Competition and monopoly

Introduction

In this chapter we discuss the basic elements of the neoclassical theory of the firm and competition. We begin with the evolution of the notion of competition as a dynamic process of rivalry of firms in their struggle for dominance and continue with the neoclassical notion of competition as an “end state” and we discuss the different types of returns to scale. Sraffa demonstrated that neither the increasing returns to scale nor the decreasing returns to scale are consistent with the assumption of perfect competition in the determination of the supply curve in the industry. The only assumption which is consistent with perfect competition is the case of constant returns to scale, which however leads to implausible results. Pierro Sraffa in his articles (1925, 1926 and 1930) where he concluded that the way out of this conundrum is to side step perfect competition and adopt in its place the notion of monopolistic or imperfect competition. His suggestion was pursued by economists in Cambridge England (mainly J. Robinson and Richard Kahn) during the 1930s. In the same time period in Cambridge-Massachusetts we had the monopolistic competition revolution (mainly E. Chamberlin, J. Bain). These developments in both Cambridges faced the criticism from the economists of Chicago University. Thus, during the 1930s we had a revolution in microeconomic analysis known as “imperefect competition” which was taking place, at the same time, with the macroeconomic revolution of Keynesian economics.

In this microeconomic revolution economists were divided into two camps. The first comprised the proponents of monopolistic competition, who were arguing that the actual economy was characterized by monopolistic elements that give rise to distortions and who tried to theorize these elements and also correct them by proposing specific antitrust and regulation policies. We shall call these economists, “imperfectionists”. On the second camp there were economists mainly from the Chicago University, who claimed on both methodological and empirical grounds that there is no such a thing as “monopolistic” or “oligopoloistic” competition and that the actual economic life is not in any empirically significant deviation from the ideal model of perfect competition. Naturally, this camp of economists may be called “perfectionists”.[1] In the ensuing debates, the “perfectionists'” view dominated over the “imperfectionist” one. Fierce as it may have been the debate between the economists in the two camps we recognize that, at the end, they both assumed the importance of perfect competition. The imperfectionists used the perfect competition concept as a yardstick to gauge the extent to which real economic life differs from the perfectly competitive state, while the perfectionists argued that there are no significant differences between the actual and the perfectly competitive economy.

It is ironic, that this process of return to perfect competition begun initially as an attempt to escape from perfect competition through the introduction of realistic elements in the economic analysis of the firm. These efforts led to the development of industrial organization, as an entirely new field of economic research, and to regulation policies that regarded the various market forms as deviations from an ideal model of the perfectly competitive economy, which should be the prototype of actual economic life.

Neoclassical Theory and Perfect Competition

The analysis of competition in the neoclassical theory is contained in the model of perfect competition, which describes the ideal conditions that must hold in the market so as to ensure the existence of perfectly competitive behavior from the typical firm and by extension the characterization of the market or industry as competitive or not. The model of perfect competition describes a market form which consists of a large number of small¾relative to the size of the market¾buyers and of a large number of small producers selling a homogeneous commodity. Both buyers and sellers have perfect information on the prices and the costs of each good. Moreover, there is perfect mobility of the factors of production. The result of the above conditions is that the producers and consumers¾because of their large number and small size¾ are incapable of influencing the price of the product. As a consequence, the price of the product becomes a datum, and the behaviour of the firms is completely passive, that is, firms display a price taking behaviour deciding only the optimal quantity that they will produce. The criterion is the maximization of profit, which is achieved, when the selling price of the good is equal to its marginal cost of production.

The intensity of competition is directly proportional to the number of producers and in general the structure of an industry. In this “quantitative notion of competition”, the firm is conceived as the legal entity that hires the services of the factors of production and combines them in order to supply goods in the market. It is important to note that the firm does not own any factors of production; it merely hires the services of the factors of production which are offered by their owners, that is, the individuals. The larger the number of firms that operate in an industry the more vigorous is their competitive behaviour and by extension we have the establishment of a uniform rate of profit across industries. By contrast, the smaller the number of firms the more oligopolistic and monopolistic is the behavior of the firm in the market and the higher the interindustry profit rate differentials.[2] In this non-competitive state of equilibrium, some prices are above the marginal cost and so society as a whole suffers losses from the underproduction and the underutilization of disposable productive resources. In the neoclassical microeconomic theory, if the firm or the industry displays profits above the normal, for a fairly long period of time, these are attributed to imperfections in the operation of the market and thus in the existence of some degree of monopoly.

We say that firms in perfect competition are price takers, but at the level of general equilibrium, we want to determine the prices which change as a result of the action of some firms. The question, however, is if each and every firm is a price taker, then how do prices change? The usual answer is that prices change exogenously; for example, consumers’ preferences change which lead to the increase (or decrease) in demand. In other words, if there is a deficit (or surplus) of the output produced, which is equivalent to saying that all firms face a negatively sloped demand curve meaning that firms in and of themselves cannot increase their price without reducing their market share. In other words, firms in this case operate as if they were in conditions of monopolistic competition. As a consequence, perfect competition exists only in conditions of equilibrium. It is important to stress that perfect competition is a mathematical assumption imposed by neoclassical economics in order to determine equilibrium and not as a market form that arises from historical observation of the way in which firms are organized and compete with each other.

Similar conclusions are drawn from Walras’s conception of attainment of equilibrium through the mediation of the auctioneer. We know that the participants in this model act independently of each other and simply react to the prices announced by the auctioneer, who is supposed to know all the facts. Clearly, if the participants in the Walrasian model act differently then the attainment of equilibrium is problematic. As a consequence, perfect competition is a sine qua non assumption in both Marshallian and Walrasian models of equilibrium. One corollary of the above is that some theories of competition, that were developed in the past, were eventually rejected not for their lack of realism, but precisely because they were out of the analytical framework of neoclassical economics which is oriented towards equilibrium.

In neoclassical economics competition is defined from the way in which technology is being used. More specifically, competition secures that the agents of production (that is firms) will tend to choose the lowest unit cost and price in order to maximize their profits and reduce the market share of their competitors. Thus, competition will combine technology with the behavior of the firms in the market. Unlike classical, neoclassical economists view production not as a process but rather as a result derived from a functional relationship between inputs and outputs. The production functions are assumed to be continuous and differentiable up to the desired degree. The techniques that are used in production are usually assumed as continuous, nevertheless the neoclassical analysis is not affected, if we have fixed input-output coefficients and L-shaped isoquant curves. Thus, the production functions in neoclassical analysis may take on various forms, such as fixed proportions or the direct opposite of it which is that of perfect substitutability between factors. The assumption of substitutability between inputs is represented with the aid of a concave production function. The proportions between inputs are convex for every single combination of inputs. Hence, we have the already known from the previous chapter isoquant curves, according to which a given level of output can be produced by a variety of input combinations. The curves that we derive are convex to the origin as shown in Figure 1. The negative slope of the isoquant curves represents the diminishing marginal rate of substitution of one factor of production from the other. The isoquants cover the positive quantrant, exactly as in the case of indifference curves, with the difference that the isoquants are measurable, that is, they are amenable of absolute, not only relative, measurement.

As in the case of consumer behaviour, where choices are made at the point of tangency of the highest attainable indifference curve to the income constraint, so in the case of production, the producer chooses the combinations of capital and labor to the point where the isoquant is tangent to the isocost curve, that is the curve C=rK+wL, where r and w are the rewards of the services of capital (K) and labor (L) respectively, and C is the total cost of production. By using the different isocost curves we can form the expansion path, that connects all the points of tangency of isoquants and isocost curves and, therefore, represents the optimal technique in use, that is, the technique with the minimal cost of production in the case of the different proportions of inputs.

From the above it becomes clear that the givens of the neoclassical theory, that is, the preferences of individuals, the endowments as well as the technology, when combined, impose a type of competition which cannot be different from perfect competition. Firms, that is, the carriers of choice of technique maximize their profits at the point where the value of the marginal product of each and every factor of production is equal to its price. The issue that we will deal with is the level and the composition of output of a firm as well as the method of production. The analysis of the firm bears many similarities with that of the consumer. For example, the isocost curves correspond to the income constraint and the isoquants to the indifference curves.

There are two major differences, between the pure exchange model and that of production. The first is that individuals and not firms own the available resources (endowment). Firms simply hire the services of the factors of production owned by the individuals and through the production process transform them into commodities. The second difference is that the isoquants, unlike isoutility (indifference) curves, are objective, that is, isoquants depend on the level of technology. And technology is not about a free choice (as in the case of individuals) but rather is imposed upon the firms through competition.

Economies of Scale

The role of the firm in the neoclassical theory of production is that of the organization of production process through the hiring of the services of the means of production (which are owned by individuals) and transform them into goods and services and subsequently sell them in the market. In other words, firms organize a process according to which the demands of individuals for goods and services are transformed to respective supplies of goods and services. Firms are viewed as price takers and do not know a priori the price at which they are going to sell their products. The size of the firm is directly proportional to its market share, and therefore, returns to scale are particularly important in determining the level of production of a firm.

It is worth mentioning that the concept of economies of scale as it develops within the neoclassical theory and especially in Marshall (1890, chs. 9-13) is static, that is, it does not arise over time, but rather at a particular moment in time. More specifically, one estimates the level of output in each increase in inputs and according to the answer, the economies of scale are distinguished to the following three categories:

• Increasing returns to scale arise, when inputs are doubled and output increases by more than double.
• Decreasing returns to scale arise, when inputs are doubled and output increases by less than double.
• Constant returns to scale arise when inputs are doubled and output doubles as well.

It is important to stress that the returns to scale imply a change in inputs and a subsequent change in output. In this sense, in the neoclassical analysis the returns to scale are derived from a unified analysis of cost. This is a quite different derivation of the returns to scale of the classical economists, whose analysis is dynamic, and therefore the variables involved are dated and evolve during time. Thus, the case of increasing returns to scale is described in Smith’s famous exemplar of a pin factory. The difference from the Marshallian and by extension neoclassical analysis is found in that Smith’s economies of scale have a dynamic dimension resulting from the division of labor, which in turn depends on the growth process of the total economy and not on the individual initiatives that are assumed at the level of production units or even at the level of industry. As a result, for the classical economists, economies of scale can only be dynamic and particularly in Smith economies of scale in industry are only increasing.

Decreasing returns to scale in the classical analysis are associated with the theory of rent. For example, Ricardo refers to the law of diminishing productivity of land, a law which is the result of the rising population and the subsequent rising demand for food that forces the cultivation of less productive parcels of land leading to a rising average cost of production. Diminishing returns to scale according to Ricardo are counteracted in part by the technological progress; nevertheless, in the long run the rise in population offsets the technological progress with the net result of the diminishing returns on land. If, however, one does not account for the technological progress and accounts only for the increase in population then we end up with diminishing returns in production, but this result is in deviation to Ricardo’s dynamic analysis. Furthermore, within the static analysis the assumption of diminishing returns to scale is questionable for it presupposes that one of the factors of production is fixed. In fact, when we double the inputs, it is always possible to repeat the production process with the optimal use of resources without reducing the output produced. Consequently, when we refer to diminishing returns to scale, we essentially presuppose that one of the factors of production remains fixed, and therefore as the other factors increase the proportions of inputs that are used differ from the optimal. The question that comes to the fore is; why should firms produce at a range of output associated with diminishing returns when the can produce at the optimal level of output associated with constant returns to scale. In other words, there is no motive what so ever for a firm to move away from the minimum cost of production associated with constant returns to scale and produce at a range of output associated with a higher cost of production and decreasing returns to scale.

Sraffa (1925) pointed out that increasing or decreasing returns to scale in the classical analysis are derived from quite different economic phenomena. Increasing returns, for example, are derived from the process of accumulation and technological change, associated with the division of labour and the extension of the market. Decreasing returns were derived from the limited availability of land, and were an important component of the theory of income distribution, being the foundation of the theory of rent.

The case of constant returns to scale is quite reasonable and is found quite frequently in economic analysis; for example, it is adopted by classical economists and Marx. Marshall on the other hand while he accepts whenever there is pressure on the raw materials that are being used in industry there is a tendency for rising prices, nevertheless he observes that because the cost of raw materials is only a small fraction of total cost it then follows that they cannot in and of themselves affect the scale of production. Walras in the first edition of his book (1874) also assumed fixed input coefficients and constant returns to scale. In the second edition of his book (1877) he allowed for more substitutability between inputs. Finally, the empirical research has shown that at least in manufacturing the average cost curves have a wide range of output associated with constant returns to scale.

Clearly, Marshall was worried about the case of increasing returns to scale as an assumption that does not fit to the neoclassical static paradigm and this is the main reason that he distinguishes between the economies of scale that are internal to the firm and to those internal to the industry and external to the firm.

Cost Curves

We know from introductory microeconomics that the cost curves of a firm are derived from the production function and the expansion curve (Figure 1b). In the beginning the firm is producing at the falling cost part of the usual U-shaped average cost curve. The shape of these cost curves has to do with the average fixed cost which is supposed to follow a rectangular hyperbola shape which when added to the average variable cost gives rise to the typical U-shaped average cost curves. If we furthermore suppose perfect competition the profit maximizing firm for the particularly given price selects the output at the point where P = MC and in the long-run at the point where P = d= AR = MR = MC = minAC (see Figure 2), where d is the demand curve faced by the firm, and the other notation is usual.

In the short run we may have P > P*, which means that firms in the industry make excess profits. The result is that firms from other industries are attracted and as the number of firms increases the supply increases and the price of the product falls. If, on the other hand, P

It is important to note that the AC curve has the same shape in both the short run and the long run (Figure 3).[3] In the short run, the average cost curve of the firm is drawn under the assumption of a fixed production capacity. In the long run
the firm has the capacity to change the initial proportions between the factors of production in an effort to achieve their optimal combination. We define the long run average cost of a firm from the points of equilibrium achieved by the firm for different levels of output. We realize that the points of tangency are not the minimum points of the short run average cost (SAC) curves and this can be contemplated theoretically by recalling that the SAC are constructed under the assumption of no optimum use of the available inputs at each output level. In the long run, however, this optimal combination is achieved for the given output. Point E is the minimum cost, which nevertheless is the highest from this which is achieved in the long run if all the productive factors are used optimally. Hence, we have the well known envelope curve which is attributed to Viner (1931), that is, the long run average cost curve (LAC) is a frontier or an envelope for the short run cost curves. The LAC curve owes its shape to the succession of increasing returns to scale, to the point of constant returns to scale, (corresponding to the optimal firm size) and past this point, to diminishing returns to scale. The plausible question is why this optimal size is not reproduced as the scale of production increases, given that in the long run there is no fixed cost to prevent this from happening. The usual answer is that there are diminishing returns to the entrepreneurship, each firm is run by a president and as the size of the firm increases it becomes more and more difficult for the same person to run effectively the firm.

Let us refer to the long run position of the economy where point ? indicates the optimal combination of all inputs. The size of the firm is determined from the minimum point of the average cost curve which is associated with a given level of production. We claim that the supply curve of the industry is the sum of the supply curves of the firms that form the industry. In other words, the supply curve of the industry is equal to the sum of the marginal cost curves of the firms for levels of output past the minimum point of the average cost curve. A precondition of the above is that we know the exact position of equilibrium of each and every firm, which is characterized as a relation between increasing and decreasing returns to scale.

John Clapham, an economic historian at Cambridge, found the discussion on economies of scale less than satisfactory for he thought there is distance between the theoretical discussion and the economic reality. His article of “empty economic boxes” impressed the economists of the time, because he pointed out the distance that separates Marshall’s theoretical discussion on the economies of scale and the well known shape of the average cost curve and the difficulties of economists in using these ideas in empirical research. More specifically, he argued that we cannot know what percentage of the performance of a firm is attributed to the economies of scale and what percentage to innovations (Clapham, 1922, p. 129). Simply put, Clapham essentially claimed that economists could not ascertain the type of economies to scale. For this reason he characterized the economic theories that could not be demonstrated empirically as “empty economic boxes”. Since we cannot discern the type of economies of scale and thus their characterization is an extremely difficult or even an impossible task, then, following this theoretical deficiency, some plausible questions follow; as for example, what kind of measures should governments follow in designing their policies with respect to taxation or the provision of subsidies and incentives in general as components of an economic policy.

In the ensuing debates, it was argued that the incongruence between Marshall’s theory of variable returns to scale and empirical observation is solely attributable to the undeveloped nature of statistical analysis and not to any weakness of the theory. We could say that this is the usual response that one gets by applying an empirical critique, which in and of itself could not overturn or create a significant theory. Empirical critique, as it repeatedly has been pointed out, can, at best, ascertain correlations between the variables and not verify causal relations, that is, it cannot derive theoretical relationships between the variables at hand. This does not mean that the empirical critique is redundant. On the contrary, the empirical critique may enhance our understanding of the underlying relationships between the variables and to reveal relationships hitherto unknown.

Sraffa’s Critique of the Marshallian Theory of the Firm

Sraffa’s criticism focused on Marshall’s hypothesis of returns to scale in production and the assumptions of the competitive firm. The assumption of increasing returns to scale for a large range of output implies that the average cost curve of the firm displays negative slope over a large part of its range and that the marginal cost curve lies always beneath it. Two are the reasons for the decreasing average cost; the first is related to the average fixed cost of the firm which, naturally, as the output expands decreases asymptotically, and thereby, since average fixed cost is a part of average total cost, the total average cost curve tend towards a negative shape. The second reason has to do with the more efficient use of the resources. Between the two reasons only the second is associated with a diminishing marginal cost, whereas the first reason leaves the marginal cost unaffected. With this description of the cost structure, if we assume the case of increasing returns to scale, which are internal for the perfectly competitive firm, then there will be a continuous pressure on the (perfectly) competitive firm to expand its size until its absolute dominance in the market.[4]

In particular, Sraffa argued that in the case of increasing returns to scale, which are internal to the firm, there would be a continuous motive by the firm to expand its production until it can supply the whole market. Clearly, such a hypothesis of returns to scale prima facie contradicts the notion of perfect competition for it leads to monopoly. Marshall had also noticed this inconsistency, for example, the case of increasing returns internal to the firm that lead to monopoly was detailed by Marshall (1920, p. 666, n. 3) who credited this idea to Cournot and as an act of intellectually honesty, Marshall characterized the increasing returns case as “Cournot’s dilemma” (Marshall, 1920, p. 380, n. 1). This is the reason why Sraffa pointed out that the case of increasing returns to scale “was entirely abandoned, as it was seen to be incompatible with competitive conditions” (Sraffa, 1926, pp. 537-8).[5] The only case of increasing economies of scale which is consistent with the requirements of perfect competition is when these economies of scale are external to the firm and internal to the industry, a case, however, which is rarely met in real economies (Sraffa, 1926, p. 540). Furthermore, this type of returns to scale cannot be limited to a single industry, and sooner or later its effects are diffused throughout the economy. The problem in this case is that the Marshallian partial equilibrium framework is inadequate to deal with the complexities emanating from the subsequent development of strong interactions between industries (Sraffa, 1926, pp. 538-9).

The same is true a fortiori with the economies of scale which are external to the firm and to the industry, since the interactions across industries are expected to be much stronger and, therefore, reinforcing the case for abandoning the analysis of partial equilibrium. Turning to the diminishing returns to scale and perfect competition, it follows that since firms buy their inputs in competitive markets they face no restrictions whatsoever in the quantities that they buy and, therefore, there is no reason for the increasing part of the usual U-shaped average cost curves. Hence, the structure of the theory of perfect competition does not allow for the case of increasing cost, as the scale of production increases, simply because there is no mechanism to force firms to abandon the minimum cost of production and move to higher cost of production.

Consequently, the only assumption that remains is that of constant returns to scale, which give rise to the constant part of the average cost curves (Sraffa, 1926, p. 540). Thus, Sraffa through a critique of the Marshallian theory of the firm was led to a description of the average cost (graphically presented as a line parallel to the horizontal axis) similar to that of the classical economists. This is the reason why he notes:

In normal cases the cost of production of commodities produced competitively […] must be regarded as constant in respect of small variations in the quantity produced. And so, as a simple way of approaching the problem of competitive value, the old and now obsolete theory which makes it dependent on the cost of production alone appears to hold its ground as the best available (Sraffa, 1926, pp. 540-1).

Hence, Sraffa endorses the theory of value of classical economists, where the price is determined by the cost of production, and not by the intersection of demand and supply curves. More specifically, in the case of perfect competition since the average and marginal cost curves will be identical to each other and since, in equilibrium, the given price (the demand curve) will coincide with the marginal cost (or supply) curve, it follows that equilibrium is not determined uniquely and so the size of the firm is indeterminate.

There are two alternatives out of this conundrum; first, abandon partial equilibrium analysis and adopt the general equilibrium; second, abandon the perfect competition model and adopt monopolistic competition. The first alternative is the best but it is extremely difficult to pursue in any satisfactory way

[T]he conditions of simultaneous equilibrium in numerous industries: a well-known conception, whose complexity, however, prevents it from bearing fruit, at least in the present state of our knowledge, which does not permit of even much simpler schemata being applied to the study of real conditions. (Sraffa, 1926, p. 542)

Sraffa concluded that the second alternative that is the imperfectly (or monopolistic) competition model might offer a simple and, at the same time, viable solution. In this second one while maintains the partial equilibrium framework and the large number of participants with the difference that their product is differentiated, at least, in the eyes of consumers (Sraffa, 1926, p.542).

Consumers’ preferences do not easily change, because they are determined by factors, such as the marketing of the product, the personal acquaintance and the loyalty of customers to a specific firm that last for long. Thus, he proposed the replacement of the assumption of perfect competition by that of monopoly:

It is necessary, therefore, to abandon the path of free competition and turn in the opposite direction, namely, towards monopoly (Sraffa, 1926, p. 542).

In short, the theory of firm cannot be built on the assumption of perfect competition, because in actual competition firms cannot sell any quantity they produce at a given price. The production is not limited by cost, but rather by demand.

The initial reaction of neoclassical economists was to assume certain fixed characteristics in the operation of the firm that give rise to diminishing returns to scale. Thus, they argued that entrepreneurship is a characteristic which does not increase with the size of the firm and so there will be diminishing returns to this factor of production.[6] The logical consequence of this argument according to Kaldor is that we are led to the idea that the optimal size of the firm is determined by the working time of the entrepreneur, in other words we have one entrepreneur firms. Another way to address Sraffa’s critique was to assume general equilibrium where entrepreneurial talents not only are unequally distributed in the economy but moreover there is a fixed supply of them which is equivalent to saying that there are diminishing returns to this factor of production. For this case Kaldor’s (1981) counterargument was that the entrepreneurial abilities are required only in the initial stage of productive activity of the firm. Once general equilibrium is achieved then there is no longer need for the entrepreneurial abilities because simply the optimal production process is repeated from the less talented businessmen. Consequently, the entrepreneur with special talents is needed only in the case where the firm is out of equilibrium. From the moment that equilibrium is achieved then there is no role for the entrepreneur because past a point his abilities are transmitted to the lower echelon of the firm. Clearly, these efforts on the part of the neoclassical economists to save the Marshallian theory of the firm were not convincing.

Another effort to rescue the neoclassical theory of the firm was undertaken six decades later by Samuelson (1990). His argument was based on the idea that once we assume general equilibrium and perfect competition some resources are fixed and so the increase in production of a good may imply the decrease of production of the other good and so we are led to diminishing returns (Samuelson, 1990, p. 269). The trouble with this view however is that Sraffa’s analysis, is focused on the level of industry and criticizes the method with which one may construct the supply curve of each industry assuming perfect competition (?atwell, 1990, p. 281). Thus, general equilibrium is out of the scope of Sraffa’s analysis.

Model Differentiation: Robinson vs. Chamberlin[7]

Up until now we showed that Sraffa’s critique was about the various types of returns to scale and the assumptions of the perfectly competitive firm. Sraffa’s contribution was not so much about the increasing returns to scale, which are internal to the firm, but rather about the strongest cases of diminishing and constant returns to scale. As for the diminishing returns to scale, he argued that they were only possible if the firm drifted further away from the optimal combination of resources and there was no particular reason in a perfectly competitive environment for firms to abandon such an optimal position, i.e., to move away from the minimum cost to a higher cost of production, unless we assume a fixed factor of production, an assumption which is inconsistent with the notion of perfect competition and also partial equilibrium analysis.[8] Consequently, only the case of constant returns to scale was found to be “consistent” with the requirements of perfect competition and partial equilibrium analysis. In this case, however, the marginal cost curve would coincide with the average cost curve and so for a given price, or what amounts to the same thing, a horizontal demand curve, it is impossible to determine the precise size of the firm and its supply decisions. Furthermore, consistent, results may be plausible in the case of increasing returns to scale internal (or external) to the industry and external (external) to the firm, two cases which are rarely met in reality. In such unlikely situations, however, Sraffa argued that the partial equilibrium framework is inadequate to capture the possible complexities that are being developed and the general (not the partial) equilibrium analysis becomes appropriate to deal with the strong interactions that are expected to be developed between industries.

Sraffa concluded that a simple and, at the same time, viable solution to the logical inconsistencies of the perfectly competitive model in the case of increasing returns to scale might be the development of the imperfectly (or monopolistic) competition model. The idea is that in this model one maintains the hypothesis of the large number of firms together with the partial equilibrium analysis and the difference from perfect competition is that the product is differentiated, at least, in the eyes of consumers. In short, the theory of the firm cannot be built on the assumption of perfect competition, because in actual competition firms cannot sell any quantity they produce at a given price. In the real world, production is not limited by cost, but rather by the downward-sloping demand curve.

Sraffa’s suggestion to abandon perfect competition inspired the development of imperfect competition in Cambridge UK by Joan Robinson, who misses no opportunity to admit her intellectual debt to Sraffa’s contribution. Chamberlin, on the other hand, claims that he was actually the first that formulated the analysis of monopolistic competition in his doctoral thesis that he defended in 1927 and published six years later (Chamberlin, [1933], 1962, p. 5, n.4).[9] Consequently, Chamberlin contends that his analysis not only was independent from Robinson’s but moreover had no connection whatsoever to Sraffa’s 1925 and 1926 articles and the pertinent literature. Furthermore, he claimed that his conceptualization of monopolistic competition and the associated with it idea of product differentiation stems from Frank William Taussig (1859-1940) in his debate with Arthur Cecil Pigou (1877-1959) over railway rates differentiation.[10] As a result, he argues that the sources of his inspiration are quite different from those of Robinson and so people mistakenly identify his “monopolistic competition” with Robinson’s “imperfect competition”. He concedes though that this identification has been so much established in the literature that perhaps it is futile to make any effort to change it (Chamberlin, 1982, p. vii). The truth however is that Sraffa had published the essential points of his 1926 article a year earlier in Italian, and that his ideas had been around for some years. Sraffa’s publication in 1926 became possible after the advice and encouragement of Edgeworth, who had read the 1925 article in Italian and certainly was in close contact with Allyn Young (1876-1929), the supervisor of Chamberlin’s thesis at Harvard. Young, on his part, was well informed about the ideas circulating on both sides of the Atlantic and besides Edgeworth he was also in contact with many of the other renowned economists of the London School of Economics, where he taught during the short period 1927-1928. Kaldor mentions that Young succeeded Cannan in the London School of Economics and taught there until his sudden death in March 1929. Thus, Chamberlin may be right, when he claims that his contribution was independent; however, we are allowed to speculate that his ideas were not independent at all of the intellectual milieu on both sides of the Atlantic, although he was not fully aware of the details of these developments in economic theory (Kaldor, 1980, p. xii).[11] It is interesting to note that Chamberlin acknowledges his intellectual debt to Young as “he encouraged with a lively interest in the project as it developed”.[12] Furthermore, it has been repeatedly ascertained that great discoveries in the history of sciences may occur at approximately the same time. After all, scientific research is not carried out by a single researcher in total isolation, but many people in various places in the world may grapple with the same questions and, therefore, it may come as no surprise that ideas disseminate among researchers; notwithstanding, that they themselves may not know the ramifications and exact routes of these ideas. In this sense, we fully share Samuelson’s ( [1967], 1986) view, which deserves to be quoted in full:

Although we have abundant evidence, after 1933 as well as before, that Edward Chamberlin was a lone-wolf scholar with infinite capacity for formulating and pushing a problem to solution in his own way, still, no man is an island unto himself. If A has any sort of communication with B who has any communication with C, […], there is no way to rule out mutual interaction between A and Z even if they have never met or had any direct contact. (Samuelson, Collected Papers III, [1967], 1986, p. 19)

Robinson’s version of imperfect competition unquestionably was created as a solution to the conundrum propounded by Sraffa. In fact, we know that Richard Kahn in his dissertation in 1932 had already developed some ideas on monopolistic competition that Sraffa had sketched out in his 1926 article. In the same time period Robinson managed to integrate some of Kahn’s arguments with regard to the demand side of the market with the cost analysis of the time to a single theory of imperfect competition.[13] More specifically, in this analysis rising costs were excluded by the formal conditions of perfectly competitive firms and given the partial equilibrium setting, the only viable and immediate solution was a downward-sloping demand curve for the industry and the firms within the industry. Robinson’s analysis of the imperfectly competitive firm was carried out on strict neoclassical principles, inasmuch as she used the exact same tools of the perfectly competitive firm. Consequently, her approach was an extension and further elaboration of Marshall’s Principles and the neoclassical tradition in general. Robinson advanced her analysis to new areas of inquiry and to new issues such as the price discriminating monopoly that constitutes, even today, a standard topic in the economics of industrial organization and the subsequent antitrust legislation.[14] She also arrived at radical conclusions regarding the presence of excess profits and capacity, and she developed the notion of labour exploitation based on application of the principles of marginal analysis. Her blunt marginal approach and the clarity with which she presented her views, soon had established her book as the basic text of microeconomic analysis for many decades not only in England but also in the USA. There is no doubt that Robinson has a theoretical starting point absolutely loyal to the Marshallian tradition and that her conclusions follow directly from a strict application of marginal analysis. More specifically, Robinson makes a clear distinction between industry and firm; thereby, couching her analysis in a partial equilibrium framework. Furthermore, she brings to the fore the industry demand curve and the associated with it marginal revenue curve. In fact, Robinson resurrected the marginal revenue and the marginal cost concepts that were laid dormant since the time of the French engineers (mainly Antoine-Augustine Cournot and Jules Dupuit). We know that Marshall used the total revenue and cost curves and his analysis was often vague and pedagogically difficult to follow. All these changed with Robinson’s contributions that explicated the exact relationship between the average and marginal magnitudes and defined the point of optimisation by the intersection of the MR and MC curves. Her models became part of the standard microeconomic apparatus and are reproduced in modern microeconomic textbooks. In what follows (Figure 4 below) we present, for comparison purposes, her model of imperfect competition:

In the left hand side graph of Figure 4, in the short-run, the downward-sloping demand curve and the U-shaped average cost curve are put together along with their respective marginal curves and determine the monopolistic equilibrium output (Qm) and through the demand curve the respective equilibrium price (Pm).[15] In this case, we have excess profits equal to the shaded rectangular area shown on the left hand side of Figure 4. In the long-run, the inflow of firms attracted by excess profits reduces the demand curve for each individual firm to the point that it becomes tangent to the AC curve and at the same time the new MR intersects the MC curve determining the long-run equilibrium pair of quantity (Q*) and subsequently the equilibrium price (P*). In this long-run equilibrium, we have P*=AC>MC and output produced (Q*) falls short of the minimum AC output (Qc) and so there is excess capacity; moreover, since P*>MC, there is loss in consumer welfare.

Turning now to Chamberlin’s notion of “monopolistic competition”, we know that he did not take issue, at least explicitly, with Marshall’s laws of return, but his theoretical model grew out from the effort to infuse realism to the established theory of perfect competition. During the period that the theory of monopolistic competition was emerging there was a tremendous development in the analysis of the firm and a variety of other characterizations of its behaviour, which can be best, called “pragmatic”. In other words, in the interwar period there was an enormous upswing in pragmatism. Economists tried to incorporate into the analytical framework of the firm more practical ideas, which seemed to have been derived from the empirical characteristics of the nature of the firm, as they could be seen operating in the market. Chamberlin, in his persistent effort to inject realism into his model, did not strictly use marginal analysis (e.g., [1933], 1962, pp. 191-3). In fact, Chamberlin objected to the determination of price via the equation of marginal revenue (“a joke”, as he characterized it) and marginal cost because in this way one determines first the equilibrium quantity and then the price, something that for him was unrealistic, as he characteristically notes:

A major deficiency in the marginal revenue technique is that it does not by itself reveal the price. This means that the discussion of equilibrium takes place primarily in terms of output; the category so neatly determined by the intersection of the two marginal curves, instead of in terms of price, the category with reference to which business decisions are most usually taken. (Chamberlin, [1952], 1982, p. 275)

Figuratively speaking his typical analysis, where firms do not take into account the behaviour of competitors is depicted, for the sake of simplicity, in a set of two graphs displayed in Figure 5, where the left hand side graph represents the short run case, where the downward-sloping demand curve (D) is put together with the average cost curve (AC) and the monopolistically competitive firm charges a price through a trial and error procedure (Chamberlin, [1933], 1962, pp. 83-84) so as to secure a desired (maximum) amount of profits which is measured by the shaded rectangular area.[16] The excess profits, however, attract other similar firms and “since the total purchases must now be distributed among a larger number of sellers” (Chamberlin, [1933], 1962, p. 84) the demand curve of each individual firm shifts inwards. The process continues until the demand curve becomes tangent to the average cost curve.

Chamberlin ([1933], 1962, p. 84) is reluctant to admit the similarity of his configuration to that of Robinson’s. It is abundantly clear though that once we have the average curves the marginal ones are implicit and their equality determines the points of optimum decisions. These optimum points may differ somewhat in the short-run and perhaps Chamberlin is right to be a bit cautious about the maximization of profits because the short-runs are fraught with uncertainties and by definition signify disequilibria situations; however, the long- run positions are identical in both Chamberlin and Robinson. This has also been pointed out by Shackle’s attentive observation:

Equilibrium of the firm is represented in Ms Robinson’s language by the output at which the marginal cost curve cuts the marginal revenue curve from below: in professor’s Chamberlin language, by the output at which the average cost curve has the same slope as the average revenue curve and does not lie above it […]. Equilibrium of the group (the “industry”) is represented in both languages by the tendency, for every firm; the average revenue and average cost curves […] the equality of two functions of output and also equality of the first derivatives. (Shackle, 1967, p. 63)

So the difference perhaps lies in the short-run, but on closer examination one discovers that Chamberlin’s “trial and error” process realistic as it might be, if profits are to be maximized then the respective pair of price and output must be exactly the same to that determined by the intersection of MR and MC curves. The lack of these two curves in Chamberlin’s analysis is what made his book less appealing to economists and the presence of these curves in Robinson’s book made her ideas of optimisation much more accessible as teaching material. In fact, not only in economics, but also in other sciences optimality is obtained once the system is set in its marginal conditions. Nevertheless, Chamberlin ([1952], 1982) time again insists on the minor importance of the marginal revenue in the determination of prices:

[…] my own book arose, not out of the marginal revenue curve, but out of the attempt to combine the two theories of monopoly and of competition into a single one which would come closer to explaining the real world, where, it seemed, the two forces were mingled in various ways and degrees. This idea does not appear in Mrs. Robinson’s Imperfect Competition. […] In my own attempt to blend monopoly and competition, the marginal revenue curve was discovered at an early stage and seen for what it is – a piece of pure technique unrelated to the central problem. (Chamberlin, [1952], 1982, p. 274)

Hence, Chamberlin essentially makes an effort to get too much credit for the work and accumulation of knowledge about monopolistic competition in the decade of 1920s, if not a century earlier. Furthermore, by downplaying Robinson’s “imperfect competition” he was essentially downplaying the importance of economists at Cambridge UK and their contributions to the microeconomic revolution. The truth is that Robinson with the term “imperfect competition” did not merely want to fill some gaps in the “intermediate zone” between pure monopoly and perfect competition, but rather she sought to underscore that the neoclassical theory of competition leads inescapably to conclusions completely opposite to those that it would like to derive. In other words, imperfect competition equilibrium is associated with excess capacity and also loss in consumers’ welfare, since the equilibrium price exceeds the marginal cost. Furthermore, the models of “price discrimination” and “exploitation of labour” arising when the marginal (revenue) product of labour exceeds the marginal resource cost were the logical consequences of the neoclassical conceptualisation of competition and marginal productivity theory of income distribution, respectively.

The next step for Chamberlin in his quest for pragmatism and also differentiation from Robinson was his idea of two demand curves. Let us suppose that all firms in the group as well as those that may enter in the group have the same cost functions. The assumption that Chamberlin makes is that the individual demand curve is much more elastic that the demand curve of all the firms that comprise the group. Hence, the individual demand curve d conveys the idea that a firm assumes that the other firms do not match its price reductions. By contrast, the demand curve D represents the share of the market curve which is drawn for the individual firm assuming that all other firms of the group match the price changes. The elasticity and exact location of the demand curve for the group depends, ceteris paribus, on the number of firms that comprise the group. In terms of Figure 6, let us start with the left hand side graph, where the number of firms is supposed to be fixed, that is, there is no entry or exit of firms.

Let us further suppose that the price set is Pm and each firm makes excess profits equal to the shaded rectangular area. This price, however, holds only in short-run equilibrium and each firm will have an incentive to lower its price by moving along the d curve and assuming that the other firms do not follow suit. But each firm has exactly the same incentives, which is equivalent to saying that all firms in the group cut their prices. As a result, the d curve will be sliding down along the D curve to the point that the d curve becomes tangent to the AC curve and also intersects with the D curve, and therefore, any incentive to lower prices is eliminated.

In the right hand side graph, we allow for the number of firms to vary and starting with a position of excess profits, as in the previous case, it follows that there is entry of firms to reap these excess profits and as the number of firms in the group increases it follows that the D curve becomes steeper to the point R of its tangency with the AC curve. Point R, however, is unstable, because there will always be an incentive for each individual firm to lower its prices assuming that the others do not follow suit and, once again, the d curve will be sliding down along the D curve until it becomes tangent to the AC curve. However, this tangency point is not yet an equilibrium proper, because at the price P* the majority of firms makes losses and so they start leaving the group and in doing so the D? curve rotates to the right and in the limit it passes through the tangency point of the d? and average cost curves. This is a stable equilibrium point attained in a more complex way than before (Chamberlin, [1933], 1962, p. 93).

In evaluating these models one wonders how is it possible for a firm to assume consistently that its competitors will not react to a probable price change. The idea to incorporate into the analysis the reaction of other firms is a step forward in the microeconomic analysis, but to assume that firms follow a strategy that is falsified consistently is perhaps worse than assuming the independence in actions of the participating agents. And in this sense, Chamberlin did not really advance the analysis much beyond the well-known models of Cournot, Bertrand and Edgeworth. However, Chamberlin’s idea of the two demand curves created an entire literature about “discontinuity” in the marginal revenue curve that leads to price rigidities in the oligopolistic markets, whereby prices are determined by demand and supply (average cost). This approach made a lot of sense in the 1930s, because it was explaining price rigidities that called forth for government intervention and also labour unions could demand higher wages without causing inflation. The idea was that the discontinuity in the MR curve allowed even substantial increases in the MC curve without affecting prices in any significant way.[17]

Apart from these differences, that is, the inclusion of strategy in the behaviour of individual firms and the two demand curves, the two economists (regardless, of what Chamberlin claims) use more or less the tools of marginal analysis, Robinson more explicitly than Chamberlin. The marginal analysis is what made Robinson’s book widely accessible and established it as a textbook in microeconomics, whereby the lack of explicit marginal analysis is what made Chamberlin’s book confusing and difficult although his insights about the demand side of the market were richer than those of Robinson. The MR is a concept that far from being a “joke” was also essential in Chamberlin’s analysis.

The Rise and Fall of a Revolution

One of the surprising results of the analysis of monopolistic competition lies in the strengthening and also wider acceptance of perfect competition. We know that the idea of perfect competition appears in Cournot (1838), whose analysis was based on the maximizing behaviour of the participating firms at the point of equality of marginal revenue and marginal cost. These concepts were also present in the writings of the other French engineers of the early nineteenth century. The often-cited didactic example of the inconsistencies that arise in the application of marginal principles has been advanced by Dupuit ([1844], 1969) and is related to the imposition of the correct price of crossing the bridge. We know that the MC of crossing the bridge is zero and so must be the optimal price (toll) of crossing the bridge. But for a price equal to zero, there is no private incentive to build bridges and a positive price (toll) on the other hand leads to resource misallocation and net welfare loss.[18] Cournot’s and the French engineers’ ideas, however, could not attract attention in the early nineteenth century because of the absolute dominance of classical economics and their view of competition as a process of rivalry and not as a static situation. The depression of 1873-1896 created the necessary conditions for the appearance of new ideas, and as it has been observed in dismal situations such as those of depressions, people, often, tend to distant themselves from the harsh reality and are ready to accept idealized situations. Clearly, such situations are those that are described in perfect competition and so Edgeworth (1881) found a fertile ground to promote the notion of perfect competition by developing its formal requirements.

Once again, this analysis could not gain broad acceptance not only because of its unrealistic assumptions but also because dominance of the ideas of classical economists. Marshall sought to circumvent the problem by assimilating the classical tradition with neoclassical economics. The classical dynamical process of competition gradually was to be translated into static terms, that is, the number of producers and the type of product may characterize the form of competition. However, even in Marshall’s time, perfect competition was not fully formulated into an operational model and this job was accomplished, to a great extent, in Knight’s (1921) book, which was essentially his dissertation written under Young’s diligent supervision. Knight in his book described in a comprehensive and meticulous way the requirements of perfect competition that could be used in the real economy and in so doing he managed to operationalize and to popularize the concept. Nevertheless, Stigler (1957) argued that this detailed description of the requirements of perfect competition was responsible for the initial appeal of monopolistic competition in the 1930s and delayed the explicit incorporation of perfect competition in neoclassical economics. Meanwhile, the books of Chamberlin and Robinson sparked a renewed interest in the static analysis of market forms: key words such as monopoly, oligopoly, rigidity of prices, price discrimination, exploitation of labour, excess capacity and the like excited and activated the interest of economists and policy makers to eliminate these undesired features of markets. The depression of the 1930s, however, changed, once again, the perception of the majority of economists about the role of these mega-corporations and there was a widespread belief that government intervention was necessary for the limitation of market power of big businesses that were also responsible, at least in part, for the depression. In fact, the usual argument (e.g., Berle and Means, 1932) was that prices in the US economy became increasingly stickier in the consumer goods industries due to the concentrated and, therefore, monopolistic structure. These “sticky prices” undermined the already constrained purchasing power of consumers. The same phenomenon was also observed in the capital goods sector and so producers were less willing to invest in new plant and equipment. Price stickiness thus inhibited the recovery of both final product demand and investment demand; thereby, precipitating the depression. Naturally, such views offered the necessary economic rationale for government intervention in the markets. In fact, governments became increasingly more interested in correcting the operation of markets in the effort to bring them closer to the hypothetical perfectly competitive markets (Bishop, 1964; Dilorenzo and High, 1988). This is equivalent to saying that the actual markets were characterized by some degree of imperfection in their operation, and hence they were found in divergence from an ideal operation, which was identified with the notion of perfect competition.

In the interwar period, as a result of Sraffa’s critique, the attention was directed to the development of the theory of monopolistic (or imperfect) competition and the suppression of perfect competition. More specifically, the theorization of competition in its imperfect form during the 1930s led to the development of the field of industrial organization, which on the one hand encompassed the new theoretical refinements of the theory of the firm and forms of competition and on the other hand made an effort to give quantitative content in these forms. Meanwhile, data on prices, costs, output and concentration ratios started to be collected on an industry basis. It is important to point out that the systematic collection of such data begins at approximately the same time with the collection of national income and product accounts data compiled for the aggregate economy and macroeconomic purposes, lending further support to the idea of two parallel revolutions in economic theory that took place at approximately the same time period. We know that the Keynesian revolution continued successfully and after WWII until the late 1960s, however, we cannot say the same about the monopolistic competition revolution which did not last for long, with economists gradually rediscovering the notion of perfect competition. As Stigler notes:

The theory of imperfect competition has raised questions which it cannot answer satisfactorily until the theory of perfect competition has been much more fully developed. […] the chief work of economic theorists should for the present still be in the theory of perfect competition. (Stigler, 1937, p. 707)

Meanwhile, many neoclassical economists in the USA (this time in Chicago) perceived the monopolistic competition revolution as a departure from the strict scientific analysis that economics ought to follow and, worse of all, as a critique of the actual market system which in turn necessitated government intervention in the economy. Stigler was very specific about the implications of monopolistic competition to the neoclassical theory of the firm. He described monopolistic competition and its implications in the following terms:

The new theory, in other words, has become something of a destructing fad. It seems often to be an escape from the very hard thinking necessary to secure a satisfactory and useful theory of perfect competition. Sound theories of price and production are indispensable to the solution of even the simplest practical problems. Yet the majority of the writers on imperfect competition seem not to realize that almost all the important concepts they have taken from perfect competition are suspect. (Stigler, 1937, p. 708)

Furthermore, Stigler (1937) claimed that the “newer literature of imperfect competition” is so complex that is incomprehensible for the legislator and the layman and so it is extremely difficult to find useful applications. In fact, both Stigler and Friedman systematically and forcefully opposed all efforts for further elaboration and possible improvement of the theory of monopolistic competition. An example of how much Stigler objected to monopolistic competition is his textbook in microeconomics ([1942], 1966), where there is no reference at all to Chamberlin and the notion of monopolistic competition, while Robinson is only mentioned en passant in the discussion of price discrimination. Stigler’s opposition was based on the idea that such a direction of research in monopolistic competition would render economic analysis more case oriented and, therefore, the lack of generalizations would make economic theory less scientific.

Friedman (1953) on the other hand, argued against monopolistic competition mainly on methodological grounds, i.e., a model is judged according to its predictive content and not the realism of its assumptions.[19] In this context, he used the example of the price effects of an indirect tax imposed on cigarettes which could be predicted with sufficient accuracy using partial equilibrium analysis and perfect competition although the cigarettes industry possessed the characteristics of monopolistic or oligopolistic competition.

A characteristically different