Economics can therefore no longer wave its preferred totem, but must instead only derive supply as a point determined by intersection of the marginal cost and marginal revenue curves.

Worse still, once we integrate this result with the fact that the demand curve can have any shape at all, the entire 'Totem of the Micro' has to be discarded. Instead of two simple intersecting lines, we have at least two squiggly lines for the demand side marginal revenue and price, both of which will be curves an aggregate marginal cost curve, and lots of lines joining the many intersections of the marginal revenue curve with the marginal cost curve to the price curve. The real Totem of the Micro is not the one shown at the beginning of this chapter, but a couple of strands of noodles wrapped around a chopstick, with lots of toothpicks thrown on top.17 There is thus very little left of conventional economic theory. The last two chapters leave very little of the 'Totem of the Micro' standing: in place of the simple intersecting supply and demand curve of conventional neocla.s.sical belief, we have wavy intersecting lines. But even this is too generous to the neocla.s.sical model, because as Sraffa pointed out almost ninety years ago there is no empirical justification for the one neocla.s.sical microeconomics concept that we have not yet critiqued: the rising marginal cost curve.

5 | THE PRICE OF EVERYTHING AND THE VALUE OF NOTHING.

Why most products cost less to produce as output rises.

We have already seen that both the demand and supply aspects of conventional economic a.n.a.lysis are unsound: first, market demand curves don't obey the 'Law' of Demand, and can have any shape at all. Secondly, a supply curve doesn't exist.

But surely, on the supply side, it makes sense that, to elicit a larger supply of a commodity, a higher price must be offered?

There is, in fact, an alternative proposition, which held sway in economics for its first century. This was the argument of the cla.s.sical school of economics that price was set by the cost of production, while the level of demand determined output.1 When this proposition is put in the same static form as economics uses to describe a commodity market, it translates as a horizontal (or even a falling) supply curve, so that the market price doesn't change as the quant.i.ty produced rises (and it can actually fall). This chapter shows that, though the modern neocla.s.sical position is superficially more appealing and apparently more sophisticated, there are logical problems with it which mean that the cla.s.sical position is a more accurate description of reality.

The kernel.

One of the peculiar aspects of modern Western society is that the majority of the population has no direct experience of how the commodities it consumes are produced. Only a small and decreasing minority is directly involved in production, and only a minority of that minority has direct knowledge of how factories are designed and managed. In contrast to consumption, the conditions under which commodities are produced are a mystery to most people, and the economic a.n.a.lysis of production appears to illuminate that mystery.

Neocla.s.sical theory argues that, in the 'short run,' productivity falls as output rises, so that higher levels of output result in higher prices. The 'marginal cost curve' therefore slopes upwards, and a higher price has to be offered to entice firms to produce a higher output.

Though this sounds intuitively plausible, when this theory was put to those who do know how factories are designed and managed, they rejected it as 'the product of the itching imaginations of uninformed and inexperienced arm-chair theorizers' (Lee 1998: 73, citing Tucker).

How could something which seems so reasonable to the inexperienced be so absurd according to those 'in the know'? The answer in part lies in the a.s.sumptions economists make about production. Though these seem sound to the uninitiated, two key a.s.sumptions are in fact contradictory: if one applies for a given industry, then the other one almost certainly does not. When one applies, supply and demand become interdependent; when the other does, the marginal cost curve is likely to be horizontal.

Economic theory also doesn't apply in the 'real world' because engineers purposely design factories to avoid the problems that economists believe force production costs to rise. Factories are built with significant excess capacity, and are also designed to work at high efficiency right from low to full capacity. Only products that can't be produced in factories (such as oil) are likely to have costs of production that behave the way economists expect.

The outcome is that costs of production are normally either constant or falling for the vast majority of manufactured goods, so that average and even marginal cost curves are normally either flat or downward sloping. This causes manufacturers no difficulties, but it makes life impossible for neocla.s.sical economists, since most of neocla.s.sical theory depends on supply curves sloping upwards.

The roadmap.

In this chapter I outline the neocla.s.sical a.n.a.lysis of production, which concludes that productivity will fall as output rises, leading to rising costs of production. This in turn leads to a need for the market price to rise if producers are to supply more, which economists represent as a 'supply curve' that slopes upward in price.

Next I detail Sraffa's argument that two crucial a.s.sumptions of this a.n.a.lysis that supply and demand are independent, and that at least one input to production can't be varied in the short run are mutually exclusive. A number of potential neocla.s.sical rejoinders to this argument are considered and dismissed.

The outline of the neocla.s.sical model of production below will probably convince you that the theory makes sense, but as with the corresponding section in Chapter 3, it is almost certain to bore you senseless. It is also unavoidably laden with jargon, and less accessible than Chapter 3 since few of us have any experience of production to the same depth as we have experience of consumption. So go back to the coffee pot, pour yourself a strong one, and read on.

Diminishing productivity causes rising price.

The neocla.s.sical theory of production argues that capacity constraints play the key role in determining prices, with the cost of production and therefore prices rising as producers try to squeeze more and more output out of a fixed number of machines, in what they call 'the short run.' The short run is a period of time long enough to change variable inputs such as labor but not long enough to change fixed inputs such as machines or for new entrants to come into the industry.

The argument has several stages: stage one puts the proposition that productivity falls as output rises; stage two takes the declining productivity argument and rephrases it as rising costs; and stage three determines the point of maximum profitability by identifying where the gap between revenue and costs is greatest.

Stage one: productivity falls as output rises Neocla.s.sical theory a.s.serts that the supply curve slopes upward because productivity falls as output rises. This falling productivity translates into a rising price. There is thus a direct link between what economists call 'marginal productivity' the amount produced by the last worker and 'marginal cost' the cost of producing the last unit.

Table 5.1 shows an example of production as neocla.s.sicals imagine it. This mythical firm has fixed costs of $250,000, and pays its workers a wage of $1,000.2 It can sell as many units as it can produce at the market price of $4. To produce output at all, the firm must hire workers: with no workers, output is zero. The first worker enables the firm to produce 52 units of output. This is shown in the first row of the table: the labor input is one unit, and total output is 52 units.

The marginal product of this worker the difference between production without him (zero) and production with is 52 units. The marginal cost of the output is the worker's wage $1,000 divided by the number of units produced 52 which yields a marginal cost of $19.20.

The average fixed costs of output at this point are enormous $250,000 divided by just 52, or $4,807 per unit. The average total cost is $251,000 divided by 52, or $4,827 per unit which implies a loss of $4,823 per unit sold, if this were the chosen level of production.

At this stage, production benefits from economies of scale. Just one worker had to perform all tasks, whereas a second worker allows them to divide up the jobs between them, so that each specializes to at least that extent. With specialization, the productivity of both workers rises. The same process continues with the ninth and tenth workers, so that the marginal product of the ninth the amount he adds to output over and above the amount produced by eight workers is 83.6 units. Similarly, the marginal product of the tenth worker is 87.5 units.

TABLE 5.1 Input and output data for a hypothetical firm.

If the firm actually produced this number of units, it would lose $257,207 dollars more than its fixed costs. However, the process of rising marginal productivity and therefore falling marginal cost comes to the rescue as output rises. By the 100th worker, the firm is still making a loss, but the loss is falling because its marginal cost has fallen below the sale price. The 100th worker adds 398.5 units to output, at a marginal cost of $1,000 divided by 398.5, or just $2.50 a unit. This is less than the sale price of $4 a unit, so the firm is making a profit on the increase in output but only enough to reduce its losses at this stage, rather than to put it into the black.

Black ink arrives with the 277th worker, who brings in $3,090 profit the proceeds of selling the 772.5 additional units the worker produces at the sale price of $4 a unit for the cost of his wage of $1,000.

This process of rising marginal productivity continues right up to the 400th worker hired. By this stage, marginal cost has fallen dramatically. The 400th worker adds 850 units to output, so that the marginal cost of his output is the wage of $1,000 divided by 850, or $1.18 (rounded up to $1.2 in the table). Average fixed costs, which were enormous at a tiny level of output, are relatively trivial at the output level of 233,333 units: they are down to just over a dollar.

From this point on, productivity of each new worker ceases to rise. Each new worker adds less to output than his predecessor. The rationale for this is that the ratio of workers the 'variable factor of production' to machinery the 'fixed factor of production' has exceeded some optimal level. Now each extra worker still adds output, but a diminishing rate. In economic parlance, we have reached the region where diminishing marginal productivity applies. Since marginal product is now falling, marginal cost will start to rise.

But profit continues to rise, because though each additional worker adds less output and therefore brings in less revenue, the revenue from the additional units still exceeds the cost of hiring the worker. In economic parlance, marginal revenue exceeds marginal cost.

We can see this with the 500th worker, who adds 800.5 units to output. The marginal cost of his output is the wage ($1,000) divided by 800.5, or $1.25 (rounded down in the table). This is higher than the minimum level of $1.18 reached with the 500th worker. But the additional units this worker produces can all be sold at $4, so the firm still makes a profit out of employing the 500th worker.

The same principle still applies for the 600th worker, and the 700th. Productivity has dropped sharply now, so that this worker adds only 401.5 units to output, for a marginal cost of $2.50. But this is still less than the amount the additional output can be sold for, so the firm makes a profit out of this worker.

This process of rising profit comes to an end with the 747th worker, whose additional product 249.7 units can only be sold for $998.8, versus the cost of his wage of $1,000. From this point on, any additional workers cost more to employ than the amount of additional output they produce can be sold for.

The firm should therefore employ 746 workers, and maximize its profit at $837,588. At this point, the marginal cost of production equals the marginal revenue from sale, and profit is maximized.

The 800th adds 52 units, for a now soaring marginal cost of $19.20. By the time we get to the 812th worker, workers are metaphorically speaking falling over each other on the factory floor, and this worker adds a mere 3.3 units to output, for a marginal cost of $300. The next worker actually reduces output.

5.1 Product per additional worker falls as the number of workers hired rises 5.2 Swap the axes to graph labor input against quant.i.ty 5.3 Multiply labor input by the wage to convert Y-axis into monetary terms, and add the sales revenue 5.4 Maximum profit occurs where the gap between total cost and total revenue is at a maximum 5.5 Deriving marginal cost from total cost 5.6 The whole caboodle: average and marginal costs, and marginal revenue From minnow to market The exposition above simply describes the situation for a single firm. To derive the market supply curve, we have to aggregate the supply curves of a mult.i.tude of producers just as to complete the derivation of the market demand curve, the demand curves of a mult.i.tude of consumers had to be added together.3 Since each individual firm's marginal cost curve is upward sloping, the market supply curve is also upward sloping.

5.7 The upward-sloping supply curve is derived by aggregating the marginal cost curves of numerous compet.i.tive firms Things don't add up.

There is no doubt that the economic a.n.a.lysis of production has great superficial appeal sufficient to explain much of the fealty which neocla.s.sical economists swear to their vision of the market. But at a deeper level, the argument is fundamentally flawed as Piero Sraffa first pointed out in 1926.

The crux of Sraffa's critique was that 'the law of diminishing marginal returns' will not apply in general in an industrial economy. Instead, Sraffa argues that the common position would be constant marginal returns, and therefore horizontal (rather than rising) marginal costs.

Sraffa's argument const.i.tutes a fundamental critique of economic theory, since, as I've just explained, diminishing marginal returns determine everything in the economic theory of production: the output function determines marginal product, which in turn determines marginal cost. With diminishing marginal productivity, the marginal cost of production eventually rises to equal marginal revenue. Since firms seek to maximize profit, and since this equality of (rising) marginal cost to marginal revenue gives you maximum profit, this determines the level of output.

If instead constant returns are the norm, then the output function instead is a straight line through the origin, just like the total revenue line though with a different slope. If (as a factory owner would hope) the slope of revenue is greater than the slope of the cost curve, then after a firm had met its fixed costs, it would make a profit from every unit sold: the more units it sold, the greater its profit would be.

In terms of the model of perfect compet.i.tion, there would be no limit to the amount a compet.i.tive firm would wish to produce, so that neocla.s.sical theory could not explain how firms (in a compet.i.tive industry) decided how much to produce. In fact, according to the conventional model, each firm would want to produce an infinite amount.

5.8 Economic theory doesn't work if Sraffa is right.

This is so patently impossible within the uncorrected neocla.s.sical model that, when told of Sraffa's critique, most neocla.s.sicals simply dismiss it out of hand: if Sraffa was right, then why don't firms produce an infinite amount of goods? Since they don't, Sraffa must be wrong.

This knee-jerk response to Sraffa's critique brings to mind the joke that an economist is someone who, when shown that something works in practice, comments, 'Ah, but does it work in theory?' Sraffa instead put the opposite case: sure, the neocla.s.sical model of production works in theory, if you accept its a.s.sumptions. But can the conditions that the model a.s.sumes actually apply in practice? If they can't, then regardless of how watertight the theory might be given its a.s.sumptions, it will be irrelevant in practice. It therefore should not be used as the theory of production, because above all else, such a theory must be realistic.

Sraffa's argument focused upon the neocla.s.sical a.s.sumptions that there were 'factors of production' which were fixed in the short run, and that supply and demand were independent of each other. He argued that these two a.s.sumptions could not be fulfilled simultaneously.

In circ.u.mstances where it was valid to say that some factor of production was fixed in the short run, supply and demand would not be independent, so that every point on the supply curve would be a.s.sociated with a different demand curve. On the other hand, in circ.u.mstances where supply and demand could justifiably be treated as independent, then in general it would be impossible for any factor of production to be fixed. Hence the marginal costs of production would be constant.

Sraffa began by noting that the preceding cla.s.sical school of economics also had a 'law of diminishing marginal returns.' However, for the cla.s.sical school, it was not part of price theory, but part of the theory of income distribution. Its application was largely restricted to the explanation of rent.

The cla.s.sical argument was that farming would first be done on the best land available, and only when this land was fully utilized would land of a lesser quality be used. Thus, as population grew, progressively poorer land would be brought into use. This poorer land would produce a lower yield per acre than the better land. Diminishing marginal returns therefore applied, but they occurred because the quality of land used fell not because of any relationship between 'fixed' and 'variable' factors of production.

Sraffa argued that the neocla.s.sical theory of diminishing marginal productivity was based on an inappropriate application of this concept in the context of their model of a compet.i.tive economy, where the model a.s.sumed that all firms were so small relative to the market that they could not influence the price for their commodity, and that factors of production were h.o.m.ogeneous. In the neocla.s.sical model, therefore, falling quality of inputs couldn't explain diminishing marginal productivity. Instead, productivity could only fall because the ratio of 'variable factors of production' to fixed factors exceeded some optimal level.

The question then arises of when is it valid to regard a given factor of production say, land as fixed. Sraffa said that this was a valid a.s.sumption when industries were defined very broadly, but this then contradicted the a.s.sumption that demand and supply are independent.

Sraffa's broad arrow If we take the broadest possible definition of an industry say, agriculture then it is valid to treat factors it uses heavily (such as land) as fixed. Since additional land can only be obtained by converting land from other uses (such as manufacturing or tourism), it is clearly difficult to increase that factor in the short run. The 'agriculture industry' will therefore suffer from diminishing returns, as predicted.

However, such a broadly defined industry is so big that changes in its output must affect other industries. In particular, an attempt to increase agricultural output will affect the price of the chief variable input labor as it takes workers away from other industries (and it will also affect the price of the 'fixed' input).

This might appear to strengthen the case for diminishing returns since inputs are becoming more expensive as well as less productive. However, it also undermines two other crucial parts of the model: the a.s.sumption that demand for and supply of a commodity are independent, and the proposition that one market can be studied in isolation from all other markets.

Instead, if increasing the supply of agriculture changes the relative prices of land and labor, then it will also change the distribution of income. As we saw in Chapter 2, changing the distribution of income changes the demand curve. There will therefore be a different demand curve for every different position along the supply curve for agriculture. This makes it impossible to draw independent demand and supply curves that intersect in just one place. As Sraffa expressed it: If in the production of a particular commodity a considerable part of a factor is employed, the total amount of which is fixed or can be increased only at a more than proportional cost, a small increase in the production of the commodity will necessitate a more intense utilization of that factor, and this will affect in the same manner the cost of the commodity in question and the cost of the other commodities into the production of which that factor enters; and since commodities into the production of which a common special factor enters are frequently, to a certain extent, subst.i.tutes for one another the modification in their price will not be without appreciable effects upon demand in the industry concerned. (Sraffa 1926: 539) These non-negligible impacts upon demand mean that the demand curve for this 'industry' will shift with every movement along its supply curve. It is therefore not legitimate to draw independent demand and supply curves, since factors that alter supply will also alter demand. Supply and demand will therefore intersect in multiple locations, and it is impossible to say which price or quant.i.ty will prevail.

Thus while diminishing returns do exist when industries are broadly defined, no industry can be considered in isolation from all others, as supply and demand curve a.n.a.lysis requires.

5.9 Multiple demand curves with a broad definition of an industry.

As you can see, Sraffa's argument here was a precursor of the Sonnenschein-Mantel-Debreu conditions that undermine the neocla.s.sical mode of the market demand curve, a.n.a.lyzed from the point of view of the producer rather than the consumer. This allows an upward-sloping supply curve to be drawn, but makes it impossible to derive an independent demand curve.

Sraffa's next argument leaves the demand curve intact, but undermines the concept of a rising supply curve.

Sraffa's narrow arrow When we use a more realistic, narrow definition of an industry say, wheat rather than agriculture Sraffa argues that, in general, diminishing returns are unlikely to exist. This is because the a.s.sumption that supply and demand are independent is now reasonable, but the a.s.sumption that some factor of production is fixed is not.

While neocla.s.sical theory a.s.sumes that production occurs in a period of time during which it is impossible to vary one factor of production, Sraffa argues that in the real world, firms and industries will normally be able to vary all factors of production fairly easily.4 This is because these additional inputs can be taken from other industries, or garnered from stocks of underutilized resources. That is, if there is an increased demand for wheat, then rather than farming a given quant.i.ty of land more intensively, farmers will instead convert some land from another crop say, barley to wheat. Or they will convert some of their own land which is currently lying fallow to wheat production. Or farmers who currently grow a different crop will convert to wheat. As Sraffa expressed it: If we next take an industry which employs only a small part of the 'constant factor' (which appears more appropriate for the study of the particular equilibrium of a single industry), we find that a (small) increase in its production is generally met much more by drawing 'marginal doses' of the constant factor from other industries than by intensifying its own utilization of it; thus the increase in cost will be practically negligible. (Ibid.: 539) This means that, rather than the ratio of variable to 'fixed' outputs rising as the level of output rises, all inputs will be variable, the ratio of one input to another will remain constant, and productivity will remain constant as output rises. This results in constant costs as output rises, which means a constant level of productivity. Output will therefore be a linear function of the inputs: increase inputs by 20 percent, and output will rise by the same amount.

Since the shapes of the total, average and marginal cost curves are entirely a product of the shape of the output curve, a straight-line output curve results in constant marginal costs, and falling average costs.

With this cost structure, the main problem facing the firm is reaching its 'break-even point,' where the difference between the sale price and the constant variable costs of production just equal its fixed costs. From that point on, all sales add to profit. The firm's objective is thus to get as much of the market for itself as it can. This, of course, is not compatible with the neocla.s.sical model of perfect compet.i.tion.

Irrational managers Sraffa's broad and narrow critiques accept that, if a firm's output was actually constrained by a fixed resource, then its output would at first rise at an accelerating rate, as the productivity of additional variable inputs rose; then the rate of growth of output would reach a peak as the maximum level of productivity was reached, after which output would still rise, but at a diminishing rate. Finally, when even more variable inputs were added, total output would actually start to fall. In the vernacular of economics, the firm would at first experience increasing marginal productivity, then diminishing marginal productivity, and finally negative marginal productivity.

However, Sraffa disputes even this proposition. He instead argues that a firm is likely to produce at maximum productivity right up until the point at which diminishing marginal productivity sets in. Any other pattern, he argues, shows that the firm is behaving irrationally.

His argument is probably best ill.u.s.trated with an a.n.a.logy. Imagine that you have a franchise to supply ice creams to a football stadium, and that the franchise lets you determine where patrons are seated. If you have a small crowd one night say, one quarter of capacity would you spread the patrons evenly over the whole stadium, so that each patron was surrounded by several empty seats?

Of course not! This arrangement would simply force your staff to walk farther to make a sale. Instead, you'd leave much of the ground empty, thus minimizing the work your staff had to do to sell the ice creams. There's no sense in using every last inch of your 'fixed resource' (the stadium) if demand is less than capacity.

Sraffa argued that the same logic applied to a farm, or to a factory. If a variable input displays increasing marginal returns at some scale of output, then the sensible thing for the farmer or factory owner to do is leave some of the fixed resource idle, and work the variable input to maximum efficiency on part only of the fixed resource.

To give a numerical example, consider a wheat farm of 100 hectares, where one worker per hectare produces an output of 1 bushel per hectare, 2 workers per hectare produces 3 bushels, 3 per hectare produces 6 bushels, 4 per hectare produces 10 bushels, and 5 workers per hectare produces 12 bushels.

According to economists, if a farmer had 100 workers, he would spread them out 1 per hectare to produce 100 bushels of wheat. But, according to Sraffa, the farmer would instead leave 75 hectares of the farm idle, and work 25 hectares with the 100 workers to produce an output of 250 bushels. The farmer who behaves as Sraffa predicts comes out 150 bushels ahead of any farmer who behaves as economists expect.

Similarly, economic theory implies that a farmer with 200 workers would spread them over the farm's 100 hectares, to produce an output of 300 bushels. But Sraffa says that a sensible farmer would instead leave 50 hectares fallow, work the other 50 at 4 workers per hectare, and produce an output of 500 bushels. One again, a 'Sraffian' farmer is ahead of an 'economic' one, this time by 200 bushels.

The same pattern continues right up until the point at which 400 workers are employed, when finally diminishing marginal productivity sets in. A farm will produce more output by using less than all of the fixed input, up until this point.

5.10 A farmer who behaved as economists advise would forgo the output shown in the gap between the two curves This might seem a minor point, but as usual with Sraffa, there is a sting in the tail. If marginal cost is constant, then average cost must be greater than marginal cost, so that any firm which sets price equal to marginal cost is going to make a loss. The neocla.s.sical theory of price-setting can therefore only apply when demand is such that all firms are producing well beyond the point of maximum efficiency. The theory therefore depends on both labor and capital normally being fully employed.

So is full employment not just of labor, but of other resources as well the norm in a market economy? If all you read was neocla.s.sical economic theory, you'd believe so right from the standard textbook definition of economics as 'the study of the allocation of limited resources to unlimited wants.'

Of course, there is recorded unemployment of labor, but neocla.s.sical economists (at least those of the 'freshwater' variety see pp. 25566) attribute that to the laborleisure choice that households make: those who are recorded as unemployed are really deciding that at the wage rates that are on offer, they'd prefer not to work. But surely firms use their capital efficiently, so that it the 'fixed resource' is fully employed?

5.11 Capacity utilization over time in the USA.

Even a cursory look at the economic data shows that this is not so. Even during the boom years of the 1960s, at least 10 percent of the USA's industrial capacity lay idle; even during subsequent booms, capacity utilization rarely reached 85 percent; and capacity utilization has rarely exceeded 80 percent since 2000, and fell below 70 percent in the depths of the Great Recession (see Figure 5.11).

This situation may seem bizarre from a neocla.s.sical point of view and there is a trend towards lower utilization over time that could indicate a secular problem but it makes eminent sense from a very realistic perspective on both capitalism and socialism put forward by the Hungarian economist Janos Kornai.

Resource-constrained versus demand-constrained economies.

Kornai's a.n.a.lysis was developed to try to explain why the socialist economies of eastern Europe had tended to stagnate (though with superficially full employment), while those of the capitalist West had generally been vibrant (though they were subject to periodic recessions). He noted that the defining feature of socialist economies was shortage: 'In understanding the problems of a socialist economy, the problem of shortage plays a role similar to the problem of unemployment in the description of capitalism' (Kornai 1979: 801).

Seeing this as an inherent problem of socialism and one that did not appear to afflict capitalism Kornai built an a.n.a.lysis of both social systems, starting from the perspective of the constraints that affect the operations of firms: The question is the following: what are the constraints limiting efforts at increasing production? [...] Constraints are divided into three large groups: 1 Resource constraints: The use of real inputs by production activities cannot exceed the volume of available resources. These are constraints of a physical or technical nature [...]

2 Demand constraints: Sale of the product cannot exceed the buyer's demand at given prices.

3 Budget constraints: Financial expenses of the firm cannot exceed the amount of its initial money stock and of its proceeds from sales. (Credit will be treated later.)5 Which of the three constraints is effective is a defining characteristic of the social system [...] (Ibid.: 803) Kornai concluded that 'With the cla.s.sical capitalist firm it is usually the demand constraint that is binding, while with the traditional socialist firm it is the resource constraint' (Kornai 1990: 27).

This meant there were unemployed resources in a capitalist economy of both capital and labor but this also was a major reason for the relative dynamism of capitalist economies compared to socialist ones. Facing compet.i.tion from rivals, insufficient demand to absorb the industry's potential output, and an uncertain future, the capitalist firm was under pressure to innovate to secure as much as possible of the industry's demand for itself. This innovation drove growth, and growth added yet another reason for excess capacity: a new factory had to be built with more capacity than needed for existing demand, otherwise it would already be obsolete.

Therefore most factories have plenty of 'fixed resources' lying idle for very good reasons and output can easily be expanded by hiring more workers and putting them to work with these idle 'fixed resources.' An increase in demand is thus met by an expansion of both employment of labor and the level of capital utilization and this phenomenon is also clearly evident in the data.

5.12 Capacity utilization and employment move together Kornai's empirically grounded a.n.a.lysis thus supports Sraffa's reasoning: diminishing marginal productivity is, in general, a figment of the imaginations of neocla.s.sical economists. For most firms, an increase in production simply means an increased use of both labor and currently available machinery: productivity remains much the same, and may even increase as full capacity is approached and surveys of industrialists, which I discuss later in this chapter, confirm this.

Summing up Sraffa.

Sraffa's critiques mean that the economic theory of production can apply in only the tiny minority of cases that fall between the two circ.u.mstances he outlines, and only when those industries are operating beyond their optimum efficiency. Then such industries will not violate the a.s.sumed independence of supply and demand, but they will still have a relatively 'fixed' factor of production and will also experience rising marginal cost. Sraffa concludes that only a tiny minority of industries are likely to fit all these limitations: those that use the greater part of some input to production, where that input itself is not important to the rest of the economy. The majority of industries are instead likely to be better represented by the cla.s.sical theory, which saw prices as being determined exclusively by costs, while demand set the quant.i.ty sold. As Sraffa put it: Reduced within such restricted limits, the supply schedule with variable costs cannot claim to be a general conception applicable to normal industries; it can prove a useful instrument only in regard to such exceptional industries as can reasonably satisfy its conditions. In normal cases the cost of production of commodities produced compet.i.tively must be regarded as constant in respect of small variation in the quant.i.ty produced. And so, as a simple way of approaching the problem of compet.i.tive 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) 5.13 Costs determine price and demand determines quant.i.ty.

If not rising marginal cost, what?

Sraffa's argument dismisses the neocla.s.sical proposition that rising costs and constant (or falling) marginal revenue determines the output from a single firm, or a single industry. This raises the question that if increasing costs don't constrain a firm's output, what does?

Sraffa's argument is simple. The output of a single firm is constrained by all those factors that are familiar to ordinary businessmen, but which are abstracted from economic theory. These are, in particular, rising marketing and financing costs, both of which are ultimately a product of the difficulty of encouraging consumers to buy your output rather than a rival's. These in turn are a product of the fact that, in reality, products are not h.o.m.ogeneous, and consumers do have preferences for one firm's output over another's. Sraffa mocked the economic belief that the limit to a firm's output is set by rising costs, and emphasized the importance of finance and marketing in constraining a single firm's size: Business men, who regard themselves as being subject to compet.i.tive conditions, would consider absurd the a.s.sertion that the limit to their production is to be found in the internal conditions of production in their firm, which do not permit of the production of a greater quant.i.ty without an increase in cost. The chief obstacle against which they have to contend when they want gradually to increase their production does not lie in the cost of production which, indeed, generally favors them in that direction but in the difficulty of selling the larger quant.i.ty of goods without reducing the price, or without having to face increased marketing expenses. (Ibid.) Economics a.s.sumes this real-world answer away by a.s.suming that products are h.o.m.ogeneous, consumers are indifferent between the outputs of different firms and decide their purchases solely on the basis of price, that there are no transportation costs, etc. In such a world, no one needs marketing, because consumers already know everything, and only price (which consumers already know) distinguishes one firm's output from another. But Sraffa says that these postulates are the exception to the rule which applies in reality.

In most industries, products are heterogeneous, consumers do not know everything, and they consider other aspects of a product apart from price. Even where products are h.o.m.ogeneous, transportation costs can act to give a single firm an effective local monopoly. As a result, even the concept of a compet.i.tive market in which all firms are price-takers is itself suspect. Instead, most firms will to varying degrees act like monopolists who, according to neocla.s.sical theory, face a downward-sloping demand curve.

Each firm has a product that may fit within some broad category such as, for example, pa.s.senger cars but which is qualitatively distinguished from its rivals in a fashion that matters to a particular subset of buyers. The firm attempts to manipulate the demand for its product, but faces prohibitive costs in any attempt to completely eliminate their compet.i.tors and thus take over the entire industry. Not only must the firm persuade a different niche market to buy its product to convince Porsche buyers to buy Volvos, for example it must also convince investors and banks that the expense of building a factory big enough to produce for both market niches is worth the risk. Therefore, with the difficulty of marketing beyond your product's niche goes the problem of raising finance: The limited credit of many firms, which does not permit any one of them to obtain more than a limited amount of capital at the current rate of interest, is often a direct consequence of its being known that a given firm is unable to increase its sales outside its own particular market without incurring heavy marketing expenses. (Ibid.) Economic theory also can't be saved by simply adding marketing costs to the cost of production, and thus generating a rising marginal cost curve. As Sraffa pointed out, there are at least three flaws with this. First, it is a distortion of the truth marketing is not a cost of production, but a cost of distribution. Secondly, it is inconsistent with the underlying economic premise that marginal cost rises because of diminishing marginal productivity. Thirdly, it is implausible in the context of the economic theory of the firm. There is no point in 'saving' the concept of a rising marginal cost curve by introducing marketing costs, since this requires acknowledging that one firm's product differs from another. If products differ from one firm to another, then products are no longer h.o.m.ogeneous, which is an essential a.s.sumption of the theory of perfect compet.i.tion. It is far more legitimate to treat marketing as a cost of distribution, whose object is to alter the demand faced by an individual firm.

Sraffa's critique strengthens the case made in the preceding chapters. Rather than firms producing at the point where marginal cost equals marginal revenue, the marginal revenue of the final unit sold will normally be substantially greater than the marginal cost of producing it, and output will be constrained, not by marginal cost, but by the cost and difficulty of expanding sales at the expense of sales by compet.i.tors.6 Sraffa's alternative However, Sraffa was not satisfied with this revised picture, which is still dominated by intersecting marginal revenue and marginal cost curves. He instead expressed a preference for a more realistic model, which focused upon those issues that are most relevant to actual businesses.

The firm faces falling average costs as its large fixed costs are amortized over a larger volume of sales, and as its variable costs either remain constant or fall with higher output.7 It will have a target level of output which it tries to exceed, and a target markup which it tries to maintain. The size of the firm is constrained by the size of its niche within the given market, and the difficulty of raising finance for a much larger scale of operation.