Growth and Distribution
The Accumulation of Capital (Robinson 1956) was the outcome of a long-held ambition to develop Keynes’s GT into a long-run theory of growth. Robinson had published an early version of her thoughts about Keynes in the long run in 1937 (see Robinson 1937a, b), but the post-war demand for reconstruction of Britain and Europe presented an urgent challenge to theorists as well as policy makers. Robinson’s post-war economic analysis was shaped by the classical concerns with growth and the distribution of income between wages and profits (returned to the analytical domain by Sraffa (1951) in his Introduction to Ricardo’s Principles). She also saw Keynes as having introduced history into the discussion of an analysis of growth with his suggestion of ‘shifting equilibria’, and the possibility of moving on from the static neoclassical method. Her ultimate goal was a theory of growth and distribution which would replace the neoclassical method of comparing equilibrium positions under different conditions with one which traced the process of moving through out-ofequilibrium situations. First, the basic logical requirements had to be made clear, and the classical schema provided a simple framework from which the implications of changing its assumptions and the forms of its relationships, extending to a Kaleckian-Keynesian theory of effective demand, could be explored. Ultimately, such a theory might lead into a description of endogeneities or path dependence, a dynamic system in which accumulation can be discontinuous and can impact on its own determinants.
Robinson developed her own theory drawing on Harrod’s (1948) categories of growth. She had studied the proofs of Harrod’s 1948 volume with Kahn and Sraffa (while on holiday in the Dolomites) and published a review (Robinson 1949) of his book. Harrod’s method of achieving ‘dynamic’ analysis was to focus on relationships between rates of change of selected instrumental variables rather than between their levels. He identified distinct categories of possible growth: gn, the natural rate of growth, reflecting a supply potential and determined by Harrod exogenously, by growth of the labour force and of labour productivity; gw, the warranted rate of growth at which investment decisions led to just the outcomes that were expected so that no adjustments need be made; ge, the expected rate of growth corresponding to the expected rate of profits; and ga, the actual rate of growth. Robinson adapted these categories, each one based on her particular hypotheses relating successive adjustments of rates of profits and accumulation depending on both investors’ expectations of future rates of profits and on productivity growth and real wages growth. Her ‘banana’ diagram illustrated possible outcomes of these. Her starting point was to clarify the conditions for steady growth, and then systematically introduce variations to these to trace their outcomes for distribution and growth. By taking account of the (initially) exogenous determinants of gn, and of the effects of Keynesian animal spirits on effective demand and on gw, she creates a set of ‘Golden Ages’.
A Golden Age exists when there is growth with neutral technical progress,
without any change in the time pattern of production, the competitive mechanism [is] working freely, population growing...at a steady rate and accumulation going on fast enough to supply productive capacity for all available labour.
[In these conditions] the rate of profit tends to be constant and the level of real wages to rise with output per man. There are then no internal contradictions in the system ... Provided that...entrepreneurs have faith in the future and desire to accumulate at the same proportional rate as they have been doing over the past, there is no impediment to prevent them from continuing to do so (Robinson 1956: 99; see also Harcourt and Kerr 2009: 88).
All expectations are realised; gw prevails and is equal to gn. Robinson then proceeds, systematically, to set out the different outcomes of different sets of conditions for variations of this equilibrium Golden Age.
Integral to understanding the process of accumulation is recognising the double-sided relationship between the rate of profits and accumulation. Investment, as well as its role in effective demand, at the same time adds to capacity: it has both ‘the relation between the rate of profit caused by the rate of accumulation and the rate of accumulation which that rate of profit will induce’ (Robinson 1962a: 48; italics in original). Kalecki and also Keynes had observed this interdependency, and Robinson developed it to clarify the responses between accumulation of the capital stock and the distribution of income. Her ‘banana diagram’ (ibid.) illustrates this interdependency in a short-period model, in which the ‘central mechanism...is the desire of firms to accumulate, and [this decision] is influenced by the expected rate of profit’ (ibid.: 47). If, for example, the rate of accumulation is higher than that justified by the associated rate of profits, planned investment will be curtailed and accumulation decline. This creates excess capacity in the sector producing machines (for both its own reproduction and producing wage goods) and investment plans are revised downward in response to lower expected profits. Expected and actual profitability are linked through Keynes’s concept of animal spirits to planned accumulation. Here, Robinson is describing a process, a sequence of responses to risky decisions in changing circumstances.
Her approach was subsequently developed by Donald Harris (1975, 1978) who introduced into Robinson’s theories of accumulation and rate of profits and growth, a production and distribution relationship, between the wage rate and the rate of profits, determined by technical factors and the current state of the class war. Technical changes associated with accumulation will shift the relationship and possibly also affect the wage rate. In terms of production, the historical and cultural elements of the workplace as well as the choice of technique and productivity growth create a potential surplus. The outcome of this relationship between the (given, money) wage rate and the maximum rate of profits consistent with it is then introduced to the relationships between effective demand, or investment, and the distribution of income between wages and profits. The expected rate of profits will be associated with a desired rate of accumulation: an animal spirits function accommodates the more nebulous elements of uncertainty and risk upon which such investment decisions largely rest, and the necessary relationship between saving and investment is effected through a redistribution of income as the margin on the price of wage goods increases and the real wage falls. Harris demonstrated Robinson’s analysis of the potential for crises in the interaction between the rate of accumulation of the capital stock and the distribution of income in historical time.
The process of technological change and choice of technique is an unavoidable part of the investment decision to maintain or expand productive capacity; new investment could incorporate new methods of production. Wilfred Salter, a Cambridge graduate student in the 1950s, described a production process with a changing composition of coexisting machines of different vintages (Salter 1960). He proposed a sequence of investment decisions, whereby the new techniques are introduced only when the existing vintages no longer cover their variable costs and when the latest best-practice machines cover their expected total costs, including the normal rate of profits (assuming competitive conditions).
Harrod had treated labour productivity as given for each long period and gn as exogenous. Robinson argued that each new investment decision could potentially introduce new techniques embodying increases in labour productivity, thereby increasing gn. Robinson’s (1956) analysis was on the basis of comparisons of different outcomes for different Golden Age conditions. At any moment in time, there is an array of methods available involving different combinations of labour and equipment, which are associated with different expected rates of profit at given levels of costs and prices. To maintain employment, real wages must grow with productivity increases and capacity changes; the real wage must also be responsive to surpluses or shortfalls in available labour.
Robinson’s endeavour to introduce the roles and impact of technical change led her into a further critique, of the neoclassical concept of ‘capital’ and of the theory of distribution (Robinson 1953-1954). Her article infuriated the mainstream and started a major controversy[1] between Cambridge (UK) Keynesians on the one hand and Cambridge (MIT-US) Neo-Keynesians and neoclassical economic theory on the other. The neoclassical approach meant that ‘the rate of return on investment itself enters into the calculation of the cost of any particular outfit of capital goods. For this reason an unambiguous measure of the cost of capital goods requires a definite and universally applicable expected rate of profit’ (Robinson 1956: 103). In order to provide a valid explanation of the distribution of income between wages and profits, in terms of marginal products, a neoclassical aggregate production function required a meaning for and a unit of measurement of ‘capital’ which was independent of prices and distribution of income itself. Her own idea of ‘capital’ recognised its different forms and roles as both finance and actual machines, and was also influenced by Marx and Keynes; the significance of capital lay in its relationships to labour. How, then, should ‘capital’ be conceptualised?
In her article (Robinson 1953-1954), she offered a possible way to deal with the measurability of capital, as real capital—capital in terms of labour time—which gave it a meaning within the aggregate production function as a ‘factor of production’ and a measure in wage units. To be a measure required overall economy equilibrium (Robinson’s) of realised expectations, so that the rate of profits is the same as the rate of interest, and no changes are made to decisions involving capital in either form. The resulting (ex ante) ‘pseudo production function’ showed a set of techniques ordered by successively lower positions of net output per unit of labour, and lower real wage rates associated with higher rates of profits. Switch points between techniques occur at a level of real wages at which the rates of profits are the same for adjacent techniques and at the ruling prices; the values of capital required for each technique are in the same ratio to each other as net profit per labour. A backward switch implies a possibility that the same technique, having been the most profitable over a set of values of wages and profits, could be reverted to at different rates of profits. This means that the production function is not a valid way of representing the neoclassical vision of prices as indexes of scarcity. Similarly, capital-reversing, where a less productive, less capital-intensive technique may also be associated with a lower rate of profits, refutes the existence of a neoclassical downward-sloping demand curve for capital.[2] The language used in this discussion around the aggregate production function suggests that a process is taking place when only comparisons of different positions can be made. Increasingly, Robinson saw this ongoing episode of theoretical and often personal and political hostility as being of little consequence and, indeed, a distraction from the development of an adequate theory of capitalist growth.
Moving on from the frustration of the (continuing) capital theory debates, Robinson’s critique of neoclassical theory returned to her earlier criterion of relevance, almost taking for granted her own sufficient refutation of its internal logic and the implications of Sraffas 1960 Production of Commodities by Means of Commodities.
For Robinson, ‘relevance’ was directed at replacing a method that was static with one that was about process, and it indicated the need to consider what happens with the passage of time, how the history of a situation is crucial for understanding how—with its present context of institutions and practices, a mixture of political, social, and economic, of institutions embodying cultural tradition and class structure—the processes of growth and distribution occur. Her analyses of growth theory had concentrated on theories in ‘logical’ time where relevance was directed to assumptions. Now, by the end of the 1960s and through the 1970s, relevance was more about method, and she wanted to address theories in ‘historical’ time.
Logical time could compare different equilibrium outcomes; it was static. Theories in historical time could (aspire to) trace a process of more complex changes which have no connection to ‘equilibrium’. She argued that the historical approach inspires more interesting and pertinent questions because it arises out of causal hypotheses. Introducing such ‘relevance’ means the theory may not be ‘scientific’ if less precisely defined and measurable factors—non- scientific—were to be included in the causal list.
Robinson’s own use of models in logical time was intended as a crucial prologue to her more complex historical theories. Such models enabled the purely deductive outcomes of systematically varying the model’s conditions, to be established prior to introducing ‘real’ features. Their solutions provided sets of simultaneously compatible equilibrium values and so carried no causal status. She constantly cautioned her readers to the limits of such abstract methods and their interpretation (see above, pp. 679-680; also see Robinson 1962a: 87, Conclusion).
Robinson was also highly critical of the associated, neoclassical use of equilibrium, as a concept and as an analytical device. Her own use of ‘equilibrium’ in her Golden Age models referred to a position or rate of growth at which an economy remains if all expectations have been satisfied and there is no reason to change the conditions or (investment) decisions by which it got there. She is careful to make the purely abstract nature of ‘equilibrium’ clear: if the economy is in ‘equilibrium’, then it must have always been there; it must have a composition of means of production perfectly suited to meeting the demands for its output so that stocks do not have to adjust between periods; the present is fully determined by its ‘history’ and to be there means its past must be exactly replicated in its present and its future. Logically, the economy could not move from one quantitative configuration of stock to another without having had a different history. Her claim of the irrelevance of the static method itself included the associated role of ‘equilibrium’ in neoclassical theory as an analytical device, of affirming that static method through the closure conditions it required. For Robinson, there was no role for equilibrium in historical analysis. The neoclassical equilibrium proposed tatonnement as the causal process of reaching this point, which Robinson dismissed. Her critique was not that there were interferences with this process reaching its equilibrium and for which ‘corrections’ could be introduced and considered theoretically; it was that her ‘vision’ of the economy and the concepts and reasoning she ultimately chose to represent it were quite different.
Robinson’s analyses in the 1970s were shifting from a long-period ‘generalisation’ of Keynes in the 1950s and 1960s to accommodating the presence of more short-period factors, familiar to her and in different ways, from Kahn and Kalecki, in which the economy experienced cycles as well as long-period trends. In 1952, she had explored the likelihood of a capitalist economy following a long-period growth trend through cycles of short-period (dis) equilibrium—in the sense that short-period off-trend behaviour caused no changes to be made to expectations and investment. The independence or interdependence and indeed recognition of these as two phases of economic activity have provoked controversy in both empirical and analytical studies, and in the respective claims of theories about trends and cycles. Robinson’s early stand was to treat the two as quite separable, but she became more ambivalent and inclined to follow the Kaleckian-Marxian implications of her existing analysis. The actual structure of Kalecki’s analysis made cyclical models part of the ‘natural order’ of capitalism. On the other hand, Pierangelo Garegnani, who was crucial in developing Sraffa’s classical concerns and method, and who supported Robinson in her rejection of neoclassical equi?librium method, completely took issue with her over the status of the long period in economic methodology and whether there actually was a centre of gravitation towards which the capitalist economy was drawn.
Garegnani argued (as did Dick Goodwin initially) that the long period was the only meaningful context for understanding growth and distribution; there was no relevant, systematic short-period behaviour. Robinson, however, identified the existence of (asymmetric) cycles and initially gave independent status to the theories of the cycle and of the trend, later identifying links between the two processes. Ultimately, she seemed to agree more with Kalecki’s (1968) and Goodwin’s (1953, 1982) conclusion that trend and cycle are ‘indissolubly mixed’. Like Robinson, they based their arguments on a Marxist understanding of contradictory forces in capitalism, demonstrating in an aggregate model the inevitability of cycles disrupting any trend, and the intersectoral constraints imposed in a disaggregated model of production interdependencies; such ‘shocks’ themselves are a product of capitalistic societies, and the trend has no independent existence; it was an ex post statistical artefact.
That both trend and cycle are integral parts of the same process of capitalist dynamics can shift the dilemma of a connection between short and long periods to a prior level of generality. The result is the coexistence of opposing visions in none of which demand and supply, and its equilibrium, play any central role and whose respective, ongoing developments underlie the emergence of separate major schools of thought in Post-Keynesian growth and distribution theory (see Harcourt 2006).
Responses to her general critique of the inherently static nature of mar- ginalist method and theory, and so too to the relationship between trend and cycle, continue, over 30 years later, and include Neo-Keynesian theorists. Paul Samuelson and Robert Solow both sustained a defiant correspondence with Robinson around issues connected, at first, with Keynesian growth theory and then capital theory. Fifty years on, they were still addressing her challenge concerning the static method of neoclassical economics, although with mathematical tools. They both use a method of analysing processes as distinct from identifying differences while remaining within a neoclassical vision of accumulation, growth, and distribution. Solow (2010) specifies separable trend and fluctuation. This changes the vision of capitalism from Robinson’s Kaleckian-Keynesian endogenously generated movements of growth, where growth and cycle processes are accounted for within a common theoretical structure, to one of exogenous and unpredictable ‘shocks’ to the otherwise steady process, a trend, of growth. Solow separates the theory of demand from the main determinants of growth, as explained by an aggregate production function. What is needed, he suggests, is ‘a theory of the evolution of invest?ment, capital stock and level of technology’ (ibid.: 60), a recurring sentiment in the history of economic thought. Hence, a ‘medium-run’ offers a way of linking the (short-period) theory of recession or inflation and the long(er)- period theory of trend. This is not a coherent solution, but it is consistent with Robinson’s 1952 suggestion and also with Garegnani’s fundamental ‘vision’, even though they held opposing theories. More than 30 years after her death, Joan Robinson has critics who still do not, or cannot, answer her arguments.
In 1980, Robinson published a joint paper (Bhaduri and Robinson 1980) in which she sought to place her historical method into a more fundamental framework, and to synthesise, or at least reconcile, those specific theories which she considered robust and relevant to her sense of the inequities of capitalism. The aim was a unifying structure with a classical vision of the circularity of reproduction set out in its schema, whose links between different economic processes could be seen to derive from the more general properties of capitalism.
The authors integrated into one structure a two-sector production analysis which they called ‘Sraffaesque’[3] and a Kaleckian-Keynesian theory of effective demand and capitalist distribution. The analysis of the distribution of the product takes place at the level of class so that the political relations involved need to be consistent with the assumptions about the class nature of expenditure and saving. This is illustrated in Kalecki’s two analyses of profits: the share of gross profit in the product of industry is determined by the level of the gross margin, which involves a discussion of market structure; and the aggregate flow of profits per annum depends upon the aggregate flow of capitalists’ expenditure on investment and consumption. With this classical framework, Robinson can introduce ‘historical’ features into the logically prior analysis of production. This means accommodating implications of ‘time’, for both the conditions and relationships, and the authors describe the assumptions that need to be made if the investment decisions are to be appropriate to the technical requirements of the composition of spending—or the distribution of income. Past expectations of profitability are manifest in the existing stocks of machines and of goods, and these will only be ‘correct’ if expected profitability is realised. As with ‘equilibrium’, the outcomes of different distributions of income, or rates of exploitation, for a uniform rate of profits, can only be analysed in a more abstract model with necessary and restrictive assumptions not applied to any particular case.
This analysis of the actual ‘share of wages in net output (and therefore the potential ratio of profits on capital) depends upon commercial, social and political influences and the fortunes of the class war’ (Bhaduri and Robinson 1980: 111): ‘[T]he workers’ share is governed by the ratio of exploitation, while accumulation...turns the potential share of the capitalists, set by the ratio of exploitation, into the corresponding rate of profit by creating enough effective demand to realise it’ (ibid.: 109). The authors conceptualised the capitalist process in a language (with which Marx may not have been comfortable) of contemporary capitalist forms and abstract methods qualified accordingly. Their approach inspired real and analytical questions and underlies much of the subsequent Post-Keynesian developments.
- [1] In her 1975 article for the Quarterly Journal of Economics entitled ‘The Unimportance of Reswitching’,Robinson states that ‘The story of what is known as the debate over the reswitching of techniques is a sadexample of how controversies arise between contestants who confront the conclusions of their argumentswithout first examining their respective assumptions. How is it possible to have a controversy over apurely logical point?’ (ibid.: 32).
- [2] A comprehensive and pellucid account of the many associated issues and proposed solutions and analyses, of the participating authors’ ingenuity and resilience in response to criticisms of logic, up to the early1970s, is to be found in Harcourt (1972) and more recently discussed in a symposium in the journal of the Union of Radical Political Economy (2014).
- [3] Sraffas work has generated many lines of research. One important direction in Cambridge growth theory looks at the process of structural transformation in response to productivity growth and changingconsumer tastes. Luigi Pasinetti (1993) developed a pure production model of structural change basedaround Sraffas (1960) scheme. His pioneering work has generated its own questions and lines of development, sometimes referred to as the Anglo-Italian School.
