Joan Robinson's first book was an attempt to forge new tools in order to deal with deficiencies in the then prevailing theory of prices, which was essentially Marshall's formalisation by Pigou. This theory has been out of fashion with later general equilibrium theorists, but in the absence of convincing applications and any substantial contact between general equilibrium theory and the real world, it is still found in elementary textbooks and constitutes the bulk of what students learn in microeconomics courses. It is therefore instructive to examine briefly what Joan Robinson deemed to be wrong with that theory.
Under competitive conditions, each firm produces such an output that marginal cost equals price. An implication of this is that firms can never be found to operate below optimum capacity as denoted by the lowest point on their average variable cost curve. Thus, the experience of the slump when most plants were working part-time and manifestly below capacity cannot be squared with this theory.
Let us illustrate this argument by means of a simple diagram (Figure 1). We look at three plants and we draw their average variable cost curves since average variable cost is what they must cover to continue in operation in the short-run.
If price is pA, then all three plants are producing well above their optimum capacity making short-period profit equal to the difference between marginal and average variable cost. If price is pB, then plant C will not cover its operating costs and it will shut down. But plants A and B will still operate at or above optimum capacity. It is evident that no matter how low the price, the plants that do not shut down will always operate at or above optimum capacity. The standard competitive theory then clearly implies that, even in the short-run, under-capacity working is impossible.
But irrelevance and incongruity with observed facts was not the only problem with the competitive model. The logic of the Marshall-Pigou theory had already been criticised by Sraffa (1926). Sraffa had argued that the notions of diminishing returns and long-run supply curves that are universally upward-sloping, which were essential to that theory, could not be logically sustained. Having demonstrated that this theory is ridden with internal contradictions, he suggested, on the grounds of realism, that a new theory should recognise that a firm's size is limited not by increasing costs but rather by 'the difficulty of selling the larger quantity of goods without reducing the price'.
Joan Robinson took up this 'pregnant suggestion' (as she called it) and, in her Economics of Imperfect Competition (1933), she considered firms confronted by downward-sloping demand curves. For this purpose she developed the concept of the 'marginal revenue curve' which, it is not difficult to show, allows the possibility of under-capacity operation with positive short-run profits.
But though imperfect competition was more realistic than perfect competition, it was still not much more than 'a box of tools for the analytical economist'. It was essentially a refinement of the Marshall-Pigou theory that hardly stepped outside the confines of that tradition. Joan Robinson's contribution consciously steered clear of 'the fundamental problems on whose solution depends the validity of the whole supply-and-demand-curve analysis'. Thus, despite the fact that her book together with Chamberlin's Monopolistic Competition (1933) opened up a whole new branch of traditional price theory that has been since well established in textbooks, Joan Robinson's increasing awareness of the fundamental problems underlying the whole supply-and-demand-curve analysis and her preoccupation with Keynesian theory and the problems of the slump caused her to outgrow her book nearly as soon as she had finished it.
What then are the fundamental problems of demand-and-supply-curve analysis? For Joan Robinson the first and foremost problem is undoubtedly that of time. Demand-and-supply-curve analysis is an analogy from physical mechanics. 'For mechanical movements in space, there is no distinction between approaching equilibrium from an arbitrary initial position and a perturbation due to displacement from an equilibrium that has long been established. In economic life, in which decisions are guided by expectations about the future, these two types of movement are totally different' (Robinson, 1974). As expectations about the future are based on past experience, history plays a paramount role in determining the reactions of economic agents. This whole area is a conceptually difficult one and we can do no more here than attempt to sketch the basic problem.
The point is that in order to decide on the quantities that will be demanded or supplied at a particular price, buyers and sellers need to have a notion of normal price or at least an expectation of a normal price-range that will be established before long. Plans to buy and sell are based on such expectations which are therefore logically prior to the plans. Such expectations are themselves based on historical experience and this is one of the ways that history affects future outcomes. The other ways that history affects future events is by determining the technical conditions of production and capacities on the supply side and the consumption habits on the demand side. (The influence of existing capacities has been explained by Marshall who clearly recognised the irreversibility of the long-run supply curve). But let us return to the more fundamental former point. We can see what it entails by considering a non-equilibrium price pA. DA is the quantity demanded and SA the quantity that the suppliers would be willing to sell at this price if this were the equilibrium price (Figure 2).
In the event, only DA can be sold and traditional analysis would argue that as a result of DA<>
The only circumstances in which it is plausible that buyers and sellers would react to a price as if it were equilibrium price, is when the market (or strictly speaking the economy) has not changed for some time and the particular price has been long-ruling. But of course for all prices other than the long-established one it still makes no sense to draw demand and supply functions without taking into account the expectations that have been formed about the normal price. It is evident that demand and supply functions that do take normal price expectations into account are not meant to determine the equilibrium price but to analyse disequilibrium situations. The traditional demand and supply analysis that is supposed to determine equilibrium price is even, in the circumstances of an unchanging market, quite inappropriate. Though it is true that only in these circumstances the traditional analysis will give an answer consistent with the ruling price, it is misleading to pretend that this ruling price has been determined by the intersection of the demand and supply curves. It is only when we assume that the price is given, and unchanging for some time past, that the demand and supply analysis can 'find' it. And apart from the point of intersection, the rest of the curves are still without any meaning and can serve for nothing. In short, it is only when the normal price has been long-established that the illusion of its determination by demand and supply schedules can be sustained.
Traditional theory can thus at best only deal with comparisons of different, unchanging imaginary states rather than analyse the effects of change. A system is in equilibrium only if expectations are such as to warrant it and this is likely only if it has been in equilibrium for some time. It is therefore possible to compare equilibrium positions that are similar in all but one respect, though such exercises in comparative statics cannot support any conclusions as regards the effects of change in any respect. The former does not at all involve time while in the latter time, that is historical experience and expectations about the future, cannot be ignored. Joan Robinson's insistence on the irrelevance of theory that does not take time seriously has two critical implications. First, it constitutes a general critique of common methodological practice in economics and secondly, it is a specific critique of the nature of neoclassical theory. As regards the first, the implication is that all conclusions from comparative statics are suspect. The method of comparative statics is- useful only as a preliminary exercise in the process of constructing and understanding the properties of a model. After this preliminary stage, the model must be capable of incorporating relevant historical and institutional detail if it is to serve for an analysis of historical change. And it is exactly this crucial step that neoclassical theory, in all its forms, consistently fails to make. Demand and supply analysis, both in its Mar-shallian partial and Walrasian general-equilibrium varieties, is constructed in a way that rules out history. Neoclassical theory could only be suitable for comparative statics; uncertainty about the future and the influence of institutional and historical factors require a drastic amendment of the fundamental assumptions of neoclassical theory in a manner that causes the neoclassical edifice to collapse (for a development of this theme, see Robinson, 1971).
It should be clear that the difficulties that the whole of neoclassical analysis has in dealing with time are, for Joan Robinson, much more grave than those associated with 'reswitching' and 'the measurement of capital'. The latter do not challenge the inability of neoclassical theory to handle time but demonstrate that, even within its chosen field of comparative statics, the results of neoclassical theory are inconclusive and, contrary to popular opinion and textbook pronouncements, do not lend themselves to any interesting or useful generalisations. Let us here digress briefly to consider the 'capital debate' of the 1960s in which Joan Robinson was one of the main protagonists. The start of the debate can be dated to Joan Robinson's critique of the neoclassical aggregate production function (1953-4). This was a particularly convenient construction as it clearly and simply established a number of distributional relationships in accord with basic general equilibrium neoclassical theory. In particular, it established that higher capital-labour ratios, which are expected to result in the process of accumulation, are associated with a lower rate of profit or, equivalently, that a higher wage is associated with a more capital-intensive technique. This went hand-in-hand with the notion that the rate of interest is equal to the marginal product of capital which here could be directly derived from the aggregate production function (being its first derivative with respect to capital). The upshot of the capital debate, which lasted well into the 1960s, was a rejection of both these neoclassical views. As regards the second notion, it was shown that it was invalid to determine the rate of interest from the marginal product of capital in the production function, since the value of capital presupposes a knowledge of the interest rate. It follows that the aggregate production function had to be abandoned as a basis on which such a proposition can be made and neoclassical theorists had to retreat to general-equilibrium analysis for a defence of marginal productivity. As regards the first view, the neoclassicals had to finally admit (Samuelson, 1966) that 'reswitching' of techniques is perfectly possible and that there is no monotonic relationship between the rate of interest and capital intensity of technique. It is therefore not the case that more capital-intensive techniques are generally associated with lower interest rates and neoclassical growth theory has thus lost what was perhaps its simplest and most appealing result. The consequence of this is that, even within comparative statics, neoclassical theory has little of interest to say and proves to be generally indefinite and inconclusive.
But of course Joan Robinson goes further than this. She has always insisted that these results are simply exercises in comparative statics and that the ultimate difficulty of neoclassical theory is that it cannot transcend the limits of comparative statics. It is for this reason that she has stated 'the problem of the "measurement of capital" is a minor element in the criticism of the neo-classical doctrines' (Robinson, 1974). It is also for the same reason that she never paid too much attention to 'reswitching' though she had been aware of this possibility since her Accumulation of Capital (Robinson, 1956) (in which she had dubbed it the 'Ruth Cohen curiosum'). Equally, she has insisted on referring to the 'pseudo-production function' to emphasise the artificiality of a construction that involves 'a comparison of timeless equilibrium positions' rather than a tool for the analysis of 'effects to be expected from a change taking place at a particular moment'.
It should be noticed that an analysis of change that takes time and history seriously does not imply for Joan Robinson that calendar time should be introduced into one's modelling. Economic models are 'tools of analysis' or 'engines of thought' and they are not meant to faithfully replicate reality. What is important is that they are not in conflict with historical events and can possibly provide rough guidance to an economic historian rather than that they portray in detail actual historical events. Given this view of economic theory, the fact that calendar time is empirically observable does not mean that it is necessarily the best notion to use in the modelling of change. On the contrary, the notion of'calendar time' is of such a low order of abstraction that it does not fit in with the other theoretical concepts that have been developed by and have, in turn, themselves formed the quite highly abstract mode of economic analysis. Leaving aside the criticism of the neoclassical theory of price, does Joan Robinson espouse an alternative theory of value? There is little doubt that she looks with favour at the classical theory of value as interpreted by Sraffa (1960). This theory explains prices on the basis of exogenous variables that are quite different from those of the neoclassical approach. The data of classical theory are: the level and composition of output; the conditions of reproduction of commodities; the real wage rate (or the profit rate). In contrast, the neoclassical theory is built on the following data: the utility functions of individuals; the existing technology; the initial factor endowments; the distribution of factor endowments (See Eatwell, 1979). On the basis of these two alternative sets of data, relative prices emerge in neoclassical theory as the equilibrium of the opposed forces of demand and supply and in classical theory as the long period position that will be established from the competitive tendency to a uniform rate of profits. But despite Joan Robinson's sympathy for the classical theory, she does not see this as a properly historical theory that overcomes her fundamental criticism of the method of comparative statics. Classical theory, for Joan Robinson, also abstracts from the influence of expectations about the future and the frequent irreversibility of the outcomes of past decisions that dominate any process of change, with the consequence that it too is limited to dealing only with comparisons of stationary states. Her regard for the classical school thus seems to be founded, on the one hand, on their common criticism of and opposition to the dominant neoclassical school and, on the other, on the more positive reason that in the classical school, and especially in Marx, economic theory is not divorced from historical analysis.
Value theory has been traditionally concerned with the determination of long-run prices. But there can also be concern for the formation of prices in certain short-period situations. Of course, there is an infinity of short-runs depending on the degree of adjustment postulated, so that there can be no general theory of short-run prices, but certain short-run conditions seem to be of some relevance for the understanding of real world events. Thus Joan Robinson has been concerned with the analysis of price formation under the short-run conditions of plant, money wages and expectations all being fixed. Following Kalecki, she distinguishes between two types of markets: those in which prices tend to fluctuate depending on the state of demand and supply; and those in which prices bear a relatively stable relationship to the cost of production. The former are common for many types of primary commodities while the latter are as a rule manufactures. In industry, firms tend to set the prices of their products by adding a profit margin, reflecting their market power and the degree of competition they face, on their costs. This is a basically common-sense approach and it has been often found useful in empirical work. In concluding this section, it should be clear that there is no theory of value that fully satisfies Joan Robinson's own criteria of a good theory. Despite her adherence to Kalecki's short-run price theory and her approval of the Classical long-run theory of prices, she notes that 'all important and interesting questions lie in the gap between the two' (Robinson, 1978, p. xx).