2.1 Godley

It is rather difficult today to wonder whether Godley was a Keynesian economist, as he is so closely associated with Kaldorian economics and Keynesian policies, and even considered as a key inspiration for MMT. Shipman (2019, pp. 140ff) recalls however that in 1974 there was a serious clash between the representative Keynesians Richard Kahn and Michael Posner on the one hand, and on the other hand the research team led by Godley – the Cambridge Economic Policy Group (CEPG) – which developed into what became known as the New Cambridge School. As Shipman (2019, p. 145) says, ‘some Cambridge Keynesians went further in their criticism, viewing the CEPG's approach as tending towards monetarism’. Advocates of the New Cambridge view were thus accused of being ‘closet monetarists’. Robert Dixon (1982/1983, p. 291) was particularly harsh. When describing the New Cambridge view of the relationship between the budget deficit and the trade deficit, he wrote that ‘JPKE readers will immediately perceive that this (virtual monetarist) vision of income distribution and the causes of income instability constitutes a total rejection of Keynes's views on economic policy’, concluding that ‘there can be no doubt that both in its analytical core and in its policy assignment, doctrines associated with the New Cambridge School represent a dramatic break with the ideas of Keynes. New Cambridge theory seems more pertinent to long-run equilibrium than the world in which we have our being’ (Dixon 1982/1983, p. 294).

Further confusion arose when readers of the book by Godley and Cripps (1983, p. 17) on Macroeconomics found its introduction arguing ‘our present synthesis may be broadly characterized by saying that we make a “monetarist” financial system (based on the behaviour of stocks of money, financial assets and debts) drive a “Keynesian” flow system based on the response of expenditure to income’. Retrospectively, it seems that it would have been better to avoid the word ‘monetarist’, as the use of the word and that of long-run or steady-state positions, plus the critiques addressed against the Keynesian orthodoxy of the time, which was said to be ‘incomplete and inadequate’ because ‘it did not properly incorporate money and other financial variables’, led several contemporary Keynesians to wonder whether Godley and his associates were not some kind of Cambridge Monetarists.

This was most obvious at the Keynes Centenary conference, held in Cambridge in 1983, where Godley (1983) presented an essentialist version of the Macroeconomics book written with Cripps. During the general discussion, someone ‘pointed out that Professor Godley's model had a real whiff of monetarism about it’ (in Worswick and Trevithick 1983, p. 174), because stock–flow relationships figured prominently and because he was assuming that the economy operated near the steady state. The confusion was so great that Francis Cripps felt obliged to state in a response that ‘what they were doing was Keynesian monetary economics; it was not neoclassical let alone general equilibrium monetary economics’ (ibid., p. 176).

Luckily, not all outsiders were bedazzled by the New Cambridge use of the fundamental macroeconomic identity and their concern with financial stocks. Cuthbertson (1979, p. 12) for instance argued that while a number of policy conclusions looked similar to those of Monetarists, ‘the economic mechanisms that lie behind these New Cambridge policy conclusions are very different from those in Monetarist models’. Furthermore, he noted that ‘in terms of the underlying structure the New Cambridge model probably lies closer to the Keynesian model than it does to the Monetarist models’ (ibid., p. 12). Indeed Godley did engage with American Monetarists in a Carnegie-Rochester conference back in 1977 (Godley and Fetherston 1978), but the paper was ill-received as one would expect in such a setting, as commentators complained about various assumptions or results: central banks being able to peg the real interest rate, the low interest-elasticity of investment, the low reaction of prices to changes in output, and the rejection of the Mundell–Fleming open-economy model.

2.2 Tobin

Well, we know that Godley did follow suit, when Tobin visited the DAE in 1984 and as portfolio analysis was eventually formally included in Godley's works. Tobin was also present at the 1983 Keynes conference, where he was asked to comment on the paper by Nicholas Kaldor. It's impossible to know from the discussion found in the proceedings whether Tobin and Godley interacted during the discussions of the papers. What we know however is that Tobin (1983, p. 28) began his comment by making clear that despite the fact that much of his work was critical of Keynes, he wore ‘the label Keynesian with pride, especially nowadays’. Indeed, those economists that Tobin associated himself with most closely, as reported by Dimand (2014, p. xv), were economists that mostly strayed away from the orthodoxy of their time: Arthur Okun, Bill Brainard, Ray Fair, Robert Shiller (the closet post-Keynesian!), and Bill Nordhaus, a student and collaborator of Tobin, as well as co-author of Godley on two occasions, in a book and a paper devoted to industrial pricing.

In his book on the contributions of James Tobin, my fellow Canadian Robert Dimand (2014, p. xvii), whose doctoral thesis at Yale was supervised by Tobin, insists that ‘self-identification as a Keynesian (later as Old Keynesian) was central to Tobin's view of himself as an economist’. Why then would some authors question whether Tobin was truly a Keynesian? The answer is that Tobin was an advocate of the ‘Neoclassical Synthesis’, hence being Keynesian in the short run and Neoclassical in the long run. His version of the synthesis was somewhat different from the standard version because he attached more attention than most of his colleagues to the workings of the monetary and financial system. In that sense, Tobin was very close to Paul Davidson (1972) and Hyman Minsky (1986), as all three of them judged that their colleagues at the time, both mainstream ones and Cambridge Keynesians (including neo-Ricardians), paid insufficient attention to monetary policy and the details of the financial system. In addition, all three of them came up with some version of q theory. However, as recalled by Dimand (2014, p. 11), ‘Tobin set himself apart from Keynes's disciples at Cambridge University (such as Joan Robinson, Richard Kahn, and Nicholas Kaldor) and their Post Keynesian allies in the United States because he objected to “throwing away the insights of neoclassical economics” (in Colander 1999, 121)’.

It is in this sense that one may wonder whether Tobin was more Neoclassical than Keynesian, despite the fact that he vigorously opposed New Classical economics and the versions of New Keynesianism based on the rigidity of nominal wages and prices. Tobin defended his approach by arguing that while q theory was ‘a neoclassical theory of corporate investment … Keynes himself presented a neoclassical theory of investment’ (Tobin and Brainard 1990, p. 543).⁶ This statement is similar to that of Kaldor (1983, p. 47) who argued that ‘the real author of the so-called “neo-classical synthesis” was not Paul Samuelson, it was Keynes himself’. This is why many modern post-Keynesians are now more attracted to Kaleckian foundations, in an attempt to get rid of this Neoclassical excess baggage.

Dimand associates Tobin with four ‘Keynesian’ propositions: (i) excess demand will impact output, and not only prices, because of slow nominal adjustments; (ii) because of downward rigidity, there are lengthy periods of unemployment; (iii) investment can be volatile and driven by bouts of optimism and confidence; and finally (iv) the most central proposition, associated with chapter 19 of Keynes's General Theory and with Fisher's debt-deflation theory, ‘a falling price level is contractionary even if a lower price level is expansionary’ (Dimand 2020, p. 41, emphasis in the original). As a consequence, in some circumstances, the bouts of involuntary unemployment will not disappear on their own through market mechanisms, they will require government intervention. Tobin's fourth proposition, which is the most heterodox, started appearing in the late 1970s when Tobin was combating Friedman's natural rate of unemployment, Barro-like arguments, and Lucas's New Classical economics which were questioning the validity of Keynesian policies.

Tobin's arguments against these rival mainstream approaches that deny the existence of involuntary unemployment are found in particular in Tobin (1981). The book generated a long book review by Hyman Minsky. Minsky (1981, p. 199) starts his review by defining three traditions in macroeconomics: ‘Macroeconomic orthodoxy or the neo-classical synthesis, which encompasses both orthodox (1950 and 1960) Keynesianism and traditional Monetarism, the new classical economics, and fundamentalist Keynesianism’, which he associates with post-Keynesian authors (namely Chick, Davidson, Kregel, Moore, Robinson, Weintraub and Minsky himself). Minsky's review is highly critical of Tobin, whom he associates with the Neoclassical Synthesis. One reason may be that while Tobin (1981, p. vii) claims to be reviewing ‘current macroeconomic theory’, there is no reference whatsoever to any of the Post-Keynesians mentioned above, in particular no reference to Minsky's work on Fisher's debt–deflation theory, despite Tobin's use of the Fisher effect.⁷ Later, Minsky (1986, p. 112) concedes that ‘from time to time Tobin shows signs of being a Keynesian rather than a neoclassical Keynesian’.

The distance between Tobin and Post-Keynesians such as Wynne Godley at the time can perhaps best be appreciated by a comparison of the two public addresses that they did at Dalhousie University, in Halifax, in the autumn of 1982. Both were asked to speak on the economic decline of the time. While Tobin (1984) presents some arguments in favour of employment hysteresis and hence concedes the need for ‘prudent expansionary policies’, thus claiming that the non-accelerating inflation rate of unemployment (NAIRU) is affected by the level of aggregate demand (definitively a Keynesian argument), he also argues that the recovery in Europe has been weaker and slower because of greater rigidity in real wages. He claims that the ‘economy would function better … if the discipline of wages by unemployed workers were exerted more directly and more quickly’, thus putting forth an insider–outsider theory of unemployment, arguing as well that minimum-wage legislation is likely to raise the NAIRU (Tobin 1984, pp. 35–36).

In his Dalhousie contribution, Godley (1984, p. 68) faults Tobin and Buiter (1976), who tried to defend Keynesian policies against the attacks of Monetarists when discussing whether expansionary fiscal policy raises aggregate demand in a permanent or temporary way, for assuming that ‘the monetary authorities … are determined to keep the stock of money (i.e., a subcategory of the total stock of financial assets) on a predetermined path irrespective of fiscal policy’. He recalls that Tobin and Buiter (1976, p. 274) develop a stock–flow IS–LM model where money is exogenous, which they start by saying that this is ‘a theoretical exercise addressed to a rather esoteric and artificial question in the logic of aggregate demand’. Godley (1984, pp. 68) insists that such use of a comparative static framework requires ‘a lot of complex and opaque mathematical paraphernalia’. By contrast, in Godley and Cripps (1983), the stock of money is endogenous and models do not rely on comparative statics. Godley (1984, pp. 69) objects clearly to the mainstream claims that ‘the long-run effects of an expansionary fiscal policy on real output and unemployment were uncertain and probably perverse’ and that fiscal policy should be subordinated to monetary policy. He argued that ‘the effect of fiscal policy on real demand, output and employment can be rapid and permanent while its effect on inflation is slow and uncertain’ (Godley 1984, p. 73). This belief was underscored by a previous statement of his, rejecting the unavoidability of the NAIRU and even of the standard Phillips curve: ‘I do not accept that it is a foregone conclusion that inflation rates will be higher if unemployment is lower’ (Godley 1983, p. 170).

3.1 Terminology

Godley and Cripps (1983) were already aware of the work being conducted by Tobin and his other Keynesian colleagues, such as Brainard, Buiter, Turnovsky, Blinder and Solow, but they thought that their analysis was overly complicated and tarnished by unrealistic behavioural hypotheses. Neoclassical Synthesis Keynesians ‘could only give vague and complicated answers to simple questions like how money is created and what functions it fulfils … . Such questions tended to produce tormented replies’ (Godley and Cripps 1983, p. 15). Indeed, in his Dalhousie lecture, Godley (1984, p. 78) had exclaimed, ‘If only the Keynesian model had started off life with the national income flow identity embedded in a system of balance sheets recording stocks of assets and liabilities at the beginning and end of each accounting period!’. There was a need for Godley to incorporate Tobin's insights into his own thinking, as suggested by discussants, but he did not yet see how this could be done.

The transition from the 1983 Macroeconomics book written with Francis Cripps to the first self-contained stock–flow consistent model, the 1996 Levy Economics Institute working paper, was a long trek, with several detours. By 1996, Godley was convinced that he had achieved a breakthrough, in having managed to provide a full integration of two approaches: on the one hand, his own views about how a realistic modern monetary economy works, based on a credit-led economy, with endogenous money and processes arising in historical time in a context of uncertainty, as they had been described in the 1983 Macroeconomics book and as they arose in the tradition of a large number of Post-Keynesian authors; and on the other hand, the systematic approach advocated by Tobin, which included several financial assets issued and held by different sectors of the economy along with their adding-up conditions, and which combined standard national accounts with flows of funds, making sure, as Godley (1996, p. 7) put it, that ‘there are no black holes’. Godley (1996, p. 3) notes that his ‘debt to Tobin is enormous; I could not possibly have made this model without his work, particularly on asset choice’. This acknowledgement is repeated later: ‘The way this has been modelled owes everything to the work of James Tobin’ (Godley 1999, p. 397).

However, in 1998, Tobin came out with a new book – Tobin and Golub (1998). When he heard about it, Godley was quite anxious for a while, as he once confided to me. He feared that Tobin would pull on him the trick that previously had hurt famous economists. We know that following the publication of Keynes's (1930) Treatise on Money Schumpeter decided to give up the completion and publication of his own Das Wesen des Geldes (1970). Similarly, Joan Robinson (1977, pp. 8–9) reports that Kalecki felt ill and lay in bed for three days after having read Keynes's (1936) General Theory, as it was the book that he had intended to write.⁸ Godley feared that Tobin would have improved upon his approach and would have already published the kind of book that he was planning to write, but these fears were alleviated when he read the book and realized that Tobin still clung to a comparative static analysis, without the dynamic traverse analysis that was incorporated into Godley's 1996 and 1997 working papers.

Indeed, I was rather appalled to find out that when Dimand (2014, p. 148) refers to ‘modeling stock–flow consistent Tobin-style Keynesian monetary growth dynamics’, he does not cite any of Godley's works, but cites instead the numerous works of Peter Flaschel and his various comrades – for instance Asada et al. (2011) – which are indeed clearly in the SFC tradition as understood today. These works, however, often entertain a number of mainstream assumptions, for instance a constant growth rate of money or expected inflation being dependent on that growth rate. Still, a chapter by Asada et al. (2011), presumably mostly written by Tarik Mouakil, compares what they call their Keynes–Meltzer–Goodwin–Tobin model with the Godley and Lavoie (2007) models. They point out that ‘the similarities between growth SFC models … and the KMG–Tobin model are striking’ (Asada et al. 2011, p. 54).

Likewise, when Tom Palley (2013; 2017), a former Yale doctoral student, presents the so-called Structuralist Post-Keynesian monetary models, inspired as he says by Tobin's portfolio approach but with bank loans being the main driver, he does not refer to the SFC models of Godley (1999) or Godley and Lavoie (2007) despite the obvious similarities. Palley's models, however, are always in implicit algebraic form, and he provides no simulations, so that an appropriate comparison between his Structuralist model and the more Horizontalist models of Godley is a delicate affair. Also, his models are not fully consistent in the current SFC meaning of the word: for instance, there are reserves, but no government sector; there is a corporate bond rate, but no equation determining the balance-sheet constraint of firms; or there is an LM curve, but no equation determining income. Palley (2017, p. 108) – despite being published ten years after Godley and Lavoie (2007) – concludes by saying that ‘the next step is to develop a fully-reconstructed IS–LM model with the full set of stock–flow consistent relations between sectors’, but he does not seem to be willing to acknowledge that such SFC models already exist, even if they embed some distinct assumptions.

Dos Santos (2021, pp. 50–51) suggests that an extensive description of Godley's approach and that of his followers would have been ‘stock–flow consistent models of capitalist economies with realistic government institutions and developed financial markets which are closed in a Keynesian way and whose dynamics are carefully analysed by means of computer simulations’.¹⁰ While this is a correct description, obviously SFC sounds better and more attractive! Nikiforos and Zezza (2017), in their survey of stock–flow consistent macroeconomic models, also wonder whether the SFC terminology is adequate. They recall that dynamic stochastic general equilibrium (DSGE) models or the Solow growth model are indeed SFC in the sense that flows add to existing stocks.¹¹ They note however that standard open-economy models are not closed by feedback effects from other countries and hence not fully SFC in contrast to two-country models in the Godley tradition. They conclude that ‘it is true that the name “SFC” is misleading and sometimes confusing for what the post-Keynesian SFC approach wants to convey’ but that ‘it is probably too late to change the name’ (ibid., p. 1229). They believe that what is meant today by the SFC approach should by now be clear: besides the pilar of accounting consistency with its integration of the real and the monetary sides of the economy, it combines ‘a demand-led closure and a sophisticated and realistic treatment of the financial side of the economy’ (ibid., p. 1229).

3.2 General differences

The rest of this section tries to explain how Godley's modern SFC approach differs from that of Tobin or from other approaches that claim consistency. I first start with general considerations, which have been described in Godley and Lavoie (2007, ch. 13).¹² The crucial differences with regard to the role of banks will be dealt in a separate section.

First, Godley's SFC models are demand-led, as underlined above by Nikiforos and Zezza (2017). Supply constraints are conventionally modelled as fuelling inflation, at least when the rates of capacity utilization or the rates of unemployment move out of a certain range.

Perhaps most importantly, Godley's SFC models provide a fully explicit traverse towards the stationary or steady-growth state. Godley's SFC models fully integrate a monetary flow analysis (linking monetary income and expenditure with a flow demand for credit) with a portfolio analysis (which explains the various demand functions for financial assets, including the demand for a stock of money, as well as debt stocks). They show visually, through simulations, how flows and stocks gradually change in line with each other through time – hence in the short, medium and long run – as the consequence of a change in any parameter. These SFC models are a true illustration of Kalecki's (1971, p. 165) statement to the effect that ‘the long-run trend is but a slowly changing component of a chain of short-period situations’. Godley's models can simultaneously determine the evolution through time of the stock of money and wealth held by households and the flows of credit, investment, consumption and income. By contrast, in the works of Tobin, one-period models are still given an enormous amount of attention; the steady-state solution is assessed, but how it is reached is usually left rather vague. What Tobin essentially does is to study how the economy evolves as agents slowly adjust their portfolios towards their desired portfolio – no more. As Randall Wray (1992, p. 86) puts it, in Tobin's approach, ‘flow variables are exogenous, so that the models focus solely on portfolio decisions’.¹³

In Godley's SFC models, market clearing through prices only occurs in some specific financial markets. Only bond and equity prices are market-determined. Other prices are generally administered, by business firms, financial institutions, or even the monetary authorities. In the goods market, there is no market clearing at all: produced goods may be unsold, there is excess capacity and fluctuations of inventories are a key part of the analysis. ‘Decisions by households, firms and banks are mainly based not on price but on quantity signals which often take the form of realised stocks of wealth or inventories’ (Godley 1997, p. 4).¹⁴ This is in clear contrast to the standard Tobinesque closure, where changes in product prices, asset prices, and rates of return provide the main market-clearing mechanism.

Godley's SFC models rely on procedural rationality, with agents reacting to past disequilibria relative to norms. Embedded in these models is the principle of adjustment to observed disequilibria by decentralized agents or institutions. By contrast, Tobin's models usually assume profit maximization and an access to an unlikely amount of information. As Dimand (2014, p. 142) puts it, ‘Tobin thus added informal optimizing foundations for each component of the IS/LM model of aggregate demand: investment, saving, liquidity preference (money demand), and money supply’. On occasions Tobin (1981, pp. 34–35) refers to complex uncertainty, rules of thumb, satisficing procedures, and full-cost pricing. But these are only weapons to criticize the Lucasian neo-Walrasian rival; they are never used for constructive purposes.

3.3 Specific differences

After these general statements, we can move on to more specific considerations. We start with households, and then move on to business firms. We shall see that in Godley's models, non-financial firms are institutions that have a life of their own, rather than being simple veils. Because producing firms are institutions, they must take a variety of decisions, not based on standard profit-maximizing principles.

In Godley's SFC models, households make their consumption and portfolio decisions sequentially. The consumption decision is made on the basis of a Modigliani-type consumption function, with disposable income or expected disposable income and past accumulated wealth as the main two arguments of the function. This generates an expected saving level for the period, which, once added to past wealth, generates the expected end-of-period wealth. Portfolio decisions – the relative proportions of the different assets – are then taken on the basis of this expected wealth. This framework allows the use of a fairly standard consumption function despite a large set of distinct financial assets. Broadly speaking, this was also the framework advocated by Backus et al. (1980, p. 273) in the empirical version of Tobin's model. In his theoretical work however, Tobin assumes that consumption is a residual: it is the amount of income which is left over once households have purchased all the additional financial assets that they desire to acquire (their saving of the period). In that theoretical framework, everything seems to happen at once: time is condensed by the need for market clearing. One would achieve the steady state at once. To reintroduce some duration, Tobin is forced to introduce the hypothesis that households cannot achieve the desired asset ratios immediately, due to the existence of high transaction costs – a hypothesis that appears highly artificial and unlikely in today's computerized world.

In Godley's SFC models, households can make mistakes or make errors in expectations, and cash money or money deposits act as a buffer for households. Whenever there are mistaken expectations, something must give. In his models, when revenues are overestimated for instance, household cash balances or deposit balances at the bank get reduced to absorb the unexpected shortfall, as would happen in real life. No such mistakes and no such mechanisms exist in Tobin's world, although this drawback is recognized by Backus et al. (1980, p. 288) when they mention that demand deposits and currency ought to serve as ‘buffers or temporary abodes of purchasing power’, pointing out that ‘the partial adjustment assumption seems particularly inappropriate’ in this case.

In Godley's models, firms must take a production decision, and must decide on fixed capital and inventories investment. In the simplest models, firms simply produce what is being demanded. In the more sophisticated models with inventories, firms must assess the demand of the period and they must also decide on the additional inventories that they would like to hold. The decision to produce is thus intertwined with the decision to accumulate circulating capital. Firms invest on the basis of stock–flow norms, assessed on the basis of expected sales. Investment behaviour is essentially demand-led, as is the rest of the model. In Tobin's models, as well as in several other Neoclassical models, fixed capital investment relies on q theory. In a non-growing world, it is claimed that net investment should be positive whenever the q ratio is above unity, that is, whenever the stock market values equities more highly than the replacement cost of tangible capital. The logic of such an investment behaviour is that it would be conducive to shareholders potentially maximizing the value of their equity. Clearly, in that instance, firms act as a veil, acting in the interest of households. This indeed was Crotty's (1990) critique of Tobin's q theory.

In Godley's SFC models, firms must make costing and pricing decisions. Prices depend on unit costs. Cost-plus pricing is part of the model: it distributes income between labour wages, entrepreneurial profits and the interest receipts of creditors. In the more complex version, a target profit level is designed to provide a reasonable dividend yield as well as to provide internal funds to finance capital accumulation, in line with the Post-Keynesian pricing mechanisms underlined by Harcourt, Adrian Wood and Alfred Eichner. In Tobin's models, either the price level is a given or the price level is determined by a market-clearing mechanism, which clears the demand for and the supply of goods. In Tobin's models, no pricing decision needs to be taken: the market does it, or it is arbitrary.

In Godley's models, firms must take financial decisions. In a Tobinesque world, firms are indifferent between finance by loans or finance by issuing equities. Only expected financing costs matter. In Godley's models, there is a hierarchy of financing means, and decisions do have consequences. Firms can decide whether they wish to issue more shares, or issue or retire corporate paper, or simply draw on their banking line of credit. In his growth models, producing firms also accumulate retained earnings, as do firms in Backus et al. (1980). The problem with the Backus et al. presentation, however, is that no sooner were retained earnings and financial decisions introduced as a realistic element of their model than another hypothesis cancelled the realism of their firms. Indeed, Backus et al. (1980, p. 266) assume that retained earnings are ‘dividends paid matched by sales of shares’. In other words, ‘business retained earnings … are imputed to shareowners … as if they were dividends … . Retention of earnings is an issue of equity by business and a purchase of equity by households and other shareowners’ (ibid., p. 268). No such strange assumption is needed in Godley's models.

4.1 Godley

Godley (1999, p. 393) starts the presentation of his SFC model by criticizing the standard assumption of an exogenous money supply, saying that he will rely on the ‘new endogenous money (EM) school’ of Post-Keynesianism, which he also associates with Schumpeter and the late Hicks. His purpose is to incorporate ‘EM ideas into a complete, if very simplified, model of the whole economy’ (ibid., p. 394). The fact that the Post-Keynesian theory of endogenous money is emphasized at the very beginning of his first published SFC article demonstrates how important it is to Godley's thinking. In Godley's SFC models, as in his earlier book with Francis Cripps, banks are creators of credit money, and they play an essential systemic role. Godley's SFC models are consistent with the theory of the monetary circuit, according to which production must be initially financed by bank loans to get the ball rolling, making the distinction between initial and final finance (Graziani 1989; Godley 2004; Lavoie 2021).

Under this view, the predominant role of banks is to create loans, providing credit to firms which carry on production in a world where goods take time to be produced and sold. As Zezza (2012, p. 156) puts it, ‘at the heart of the TMC [theory of the monetary circuit] is the notion, shared by Godley, that production requires time, and that costs of production have to be paid before receipts from sales can be obtained’. Godley's banks are Kaldorian, responding to the financial needs of their creditworthy clients. In addition, bank loans to firms act as a buffer for the fluctuations in inventories and are required for dividends to be distributed. Inventories, or unsold production, must be financed by bank loans, otherwise the profits of producing firms, as assessed by accountants, just could not be distributed to shareholders. The money to be distributed would not be there.

Godley's banks are also institutions that do take pricing decisions: they set deposit and lending rates. By contrast, in Tobinesque models, deposit and lending rates are market-clearing prices, which adjust the demand for and supply of deposits and loans respectively. In the simplest Godley models, lending rates are set as a mark-up over deposit rates, as in many Post-Keynesian macroeconomic models.¹⁶ In the more realistic models, lending rates are such that they allow banks to attain their target level of profits, as was the case with the prices set by producing firms.¹⁷ As to deposit rates, it is assumed in Godley (1999) and in Godley and Lavoie (2007) that rates on term deposit are hiked up or pushed down whenever the bank's liquidity ratio goes outside its target range.

4.2 Tobin

Tobin's view of banks is different. As is clear from Brainard and Tobin (1968), banks, like households, are assumed to make portfolio asset choices, based on rates of return, among free reserve assets, loans, and government bills. Loans play no special role in this approach – they have no priority – and banks could as well be non-banking financial institutions. This impression is reinforced by a reading of Tobin (1969) and Tobin (1982).¹⁸ There, bank loans are omitted altogether from the formal model, and when banks are mentioned, it is claimed that the ‘traditional business of commercial banks is to accept deposits … and to acquire assets of less liquidity and maturity … . Other intermediaries likewise transform their assets into forms better tailored … to the preferences and circumstances of their creditors’ (Tobin 1982, p. 193).

Backus et al. (1980, p. 265) also formally describe banks as pure intermediaries, but they do concede that, more realistically, bank loans play a special role in monetary production economy, admitting that ‘banks regard business loans as a prior claim on their disposable funds, and meet these demands at the prevailing rate, only later adjusting this rate in the direction that brings loan demand closer to the bank's desired supply’. Similarly, Brainard and Tobin (1968, p. 110), concede that ‘in the short run banks meet from excess reserves whatever loan demand comes their way at the established interest rate’. This ‘more realistic’ accommodating bank behaviour is more in line with the role of banks in Godley's models, but as was said earlier, it disappears from Tobin's theoretical models.

Dimand (2014, p. 141) praises Tobin for not relying on the quantity theory of money expounded by the Monetarists, where the quantity of money is set exogenously by the central bank through the money multiplier. Rather, he says, in Tobin's view, ‘the central bank sets the monetary base (outside money, that is currency plus reserves with zero or other exogenously fixed rate of return), and then optimizing financial institutions created financial assets including inside money’ (ibid., p. 141). The money supply is thus endogenous, but not endogenous according to the meaning given to the term by Post-Keynesian authors (including Godley) belonging to the so-called Horizontalist tradition, who focus on reversed causation, with loans determining the level of deposits and deposits determining the amount of reserves in the system. What is endogenous in Tobin's works is the money multiplier or the velocity of money, something which at best one could call weak endogeneity. As Dimand (2014, p. 142) continues, ‘Tobin … explicitly modeled money creation by optimizing banks to derive an upward sloping supply curve for money: higher interest rates induce banks to create more money by choosing lower reserve/deposit ratios’.¹⁹

In Tobin's models, although the money supply appears to be endogenous, ultimately it is not. As Palley (2013, p. 410) puts it, ‘bank lending remains completely invisible and is attributed no role in the money supply process … . Though the money supply is endogenous, the monetary base remains exogenous which is what gives the model its verticalist character’. This is even clearer when Tobin deals with fiscal policy. Both Tobin (1982, p. 182) and Backus et al. (1980, p. 267) assume that a given fixed proportion of the government deficit is being monetized, which goes against the principle of a truly demand-led endogenous supply of money. In the Post-Keynesian view and that of MMT, the proportions of money-financed and bond-financed government deficits are determined by the portfolio choices of the public.²⁰

4.3 The controversy around Tobin's old and new views of banking

Over the last few years there has been a series of papers devoted to the various conceptions of banks, many of which have arisen from researchers working in central banks. Best known is the paper by McLeay et al. (2014, p. 14), from the Bank of England, who argued that ‘money creation in practice differs from some popular misconceptions – banks do not act simply as intermediaries, lending out deposits that savers place with them, and nor do they “multiply up” central bank money to create new loans and deposits’. These papers have criticized the textbook money-multiplier story as well as the loanable-funds approach, providing support to the long-held Post-Keynesian credit-creating view of banking which can also be found in Godley's works. Different names have been proposed for this third view of banks: credit-creation theory (Werner 2016); endogenous-money theory (Ábel et al. 2016); financing through money-creation theory (Jakab and Kumhof 2015; 2019); originate inside money theory (Bianco and Sardoni 2018); bank-originated money and debt theory (Keen 2020).

Although apparently different, the loanable-funds theory of banking and the money (or deposit) multiplier theory are in fact related. In the loanable-funds theory of banking, banks are just like all other financial intermediaries, which lend funds that they have been entrusted with. In the caricature provided by Jakab and Kumhof (2015), which they associate with what they call the intermediation of loanable-funds theory of banking, savers bring gold or gravel and deposit these at banks, which can then be lent to borrowers. Ábel et al. (2016) and Werner (2016) simply call it the financial intermediation theory. In a sense, this is no different from the fractional-reserve theory of banking (as Werner 2016 calls it) associated with the money-multiplier story. Within that context, as Bianco and Sardoni (2018, p. 170) put it, ‘banks are intermediaries of loanable funds in a fractional reserve environment’. With bank reserves at the central bank also being called outside money, ‘saying that banks are intermediaries of loanable funds is just the same as saying that banks are intermediaries of outside money’ (ibid., p. 170). The reserves play the role of loanable funds; without excess reserves, no loans could be made.²²

All these authors blame Tobin (1963) ‘for having played a critical role in establishing the ILF [intermediation of loanable funds] view of Gurley and Shaw as the new dominant paradigm’ (Jakab and Kumhof 2015, p. 7), as he argued that banks faced the same constraints as other financial institutions, because bank deposits are also subjected to the portfolio decisions of households.²³ I must say that, many years ago, as a young scholar trained in the Post-Keynesian and Circuitist version of endogenous money, I was utterly confused when I read and tried to understand Tobin's (1963) new view, now said to be at the origin of the intermediary theory of banking, as compared to what he called the old view. Tobin seemed to associate the fractional-reserve or money-multiplier theory of banking, what he called the old view, with the idea that bank loans can be created ‘ex nihilo’, at the stroke of a pen, as Post-Keynesians would have it, so that loans led to deposits. My understanding of the money-multiplier story was quite different from Tobin's: in the fractional-reserve world, it is the initial acquisition of cash or deposits by banks that leads to the appropriation of excess reserves, which then allows the creation of bank loans. No wonder I got confused by Tobin's old and new views of banking!

Despite the fact that Tobin claims that in his portfolio models one of the closures could be a closure where income is endogenous, in fact he does not deliver on that closure. As recalled earlier by a quote from Wray (1992), when dealing with banks Tobin essentially assumes that output is a given, while actual portfolios adjust to the desired ones. His analysis of banking is more static than dynamic. While Tobin recognizes that banks can create credit money ex nihilo in contrast to other financial institutions, he believes that banks can only do so if agents are willing to modify the composition of their portfolio in favour of bank deposits, thus being subjected to the same constraints as any financial intermediary. But the constraints are not identical. Non-banks, if they wish to create more credit, must first get either more deposits from the public or advances from banks. Tobin would argue that the situation is similar for banks: if their depositors decide to transfer their bank deposits at accounts in non-banks, then the banks in their turn will need to access funds from non-banks.

The difference, however, is that when non-banks acquire assets, they must obtain outside financing to start with, at the very first stage; when banks borrow, this may happen, but only at a later stage. This distinction is hard to make if one assumes a static economy where production does not change and where everything happens at once. The distinction is made in particular by Post-Keynesian Circuitists. At the start of the circuit, the credit creation by banks is associated with initial finance; it leads to an immediate increase on both sides of the balance sheets. The funds that need to be recovered by the end of the circuit are final finance, and are mainly the consequences of the portfolio decisions of non-financial agents. They arise from the decision about where to assign the flow of saving and the reallocation of wealth. As Jakab and Kumhof (2015, p. 8, emphasis in the original) say, ‘the critical insight is that banks can create their own funds instantaneously’.²⁴

Godley has a completely different perception from Tobin of how banks can handle leaks in the deposits that they have created when they granted loans. This is quite clear in the Godley and Cripps (1983) book. First, they note that firms often draw on overdraft facilities, so that the outstanding quantity of loans is not under the full control of banks. They then underline that in practice it would be impossible for banks to fulfil their role if there had to be a continuous equivalence between the loans granted by a bank and the deposits it collects. Godley and Cripps (1983, p. 77) remark, however, that ‘this problem is readily overcome … given a system of inter-bank credits’, making it even possible for some banks to specialize in making loans while others specialize in deposit-taking.²⁵ They conclude that ‘once a system of inter-bank credits is in existence, there is no logical or institutional constraint on the extent to which the whole banking system can supply additional loans … . Changes in the stock of loans and money will be governed solely by demand for loans and the credit-worthiness of would-be borrowers’ (ibid., p. 77).