated with the standard money-multiplier story, whose existence is denied by post-Keynesian economists, who advocate instead a credit-divisor story, based on the notion of reversed causation, with central banks supplying reserves defensively, so as to achieve their target interest rate. Furthermore, Marglin sticks to rising marginal costs and profit-maximization, also making use of utility functions in other parts of his book that have not been discussed in this review. He reasons, as noted at the beginning of this review, that ‘like Keynes, I am committed to engaging mainstream economics on its own turf, of which the aggregate production function is an important constituent’ (Marglin 2021: 661). We will have to wait and see to discover whether the mainstream is willing to engage with him on this account.

  • 1

    There is a longer discussion on this theme in Lavoie (2022) and in other contributions to the roundtable on Marglin's book on the blog.

  • 2

    Indeed, elsewhere Marglin (2020) clearly endorses the credit-creating view of banking, mainly on account of the fact that banks in need of reserves can always borrow them on the federal funds market.


  • Bhaduri A. , 'On the dynamics of wage-led and profit-led growth ' (2008 ) 32 (1 ) Cambridge Journal of Economics : 147 -160.

  • Bhaduri A. & Marglin S.A. , 'Unemployment and the real wage: the economic basis for contesting political ideologies ' (1990 ) 14 (4 ) Cambridge Journal of Economics : 375 -393.

    • Search Google Scholar
    • Export Citation
  • Colander D. , 'Overview ', in D. Colander (ed), Beyond Microfoundations: The Post Walrasian Macroeconomics , (Cambridge University Press , Cambridge, UK 1996 ) 1 -17.

  • Davidson P. , 'Post Keynesian employment analysis and the macroeconomics of OECD employment ' (1998 ) 108 (448 ) Economic Journal : 817 -831.

  • Foley D.K. , 'Varieties of Keynesianism ' (2014 ) 43 (1 ) International Journal of Political Economy : 4 -19.

  • Kalecki M. , 'Professor Pigou on the classical stationary state: a comment ' (1944 ) 54 (213 ) Economic Journal : 131 -132.

  • Lavoie M. , Foundations of Post-Keynesian Economic Analysis , (Edward Elgar Publishing, Aldershot, UK and Brookfield, VT 1992 ).

  • Lavoie, M. (2022): Raising Marglin, URL:

    • Search Google Scholar
    • Export Citation
  • Marglin S.A. , 'Wages, prices, and employment in a Keynesian long run ' (2017 ) 5 (3 ) Review of Keynesian Economics : 360 -425.

  • Marglin, S.A. (2020): What do banks do?, URL:

  • Marglin S.A. , Raising Keynes: A Twenty-First Century General Theory , (Harvard University Press, Cambridge, MA 2021 ).

  • Pasinetti L.L. , 'The economics of effective demand ', in L.L. Pasinetti (ed), Growth and Income Distribution: Essays in Economic Theory , (Cambridge University Press , Cambridge, UK 1974 ) 29 -53.

    • Search Google Scholar
    • Export Citation
  • Robinson J. , 'Time in economics ' (1980 ) 33 (2 ) Kyklos : 219 -229.

  • Schefold B. , 'Kahn on Malinvaud ', in J. Eatwell & M. Milgate (eds), Keynes's Economics and the Theory of Value , (Oxford University Press , Oxford 1983 ) 229 -246.

  • Smithin J. , 'Real interest rates, inflation, and unemployment ', in B.K. MacLean & L. Osberg (eds), The Unemployment Crisis: All for Nought? , (McGill-Queen's University Press , Montreal and Kingston, ON 1996 ) 39 -55.

    • Search Google Scholar
    • Export Citation
Since 2022 Since May 2022 Past 30 Days
Abstract Views 438 0 0
Full Text Views 433 291 40
PDF Downloads 284 212 22