Kalecki s 1934 model VS. the IS-LM model of Hicks (1937) and Modigliani (1944)*

Transcription

1 Euro. J. History of Economic Thought 14: March 2007 Kalecki s 1934 model VS. the IS-LM model of Hicks (1937) and Modigliani (1944)* Michaël Assous 1. Introduction In his influential book Anticipations of the General Theory? Patinkin (1982) concluded that before the publication of the General Theory Kalecki did not deal with the notion of unemployment equilibrium in terms of a general equilibrium system of simultaneous equations. In short, Patinkin claimed Kalecki did not anticipate the Keynesian model, 1 of which the more relevant interpretation, according to him, is the IS-LM model (Patinkin 1990a,b). In 1995, Simon Chapple claimed in a closely argued article that: an early version of the mainstream Keynesian model was constructed and published by Kalecki before 1936 (Chapple 1995: 521). 2 Focusing on Address for correspondence PHARE-CNRS, Maison des Sciences Economiques, , boulevard de l Hôpital, Paris Cedex 13, France; michael.assous@wanadoo.fr * I am grateful to Professors Richard Arena, Rodolphe Dos Santos Ferreira, Gilbert Faccarello, Harald Hageman, Heinz Kurz and Antoine Rebeyrol for helpul comments and suggestions on an earlier draft. I am especially indebted to Professor Alain Béraud, with whom I had lengthy exchanges. I also gratefully acknowledge Claude Marguet for detailed comments and useful observations. Helpful remarks of two anonymous referees are gratefully acknowledged. Any remaining errors in this paper are mine. 1 In his 1982 study, Patinkin affirmed that Kalecki had not analysed the mechanisms by which the economy is likely to reach equilibrium with unemployment without contrasting it with classical mechanisms. Moreover, Patinkin did not think that Kalecki defined a general equilibrium model like the one described by Hicks in 1937 (Patinkin 1982: 10 11). 2 Chapple aimed to demonstrate that Kalecki anticipated the key features of the General Theory, which, as Patinkin defined them, are threefold. First, he claimed Kalecki s works prior to the General Theory s publication contained the notion of effective demand whose essence is, according to Patinkin, the well-known forty- The European Journal of the History of Economic Thought ISSN print/issn online Ó 2007 Taylor & Francis DOI: /

2 Michaël Assous Kalecki s (1934) article, 3 Chapple showed that Kalecki had constructed three variants of the IS-LM model that allowed him to mimic the principle conclusions of the neoclassical theory and to explain the persistence of unemployment. Centring on a discussion of Patinkin s argument, Chapple pushed to the background the differences between Kalecki s (1934) model and the IS-LM model. The aim of this paper is to highlight these differences 4 by showing how Kalecki s model differs significantly from the two main variants of the IS-LM model, those of Hicks (1937) and Modigliani (1944). 5 Based on this twofold comparison, the paper then shows that Kalecki s model offers an original explanation of the difference between classical models (based on Say s law) and types of models that were to be called later Keynesian models. Showing that Kalecki s theory is concerned, strictly speaking, with a situation of unemployment quasi-equilibrium, one then understands that the validity of his analysis does not depend on the existence of either of these special assumptions of the liquidity trap (Hicks) or alternatively absolute rigid money wages (Modigliani). Indeed, as Kalecki stressed in the conclusion of his 1934 article, his theory aims at analysing the situation of five-degree diagram (Chapple 1991). Second, contrary to Patinkin, Chapple suggests that Kalecki provided an integrated treatment of goods market equilibrium with money market equilibrium (Chapple 1995, Osiatynski 1985, 1992). Last, he rebutted Patinkin s argument that Kalecki did not link aggregate demand with the marginalist theory of short-run aggregate demand (Chapple 1995). 3 Kalecki s 1934 article was originally published in Polish in the main Polish economic review Ekonomista and was translated into English only in volume 1 of Kalecki s Collected Works. The fact that Kalecki did not choose to translate this article to claim anticipation of the General Theory continues to be ignored by Patinkin s criteria. (See Chapple 1991 on the discussion of Patinkin s criteria.) 4 Chapple noticed briefly how Kalecki s model differs from the textbook IS-LM version, emphasizing only in passing the specificity of Kalecki s treatment of the labour market in the unemployment variant of his model. 5 In his influential 1944 article, Modigliani recast Hicks initial model into what was to become the standard version of the IS-LM model (see Darity and Young 1995, Barens and Caspari 1999, De Vroey 2000, Young 1987). Indeed, when Hicks opposed Keynes and the classics, he admitted that money wages are given both in the Keynesian and the classical models. Hence, Keynes s main contribution is that of having built a model based on the theory of liquidity preference. Modigliani, however, found Hicks analysis flawed. In 1944, he presented a new version of the difference between Keynes and the classics. To him, the liquidity preference theory is fully acceptable in a classical model. Keynes s essential contribution would be that he showed that a macroeconomic equilibrium with unemployment is possible when money wages are rigid. It is around this idea that a synthesis between the classical tradition and the Keynesian revolution was developed. 98

3 Kalecki s (1934) model disequilibrium unemployment and not the situation of unemployment equilibrium. As soon as the assumption of a given volume and structure of capital equipment is abandoned, then, as a result of changes in capital stock, there will be a continual movement through a series of quasi-equilibria. Thus, even if money wages adjust in response to unemployment movements, the economy will not necessarily reach a position of full employment. 6,7 This paper is organized as follows. The first section outlines the construction of Kalecki s (1934) article. Starting from Kalecki s analysis of classical economics, this section reconsiders the crucial steps in the process of constructing Kalecki s unemployment model and proposes a formalization of Kalecki s argument. The last two sections then compare, respectively, Kalecki s article with the Hicks and Modigliani IS-LM models, focusing on differences that affect the structure of the economy, the effect of demand shocks on employment and unemployment analysis. 2. A reconstruction of Kalecki s Three Systems 2.1. Systems I and II Kalecki s 1934 model describes a perfectly competitive economy whose employed workers consume their entire wages. 8 The first variant of this 6 It is worth stressing Kalecki s analysis differs also from Patinkin s own IS-LM model in terms of unemployment disequilibrium whose differences with Modigliani s 1944 model are discussed by G. Rubin in the 2004 supplement to History of Political Economy. Patinkin s model is based on the idea that when money wages decline in the face of excess supply of labour, the economy does not steer itself to full employment. His message is that even if full employment equilibrium is globally stable, disequilibrium can be protracted and stubborn. By assuming money wages do not fall in the face of excess supply of labour, Kalecki underlined on the contrary that disequilibrium does not depend on money wages adjustments although induced variations on money wages play a part on employment variations but on investment variations caused by the evolution of the profitability of equipment. 7 It is important to stress that in his 1934 perfectly competitive framework, the focus of attention in terms of sectors of the economy was not the product markets. In Kalecki s model, prices are viewed as moving in line with marginal costs so that the major cause of unemployment cannot be seen to be a mismatch between the degree of monopoly, equal to zero, and the level of investment expenditures (see Sawyer 1985, Lopez and Assous 2007 on the importance of imperfect competition in Kalecki s latter works) but only on the weakness of capitalist expenditures. 8 Both production sectors operate with a constant and historically given capital stock in which technology exhibits decreasing marginal productivity of labour. 99

4 Michaël Assous model System I is a classical model founded on Say s law. 9 Kalecki emphasized this point by considering two shocks: a rise in the labour supply and an exogenous reduction in capitalists consumption capitalists consumption being itself considered exogenously given. In both cases he showed that the production of investment goods increases. As Kalecki stressed, an excess supply of labour reduces money wages, causing on the one hand a rise in employment and aggregate production because of the fall in real cost and on the other hand a rise in investment. Indeed, according to Say s law and capitalists consumption assumed given, capitalists invest the profits due to the fall in money wages. Finally, because there is at the same time a rise in demand and in profitable output, a level of macroeconomic equilibrium, characterized by a higher level of employment and of production of investment goods, is reached. Considering the labour supply as constant, Kalecki envisioned a second shock: an exogenous fall in capitalists consumption. Again, his analysis rested on Say s law. Thus, by reducing their consumption, capitalists correspondingly increase investment. The price of investment goods rises because demand is greater whereas the price of consumer goods falls because demand is smaller. Finally, employment and production rise in the investment goods sector and shrink in the consumption goods sector (Kalecki 1990: 205). Then, Kalecki concluded, the production of investment goods is an increasing function of the supply of labour (assumed inelastic) and a decreasing function of capitalists consumption: 10 I ¼ f N ; C p ð1þ Investment demand is assumed to depend negatively on the interest rate and positively on the current profitability of equipment for which entrepreneurs expect the return of their investment projects: The number of investment projects which pass the profitability test depends on the mutual relation at a given moment between prices of consumer goods, prices of Profit maximization under perfect competition is then assumed as prices are equal to marginal costs. Implicit assumptions include a closed economy and no government sector. 9 Kalecki characterize Say s law as follows: In System I, the principle of preservation of purchasing power is pushed to the extreme: all income must be spent immediately on consumer or investment goods. This model is in fact accepted by all classical economists (Kalecki 1990: 201). 10 The notation used by Kalecki has been replaced by the more conventional ones. 100

5 Kalecki s (1934) model investment goods, and wages (which are determinants of the expected gross profitability), and on the rate of interest. (ibid: 206) Hence, since the supply of labour and capitalist consumption entirely determines the relation of prices and wages, investment demand can be presented as the function C N ; C p ; r (ibid: 206). Assessing the production of investment goods is determined by equation (1) and the demand for investment goods is represented by the function C N ; C p ; r one obtains the condition of equilibrium in the investment good market from which the equilibrium rate of interest is obtained: I ¼ C N ; C p ; r ð2þ The functions f and C thus determine investment goods output and the rate of interest. The formal model underlying Kalecki s System I can be represented as follows: C ¼ f C ðn C Þ I ¼ f I ðn I Þ ð1:1þ ð1:2þ W ¼ p C f 0 Cð N CÞ ð1:3þ W ¼ p I f 0 I ðn I Þ N I þ N C ¼ N p C I ¼ I W ; p I W ; r; g C ¼ C p þ WN p C M ¼ kðp I I þ p C CÞ N ¼ N ð1:4þ ð1:5þ ð1:6þ ð1:7þ ð1:8þ ð1:9þ Equations (1.1) and (1.2) represent the sectoral production functions where C is the output of consumer goods and I is the output of investment goods. N c, N I is employment in the consumer-good (investment-good) sector. The marginal productivities in both sectors are equal to product 101

6 Michaël Assous wages (equations (1.3) and (1.4)). N I plus N c results in aggregate employment demanded (equation (1.5)). Real investment depends on the inverse of the product wages of the two production sectors 11 and on the rate of interest (equation (1.6)). The parameter g has been added to represent explicitly a propensity to invest. 12 The level of consumption demand is equal to the demand of capitalists and the demand of workers who consume their entire wages (equation (1.7)). Nominal money demand function is written, in accordance with the quantity theory, as a function of nominal income. By equating this demand function with the quantity of money, M, one gets the equilibrium condition of the money market (equation (1.8)). Finally, because the labour market is balanced, employment is equal to labour supply (equation (1.9)). The endogenous variables are: N c, N I, N, C, I, p c, p I, r, W. The exogenous variables are: N ; M ; C p. Equations (1.1), (1.2), (1.3), (1.7) and (1.9) result in Kalecki s equation (1). Equations (1.1), (1.3), (1.4), (1.5), (1.6), (1.7) and (1.9) result in Kalecki s equation (2). (The solution of the model is discussed in Appendix 1.) Thus, by constructing a model based on Say s law, Kalecki described an economy for which real variables and nominal variables are respectively determined by the real and the monetary parts of the model and in which market mechanisms spontaneously re-establish full employment. In order to determine whether this result depends on the absence of hoarding, Kalecki considered in his System II the implications of variations of cash reserves owned by firms. In Kalecki s System II, money supply is first assumed given. 13 Money demand is instead assumed to increase with income and to decline with the interest rate. More precisely, Kalecki argued that agents choose between cash reserves, which they need in order to make transactions insisting on the transaction motive for financing production and financial assets, which do not allow making transactions but yield interest. In contrast to System I, individual economic agents in System II hold cash reserves which can be increased or decreased. A cash reserve is necessary to run an enterprise at a given turnover smoothly. The volume of this reserve depends not only on the turnover of the enterprise, but also on the rate of interest. The higher the rate of interest, the smaller the cash reserve held by an enterprise at a given turnover. Hence if sales increase while the volume of money in circulation remains constant, that is, if 11 Current real profits by unit produced in each production sector depend respectively on p c /W and p I /W; they in turn determine expected profitability and hence investment. 12 In Kalecki s analysis, investment can be increased in response to a Schumpeterian new production combination (Kalecki 1990: 206). 13 After having presented in depth his second system, Kalecki dealt with the implications of increasing money supply with interest rates (Kalecki 1990: 213 4). 102

7 Kalecki s (1934) model the velocity of money circulation increases, the rate of interest rises, since there will be a tendency to increase reserves in the same relation, which must be counteracted by an increase in the rate of interest. The rate of interest in System II is determined in this way by the velocity of money circulation. (ibid: 207) Formally, by assuming that the elasticity of money demand with regard to nominal income is equal to 1, the money demand function described by Kalecki can be written as follows: M d ¼ (p I I þ p c C)L(r), where the function L is a decreasing function of the rate of interest. From the condition of equilibrium on the money market, M ¼ðp I I þ p C CÞLðr Þ, one obtains the velocity of money circulation: V ¼ðp I I þ p C CÞ=M ¼ 1=Lðr Þ. It thus appears that when nominal income rises, the velocity of money circulation increases and equilibrium on the money market is re-established by a rise of the interest rate. By adding this money market conception to his System I, Kalecki showed, however, that the final position of equilibrium in this system is the same as under Say s law. Consider his analysis of the impact of a rise in labour supply. 14 Due to the complete flexibility of money wages, an excess supply causes money wages 14 Kalecki also illustrated this point by considering the impact of an exogenous decrease in capitalists consumption and a rise in the incentive to invest. Kalecki dealt with the impact of an exogenous reduction in the volume of capitalists consumption given supply labour. Capitalists, instead of investing, increase their money reserves. In the sector of consumption goods, supply exceeds demand, so the price decreases until equilibrium is re-established, which causes a rise in real wages and the reduction of employment (ibid: 210). With an excess supply of labour, money wages decrease, allowing firms of the investment sector to hire the workers dismissed from the consumer sector (ibid: 210). Production increases in the investment goods sector, which enables a lowering in prices and a rise in real balances. More real balances are then available for the financing of production, which lowers the interest rate and enables a rise in investment (ibid: 211). Thus, capitalists finally put their demand for consumption goods entirely on the investment goods sector so that the economy reaches a full employment equilibrium. Then, Kalecki focuses on the implications of a rise in the inducement to invest, given the supply of labour and of capitalists consumption. Because of the rise in demand, prices of investment goods increase, causing a decrease in real costs and a rise in labour demand. Some workers move from the consumption to the investment sector, which decreases consumption goods output. For a given volume of capitalist consumption, demand exceeds supply and consumption. Prices rise until a new equilibrium is established, which involves a lower real wage rate, and therefore a somewhat higher output of consumer goods than without a real wage reduction. Consequently, as Kalecki argued, wages go up, and a number of workers now return to this industry from the investment sector. Production of the latter falls. In this way, we return to the initial position, except that the general level of prices and wages has risen (ibid: 209). As a 103

8 Michaël Assous to fall. Real wages decrease, causing a rise in employment and production. As a result, prices decrease due to the appearance of an excess supply of goods, which results in a rise in the value of money holdings. More real balances are then available for financing production activities, causing the interest rate to fall and permitting investment to increase on account of the falling money value of sales the velocity of money circulation declines and with it also the money rate of interest, which encourages entrepreneurs to make investments (Kalecki 1990: 212). A set of real variables identical to the one defined by Kalecki s first model is thus determined. (The solutions of Kalecki s System II are discussed in Appendix 2.) This adjusting mechanism, through which lower prices and wages could eventually generate a move towards full employment, relies entirely on the Keynes effect. Disequilibrium on the labour market indeed entails a variation in money wages, which causes a variation in price. This variation of price modifies the real value of money supply, which lowers the interest rate and stimulates investment. This process occurs until income and production reach a level ensuring equilibrium in all markets. As Kalecki stressed: [T]his is the essence of arriving at equilibrium identical with one which would be established in System I (Kalecki 1990: 214 5). So, when prices and money wages are completely flexible and the Keynes effect applied, Say s law is still valid. It is by modifying the conception of the labour market in this second model that Kalecki suggests Say s law could be invalidated, thus showing that the economy could get stuck in a position of quasi-equilibrium System III There is a radical difference between Kalecki s third model and his first two models with regard to the functioning of the labour market. The central hypothesis at the core of this difference is that unemployment, as such, is no longer supposed to push money wages down. Kalecki argues as follows: [A]s long as it remains unchanged, existing unemployment does not pressure the market. Without going into the reasons for this, we shall continue to study System II, except that now it permits the existence of some reserve army of the unemployed. This we call System III. (Kalecki 1990: 215) consequence, less real balances are available for financing production. So the interest rate increases until the volume of investment projects is reduced to the initial level (and naturally new production combinations are realized by cancelling other projects which are unprofitable at a higher rate of interest) (ibid: 209). 104

9 Kalecki s (1934) model Was Kalecki insisting on the difficulty and the time necessary to render money wages flexibly downwards or was he referring to some different adjustment mechanism? The second characteristic of his conception of the labour market provides some clarification: Namely, while the existing [emphasis in the original] unemployment does not exert any pressure on the market, we postulate that changes [emphasis in the original] in unemployment cause a definite increase or fall in money wages, depending on the direction and volume of these changes. (Kalecki 1990: 215) This conception of the labour market obviously has its roots in Marxian economics. It is indeed Marx who developed the concept of the reserve army of the unemployed, the role of which was to regulate the capitalist system by exerting a disciplinary effect. Kalecki certainly thought that falling (rising) unemployment increases (decreases) the power of workers to press for higher (lower) wages. 15 The first hypothesis allows the determination of what Kalecki called a position of quasi-equilibrium; it can be defined by a set of equations identical to that of Kalecki s second model, except that in each equation the level of the supply of labour has been replaced by the level of actual employment. Thus, as soon as actual employment is known, the quasiequilibrium is determined. Yet if this level of employment is undetermined, then so are quasi-equilibria. Kalecki s second hypothesis, according to which money wages are related to the level of unemployment referred to as follows with the equation W ¼ g ðn N Þ, where g 5 0 allows one to define a quasi-equilibrium (Kalecki 1990: 215 6). By replacing equation (1.9) with the equation W ¼ g ðn N Þ, Kalecki s third model is obtained. The endogenous variables remain N c, N I, N, C, I, p c, p I, r, W. and the exogenous ones are N ; M; C p. The model still has nine equations (see Appendix 3). However, contrary to the other model, it is not dichotomic so that shocks in demand now have an impact on employment. To show this, Kalecki carries out two comparative statics exercises: first, an improvement in the inducement to invest; and second, a cut in capitalists consumption expenditures. Consider the effects of an increase in the inducement to invest. This leads to an increase in the price of investment goods. As a result, production and employment expand in the investment sector. In turn, this causes increased worker s consumption, which boosts price and production 15 In an imperfect competition framework, Kalecki represented the increase in worker s power associated with a boom by a decline in mark-up in the pricing equation (Kalecki 1971). 105

10 Michaël Assous in the consumption good sector. As capitalists consumption is given, aggregate production will expand until profits increase by the same amount as the increase in real investment. Kalecki s System III allows then the expression of his theory of profit whereby capitalists get what they spend (Kalecki 1990: 216 7). However, this is not the end result. As Kalecki emphasized, the rise in prices and in money wages due to increases in employment and production, leads to a rise in the money value of turnover ; this also causes a rise in the transaction demand for money that can only be met by an increase in the rate of interest, which in turn reduces the volume of investment (see Kalecki 1990: 217). But despite this depressive effect, the new quasi-equilibrium is established at a higher level of employment because of the upward movement of the schedule of marginal profitability of new investment projects: the increased output and rise in prices in relation to wages in turn increase profitability, which additionally stimulates investment activity (Kalecki 1990: 217). Now consider how Kalecki envisions the effect of an exogenous decrease in capitalists consumption. The price of consumption goods decreases and production falls, which results in workers being pushed to the reserve army of labour. Higher unemployment reduces consumer goods demand. Prices, output and employment in the consumer goods sector decrease until profits have fallen by the amount of the capitalist consumption decrease. Then, because of the rise in unemployment, wages eventually go down. However, as long as investment does not vary, prices in the consumption goods sector fall pari passu as the money wages do, without entailing a reduction in real cost. But if the lowering of money wages does not affect firms costs, they reduce, however, the interest rate, which causes a rise in investment and the hiring of some workers pushed initially into the reserve army of the unemployed. Yet, in spite of the decrease in interest rate, investment is likely to fall due to profitability deterioration. Thus, Kalecki came to the conclusion that a decrease in capitalists consumption, and so a rise in savings, can reduce investment and drive the economy into a position where unemployment is higher. Having explained the three variants of Kalecki s 1934 model,now compare it with the IS-LM model, focusing attention on the versions described by Hicks (1937) and Modigliani (1944). 3. Three Systems vs. the IS-LM model of Hicks (1937) To draw a contrast between the classical and the Keynesian perspectives, Hicks also constructed three models: the first he qualified as being classical; 16 the second Keynesian; 17 and the third a synthesis, 18 two variants 106

11 Kalecki s (1934) model 16 Hicks first system is a classical system in which money demand, in accordance with the quantitative theory of money, does not depend on the interest rate. Hicks presented it as follow: M ¼ ky n ; I n ¼ I n ðrþ; I n ¼ S n ðr; Y n Þ Y n, is nominal income, I n is nominal investment, r is the interest rate, M the quantity of money in circulation supposed given and k a constant corresponding to the inverse of the velocity of money circulation. Hicks showed how a rise in the inducement to invest in this model affects only the interest rate and leaves nominal income as it is. Consequently, employment will vary only if the supply elasticity of each sector is not equal so that as he pointed out: labour will be employed more in the investment trades, less in the consumption trades; this will increase total employment if elasticity of supply in the investment trades is greater than that in the consumption-goods trades diminish it if vice versa (Hicks 1937: 149). In this model, curiously, it is necessary to note that an increase in the quantity of money, by raising nominal income, will cause an increase in employment. This first model, although Hicks calls it classical, is neither dichotomic nor neutral. This characteristic comes from the fact that it is nominal investment and nominal savings and not real investment and real savings that depend on interest rate. Thus the investment function is not homogeneous of degree one vis-à-vis nominal variables, which, as d Autume remarks translates a generalised money illusion (2000: 421), a characteristic that can be found in each of these models. 17 A Keynesian model opposes the above in that the demand for money depends on interest rate and in that nominal savings, in accordance with the multiplier, depends only on nominal income. Hicks wrote it as follow: M ¼ Lðr Þ; I n ¼ I n ðr Þ; I n ¼ S n ðy n Þ The singularity is that it is the interest rate and not nominal income that is determined by the quantity of money: the interest rate determines nominal investment, which, via the multiplier, determines nominal income. It results in a rise in the inducement to invest, which increases national income without affecting interest rate. Obviously a rise in the quantity of money, by reducing the interest rate, increases nominal investment and employment. Keynes s essential contribution is therefore, according to Hicks, his liquidity preference analysis, because without it the multiplier would have no role. However, Hicks thought the economy described by Keynes corresponds more closely to the following model: M ¼ LðY n ; rþ; I n ¼ I n ðrþ; I n ¼ S n ðy n Þ in which nominal income has been introduced in the function of the demand of money. For Hicks, this modification restricts considerably the opposition between Keynesian theory and classical theory. Indeed, henceforth, a rise in the inducement to invest triggers an increase in nominal income as well as in interest rate, whereas a rise in the quantity of money reduces the interest rate and increases employment. Graphically this result appears clearly. If LL, the curve representing equilibrium of the money market in the plan (r, Y n ) is increasing, a rise in the inducement to invest shifts IS to the right and generates 107

12 Michaël Assous of the first two enabling passage easily from one to the other. He stressed that the opposition between Keynes and the classical authors is neither a conflict between rigidity and flexibility of money wages nor a conflict between unemployment and full employment, but originates in liquidity preference theory. Now compare Hicks model with Kalecki s 1934 model. It is worth noting that the conceptions of the labour market advocated by Hicks and Kalecki are radically different from one another when one considers classical theory. Whereas Hicks assumed that the rate of money wages per head can be taken as given (Hicks 1937: 148), Kalecki supposed on the contrary that the money wage rate decreases with an excess supply of labour. Moreover, while Hicks article lacked an explicit account of how the labour market works and in which state it happens to end up, Kalecki insisted on the idea that for a system to be accepted by classical economists (Kalecki 1990: 201) it must display full-employment equilibrium. As a result, the impact of a rise in the inducement to invest and in the quantity of money differs significantly in Hicks and Kalecki s classical models. Focus, to start with, on the way Hicks and Kalecki respectively envisioned the effects of a rise in the inducement to invest. In his system of two production sectors, Hicks showed that such a shock modifies the structure of production. Thus, because total employment depends on how production is divided between sectors, it will not necessarily remain unchanged. Only if sectoral supply elasticities are identical will there be no change in employment. On this point, Kalecki s classical models are fully at odds with Hicks classical model. Indeed, market clearing and full employment exists in both of Kalecki s classical models. Consequently, an increase in the inducement to invest (i.e. a rightward movement of the schedule of marginal profitability of new investment projects) always elicits a rise in the rate of interest, which results in unchanged total a rise of national income and of the interest rate. It is only if LL is horizontal in the case of the liquidity trap that a rise in the inducement to invest only causes a rise of national income. 18 Last, aiming to show that it is possible to realise a complete synthesis between classical tradition and the Keynesian theory, Hicks built a variant of the latter, where the nominal income and the interest rate are the arguments for the demand functions of money, investment, and savings, the model of generalized General Theory, which he wrote as such: M ¼ LðY n ; rþ I n ¼ I n ðy n ; rþ I n ¼ S n ðy n ; r Þ Thanks to this, Hicks can also show that a rise in the inducement to invest causes an increase in nominal income and in the interest rate, whereas a decrease in the quantity of money reduces the interest rate and raises the nominal income. 108

13 Kalecki s (1934) model employment. 19 In the same way, an exogenous decrease in capitalists consumption will not affect total employment. Indeed, according to Say s law, if saving rises, investment spending rises by the same extent. Thus, whatever the differences of supply elasticity between production sectors are, workers unemployed in the sector of consumption goods are hired in the investment sector. Because, as long as they are still unemployed, money wages will fall, inciting capitalists to increase their spending until full employment is reached. And this result is not modified when the demand for money depends on the interest rate as in Kalecki s System II. With regard to the effects of monetary expansion, the differences between Kalecki s and Hicks analysis also have their roots in the treatment of the labour market. In Hicks model, an increase in the supply of money causes a rise in employment, due to the rigidity of money wages, whereas for Kalecki, money is neutral due to the flexibility of money wages. Indeed, whether it is in his System I, founded on quantity theory, or in his System II, in which nominal income and the interest rate are the two arguments of money demand function, any rise in the supply of money entails only a change in nominal variables. Contrary to Hicks, Kalecki claimed that introducing the interest rate in the money demand function alone is not sufficient to get a system that leads to non-classical conclusions. What is needed is to add a particular conception of the labour market. This paper now turns to the differences between Kalecki s unemployment model and Hicks Keynesian model. In order to build a model with unemployment Kalecki developed a different conception of the labour market from Hicks. The central hypothesis of this conception is that unemployment, as long as it remains unchanged, is not supposed to pressure money wages downwards. However, if money wages do not fall and there is an excess supply of labour, Kalecki did not conclude that wages are completely rigid. On the contrary, he believed that money wages respond to variations in unemployment. Unfortunately, this approach is mentioned but not explained, even if it is highly likely that Kalecki was referring to Marx s theories. Whatever it may be, however, it is clear that Kalecki believed that the labour market, due to the existence of a reserve of unemployed workers being available, is characterized by a gap between supply and demand. This analysis can hereby be distinguished from that of Hicks. For Hicks, on the one hand, money wages are given and on the other hand the supply of labour is not specified, making it difficult to say whether or not unemployment exists (see De Vroey 2000). 19 As real savings do not depend on interest, the distribution of employment between sectors will not be affected. 109

14 Michaël Assous Despite this difference, Kalecki s model with unemployment behaves fundamentally in the same way as Hicks. Concerning the effects of a rise in the inducement to invest and the supply of money, both models react in exactly the same way. The only difference between Kalecki s analysis and Hicks is the existence of a liquidity trap in the latter. Kalecki did not refer to a situation in which the liquidity preference schedule is interest inelastic. Consequently, whereas in Hicks model, a rise in the inducement to invest can trigger a rise in employment without affecting the interest rate, such a shock in Kalecki s model obviously creates a rise in employment and in the interest rate. In his attempt to highlight the differences between classical theory and Keynesian theory, Modigliani also came up with three models but reached radically different conclusions from Hicks. Whereas to Hicks the distinguishing feature is liquidity preference analysis, to Modigliani it is the rigidity in money wages. Although Kalecki adopted a representation of the classical theory that is not very different from Modigliani s, his model including unemployment is different from Modigliani s Keynesian system. Kalecki s 1934 article offers both anticipation of the IS-LM model on the one hand and of the difference between the classical and the Keynesian models on the other. 4. Three Systems vs. the IS-LM model of Modigliani (1944) In his 1944 article, Modigliani reconsidered the difference between Keynesian theory and classical theory. Keynesian theory is now defined by the hypothesis of rigidity of money wages that Hicks considered common to classical and Keynesian models. Henceforth, the opposition between Keynes and classical authors becomes an opposition between rigidity and flexibility of wages and between unemployment and full employment. Modigliani s analysis of the labour market, 20 coupled with two conceptions of the money market, then allows the definition of three models: a crude classical model; a generalized classical model; and a Keynesian model. The specificity of the crude classical model is that the real part of the system, namely, employment, interest rate [emphasis in the original] output, or real income, do not depend on the quantity of money. The quantity of 20 From the idea that in a classical model the workers are rational, Modigliani wrote the supply of labour in a conventional way: N s ¼ F(W/P) or in the inverse form: W ¼ F 71 (N)P. Therefore, by introducing a hypothesis of rigidity of money wages, corresponding for him to the benchmark between classical and Keynesian models, he rewrote this equation as W ¼ W

15 Kalecki s (1934) model money has no other function than to determine the price level (Modigliani 1944: 68). In this model, one does not however find this property in an obvious way. In fact, the quantity of money determines national nominal income and the interest rate. Thus, it is only if one supposes nominal investment and savings to be homogenous of degree one with regard to the price level that this occurs. In 1944, Modigliani curiously did not totally resolve Hicks (1937) problem. In his second model, Modigliani replaced the quantity equation by a function of money demand for which the arguments are nominal income and interest rate. This meant to show that the introduction of the interest rate in the demand function for money is perfectly acceptable in a classical model when money wages are perfectly flexible. Indeed, as long as the supply of labour depends on the level of real wages, the equilibrium reached by the economy is not modified. Once again, this is true only if the functions of nominal investment and nominal savings are homogeneous of degree one in prices. It is worth noting that Modigliani s classical models are characterized by the flexibility of money wages and prices and its ensuing clearance of the labour market; it is also characterized by the ineffectiveness of a monetary expansion in increasing employment and by the failure of an increase in the inducement to invest to reach the same goal. Last, Modigliani elaborated on a model representing the Keynesian theory. He claimed a Keynesian outcome arises when two factors are jointly present: rigidity of money wages and money demand depends on the interest rate and nominal income. Thus, Modigliani argues that the Keynesian model is characterized by a basic maladjustment between the quantity of money and the wage rate, which explains the low level of investment. He expands as follows: What is required to improve the situation is an increase in the quantity of money (and not necessarily in the propensity to invest); then employment will increase in every field of production including investment. (Modigliani 1944: 76 7) The contrast between Kalecki s and Modigliani s approaches can be easily drawn and synthesized in Table 1. As seen, Modigliani s classical models are characterized by the flexibility of money wages and its ensuing clearance of the labour market, on one hand, and by the ineffectiveness of an increase in the inducement to invest in increasing employment on the other. It thus seems these are exactly Kalecki s classical models. Kalecki s models differ from Modigliani s only by distinguishing between two classes (capitalists and workers) and two sectors (consumption and investment 111

16 Michaël Assous Kalecki Hicks Modigliani Table 1 The features of the Kalecki, Hicks and Modigliani models Labour market Demand for money Impact of shocks Flexible money wages in classical models (Systems I and II), resulting in full employment; and flexible money wage in System III, which results in unemployment. Fixed money wages in the classical and Keynesian models, resulting in a non-specified situation in the labour market. Flexible money wages in classical models and rigid money wages in the Keynesian model, resulting respectively in full employment and unemployment. Demand for money independent from interest rate in System I. Demand for money dependent on interest rate and national income in System II and III. Demand for money independent from interest rate in the classical model and dependent on interest rate in Keynesian models. Demand of money independent from interest rate in the crude classical model. Demand for money dependent on national income and interest rate in the amended classical and Keynesian models. Rises in the inducement to invest and in the quantity of money do not affect employment in Systems I and II and entail both a rise in employment and interest rate in System III. Rises in the inducement to invest (when sectoral supply elasticities are not identical) and in the quantity of money affect the level of employment in the classical model but may have no effect on it in the Keynesian model due to the existence of the liquidity trap. Rises in the inducement to invest and in the quantity of money affect employment only in Keynesian model. goods). But, whereas Kalecki s and Modigliani s classical models happen to be so closed, their models with unemployment display some important differences Thus we have : N di ¼ f 0 I W p I ; N dc ¼ f 0 C W W p C ; N d ¼ N di þ N dc ; N d ¼ N d p I ; W p C : 112

17 Kalecki s (1934) model As previously stated, the specificity of Kalecki s unemployment model hinges on his conception of the labour market. No difference between his unemployment model and the classical models would remain were this argument proved to be flawed. Kalecki s unemployment model is, however, at odds with Modigliani s Keynesian model, which rests on an exogenous wage. Indeed, although money wages do not adjust in response to an excess supply of labour, Kalecki argues that they depend on unemployment movements. Money wages are thus endogenous. It is, however, clear that if Kalecki had proceeded to make use of his unemployment model to discuss the effect of an exogenous decrease in money wages, he would have reached Modigliani s conclusion. He would in particular have argued that the only way a decline in wages could increase employment is through its effect in increasing the real quantity of money, hence decreasing the rate of interest and thereby increasing investment and aggregate demand. However, contrary to Modigliani, Kalecki s main interest was not comparative static equilibria. Instead, Kalecki referred to a temporary equilibrium position in the Marshallian sense, a position that would subsequently change as variables that had been held constant would be permitted to change. In the conclusion of his 1934 paper, he indeed noted that if the assumption of a given volume and structure of capital equipment were abandoned, then as a result of changes in capital stock there would be a continual movement through a series of equilibrium or quasi-equilibria until the final equilibrium is attained, i.e. a position in which investment activity no longer changes the volume and structure of capital equipment. Moreover, when the time of construction of investment goods is taken into account, this movement will be cyclical and the position of final equilibrium will never be reached, giving rise to endogenous business fluctuations instead. 22 Thus, Kalecki s unemployment theory should not be interpreted as a static theory of unemployment disequilibrium. More specifically, what concerned Kalecki, according to this interpretation, is not an economy whose level of unemployment remains constant over time, it is instead an economy whose capital stock is continuously varying, entailing unemployment movement that causes wages to vary but in which aggregate demand is not thereby adequately stimulated, so that unemployment fluctuations continue to prevail, although the intensity changes over time. Correspondingly, once it is recognised that Kalecki s unemployment theory is concerned, strictly speaking, with a situation of 22 On this point, Kalecki s 1939 Essays are directly related to Kalecki s 1934 model. For an account of the relationship of Kalecki s 1934 model and Kalecki s 1939 Essays, see Assous (2003). 113

18 Michaël Assous unemployment quasi-equilibrium, it is also understood that the validity of its analysis does not depend on the special assumption of absolutely rigid money wages. 5. Conclusion As the 1934 article proved, before the General Theory appeared, Kalecki had already built a model able to express the main conclusions of the classical theory and to express the persistence of unemployment. In the case of a complete flexibility of prices and wages, he first elaborated a model of full employment founded on Say s law and then, considering the case in which the demand for money depends on the interest rate, showed that the economy reaches an identical equilibrium. In a third model, dedicated to allow for unemployment, he referred to a conception of the labour market for which, as long as unemployment remains unchanged, it does not push down money wages. In this case, movements of employment can be explained in terms of movements in aggregate demand, resulting in Kalecki s famous doctrine, which states that capitalists get what they spend. A formal representation of this argument has made it possible to show that Kalecki did elaborate on an original IS-LM model that differs from the models of Hicks and Modigliani. On the one hand, it seems that Kalecki and Hicks developed a radically different analysis of the classical theory. Contrary to Hicks, Kalecki did not think that the introduction of the interest rate as an argument in the money demand function necessarily cast a shadow on the classical theory, a conclusion Modigliani stressed again ten years later. On the other hand, this comparison has highlighted the fact that Kalecki developed a different model with unemployment from Modigliani s. Whereas in Modigliani s Keynesian model, money wages are exogenous, they are endogenous in Kalecki s model. As a consequence, while Modigliani, in a static comparative framework, attributed unemployment to the rigidity of money wages, Kalecki originally developed, with his concept of quasi-equilibrium, a dynamic theory of unemployment disequilibrium in which unemployment variations are due fundamentally to the fluctuations of investment. However, despite the originality of this model, Kalecki did not find it timely to have his 1934 article translated. To explain this decision, three hypotheses can be suggested. In 1944 Kalecki wrote that the flexibility of prices and money wage could cause distribution effects making fullemployment equilibrium unstable and he thus put implicitly into doubt his 1934 analysis of the classical theory. Moreover, in his 1934 article 114

19 Kalecki s (1934) model Kalecki reasoned in a perfect competition framework, whereas he adopted the hypothesis of imperfect competition when Hicks and Modigliani s articles came out, which drove him to develop a new analysis of the distribution of income. Hence, contrary to what his 1934 article showed, Kalecki insisted on the fact that no negative correlation exists between real wages and employment. Last, as noted previously, Kalecki thought that the adequate frame to his theory was dynamics and not comparative static, a point that he acknowledged to Hicks when criticizing his IS-LM model (Kalecki 1939: 313). Considering certainly that this latter theory filled the gap, Kalecki might have thought it useless to translate this article. References Assous, M. (2003). Kalecki s contribution to the emergence of endogenous business cycle theory: An interpretation of his 1939 essays. History of Economic Ideas, 11: Barens, I. and Caspari, V. (1999). Old views and new perspectives: On re-reading Hicks Mr. Keynes and the Classics. The European Journal of the History of Economic Thought, 6: Chapple, S. (1991). Did Kalecki get there first? The race for the general theory. History of Political Economy, 23: (1995). The Kaleckian origins of the Keynesian model. Oxford Economic Papers, 47(3): Darity, W. and Young, W. (1995). IS-LM. An inquest. History of Political Economy, 27: D Autume, A. (2000). L essor de la macro-économie. In A. Béraud and G. Faccarello (eds), Nouvelle Histoire de la Pensée Economique, tome 3. Paris: La Découverte. De Vroey, M. (2000). IS-LM à la Hicks versus IS-LM à la Modigliani. History of Political Economy, 32: Dos Santos Ferreira, R. (2000). Keynes et le développement de la théorie de l emploi dans une économie monétaire. In A. Béraud and G. Faccarello (eds), Nouvelle Histoire de la Pensée Economique, tome 3. Paris: La Découverte. Hicks, J. (1937). Mr. Keynes and the Classics : A suggested interpretation. Econometrica, 5: Kalecki, M. (1934). Trzy uklady. Ekonomista, 34: Translated in Kalecki (1990: ) as Three Systems. (1939). Essays in the Theory of Economic Fluctuation. London: Allen and Unwin. (1944). Prof. Pigou on The Classical Stationary State. A comment. Economic Journal, 1: (1971). Selected Essays on the Dynamics of the Capitalist Economy Cambridge: Cambridge University Press. (1990). Collected Works of Michal Kalecki. Volume I: Capitalism, Business Cycles, and Full Employment. Oxford: Clarendon Press. Keynes, J.M. (1936) [1973a]. Collected writings of John Maynard Keynes, vol. VII. In D. Moggridge (ed.), The General Theory of Employment, Interest, and Money. London: Macmillan. 115