TABLE 3.3 The commodities in Sippel's 'Revealed Preference' experiment To his great credit, Sippel concluded with an understated but accurate reflection on the implications of his experiment for economic theory: We conclude that the evidence for the utility maximization hypothesis is at best mixed. While there are subjects who appear to be optimizing, the majority of them do not. The high power of our test might explain why our conclusions differ from those of other studies where optimizing behavior was found to be an almost universal principle applying to humans and non-humans as well. In contrast to this, we would like to stress the diversity of individual behavior and call the universality of the maximizing principle into question [...]

We find a considerable number of violations of the revealed preference axioms, which contradicts the neocla.s.sical theory of the consumer maximizing utility subject to a given budget constraint. We should therefore pay closer attention to the limits of this theory as a description of how people actually behave, i.e. as a positive theory of consumer behavior. Recognizing these limits, we economists should perhaps be a little more modest in our 'imperialist ambitions' of explaining non-market behavior by economic principles. (Sippel 1997: 14423) Sippel did not speculate as to what his subjects were actually doing if they weren't in fact maximizing their utility, but it is fairly easy to show that these subjects were behaving rationally in the face of a real-world phenomenon of which armchair economic theorists are blithely unaware: the 'curse of dimensionality.'

Rational behavior and the curse of dimensionality.

The neocla.s.sical definition of rational behavior argues that a rational person, when confronted with a set of options, will attempt to choose the best option available. It appeared to Sippel that this was exactly what his subjects were doing: A closer look at the actual demand data corroborates the view that the subjects did not choose randomly. Every subject showed a marked preference for some of the goods while other goods were not chosen at all, even at low prices. Some subjects' demand was quite price inelastic, whereas others subst.i.tuted cheaper goods for their more expensive counterparts, e.g. c.o.ke for orange juice, sometimes to the extent that they always switched from one to the other, depending upon which was cheaper in the particular situation. There can be no doubt that the subjects tried to select a combination of goods that came as close as possible to what they really liked to consume given the respective budget constraints. (Ibid.: 1439) However, despite this intention to choose the best option, they failed to do so rationally according to Samuelson's rules. So what's at fault human behavior, or the neocla.s.sical model of rationality?

The latter, of course. It is a 'toy' model that looks OK on paper, but fails completely when one takes even a tiny step into the real world as Sippel's experiment did.

Let's look at what his subjects were being asked to do more closely. Sippel gave them a choice between eight different commodities, and let them choose any amount of them that they could afford with their budget. How many different 'shopping trolleys' could this mean they were looking at each containing a different combination of goods?

Unfortunately, the answer is 'an infinite number of shopping trolleys,' so let's simplify it and imagine that students considered their choices in discrete units say 5-minute segments for the videos and computer games (30 minutes, 35 minutes, and so on out to 60 minutes), 250ml units of drinks (400ml, 650ml, out to 2 liters), and 250 gram units of sweets (400 grams, 650 grams, out to 2 kilos). This means roughly eight different quant.i.ties for each of the eight goods. How many different shopping trolleys does that give us?

The answer will probably surprise you: you could fill over 16.7 million shopping trolleys with different combinations of these eight goods. Sixty-four would contain varying amounts of only one good from 30 to 60 minutes of video, from 400 grams to 2 kilos of candy. The other 16.7 million-plus would have varying combinations of all the goods available.

This is a consequence of the real-world phenomenon that computer scientists have dubbed 'the curse of dimensionality.' The standard neocla.s.sical 'toy' model of consumption shows you choosing between two different commodities. Most of these drawings don't show quant.i.ties on their axes, but if the quant.i.ties being considered were between zero and ten units of each good, then there would be 121 different combinations you could choose: zero units of both ([0,0]), ten units of both ([10,10]), and another 119 combinations in addition to that ([0,1], [1,0] right out to [10,9] and [9,10]).

The general rule for choices involving many commodities is that the number of different combinations equals one plus the number of units that you could buy of each commodity,17 raised to the power of the number of commodities you are considering. In the simple two-commodity case, this results in 11-squared choices or 121. Your budget might allow you to rule out 90 percent of these, leaving just 10 or so choices to consider.

In Sippel's experiment, however, this resulted in 8 raised to the power of 8 or in longhand 8 by 8 by 8 by 8 by 8 by 8 by 8 by 8 by 8, which equals 16.7 million.18 Many of these 16.7 million combinations would be ruled out by the budget the trolley containing the maximum amount of each item is clearly unattainable, as are many others. But even if the budget ruled out 99.99 percent of the options for being either too expensive or too cheap compared to the budget there would still be over 1,600 different shopping trolleys that Sippel's subjects had to choose between every time.

The neocla.s.sical definition of rationality requires that, when confronted with this amount of choice, the consumer's choices are consistent every time. So if you choose trolley number 1355 on one occasion when trolley 563 was also feasible, and on a second occasion you reversed your choice, then according to neocla.s.sical theory, you are 'irrational.'

Nonsense. The real irrationality lies in imagining that any sentient being could make the number of comparisons needed to choose the optimal combination in finite time. The weakness in the neocla.s.sical vision of reality starts with the very first principle of 'Completeness': it is simply impossible to hold in your head or any other data storage device a complete set of preferences for the bewildering array of combinations one can form from the myriad range of commodities that confront the average Western shopper. With this principle being impossible, any sane person's shopping behavior will certainly also violate the neocla.s.sical rules of Transitivity and Convexity (and probably Non-satiation as well). But it will be because the neocla.s.sical principles themselves are irrational, not because the shopper is.

Consider, for example, your regular visit to a supermarket. The typical supermarket has between 10,000 and 50,000 items, but let's segment them into just 100 different groups. How many different shopping trolleys could you fill if you limited your decision to simply whether to buy or not buy one item from each group?

You would be able to fill two to the power of one hundred shopping trolleys with different combinations of these goods: that's 1,267,650,600,228,229,401,496,703,205,376 trolleys in total, or in words over 1,000 million trillion trillion shopping trolleys. If you could work out the utility you gained from each trolley at a rate of 10 trillion trolleys per second, it would take you 100 billion years to locate the optimal one.

Obviously you don't do that when you go shopping. Instead, what you do is use a range of commonplace heuristics to reduce the overwhelming array of choices you face to something manageable that you can complete in less than an hour. You part.i.tion your choices into a few basic groups, rather than looking at every separate product; and within the groups you use habit to guide your purchases if you normally have muesli for breakfast, you ignore cornflakes. Truly rational behavior is therefore not choosing the best option, but reducing the number of options you consider so that you can make a satisfactory decision in finite time.

This is a commonplace observation in computer science, which unlike economics has built its knowledge of how decisions are made from experimentation and experience. What are sometimes called the 'Laws of Computational Theory' put front and center in a rather paradoxical way the fact that most real-world problems have so many potential solutions that an optimum cannot be found: 1 You cannot compute nearly all the things you want to compute.

2 The things you can compute are too expensive to compute. (Ballard 2000: 6) The first law reflects research by Turing which established that most logical problems cannot be solved by a computer program. The second states that for the minority of problems that can be solved, the 'Curse of Dimensionality' means that an optimum solution cannot be found in finite time, no matter how much computing power is thrown at it. Computer scientists are much more informed than economists about the capacity of any reasoning system to solve even the simplest problems, and they are much more cautious as a result.

Economists should respect their greater knowledge, and accept that individual behavior will be 'satisficing' in nature rather than optimizing, as the behavioral economist Herbert Simon put it (Simon 1996).

Conclusion.

There are of course reasonable grounds to expect that, for many commodities, demand will rise as price falls. One, given by the marketer and statistician Andrew Ehrenberg, was that consumers allocated a fairly constant percentage of their spending to different cla.s.ses of commodities (shelter, food, clothing, etc.) and a fall in the price of any item within a cla.s.s resulted in an increase in the purchases of it, though very little change in the aggregate amount spent on that cla.s.s of commodities overall (Ehrenberg 1975: 2759).

This empirical reality cannot, however, rescue the neocla.s.sical theory of consumer behavior from its result that market demand curves derived from consumers having 'rational' preferences (as neocla.s.sical theory defines them) can have any shape at all. As I note later, this is an example of what is known in complexity theory as 'Emergent Behavior' that the behavior of the sum of a set of isolated individuals cannot be deduced from the behavior of any of them in isolation.

This could imply that the best research strategy to develop economics is to abandon the model of rational behavior as neocla.s.sical economics defines it and adopt the behavioral perspective of satisficing or bounded rationality instead.

I am more inclined to take Alan Kirman's lead here: that the failure of the endeavor to derive market rationality from individual rationality implies that the whole agenda of trying to derive systemic economic laws from the a.n.a.lysis of the isolated individual known as 'methodological individualism' is a waste of time. Instead, as Kirman put it, 'If we are to progress further we may well be forced to theories in terms of groups who have collectively coherent behavior' (Kirman 1989: 138). This implies that the old cla.s.sical economics focus on social cla.s.ses as the ideal level of a.n.a.lysis was correct even if many of the conclusions derived from that in the 1800s were false.

That is the approach I take to macroeconomic modeling as you will see in Chapters 13 and 14. I am inclined to leave studies of satisficing behavior to the psychologists.

So one half of the iconic 'supply and demand' model is unsound: what about the other half, the supply curve?

4 | SIZE DOES MATTER.

Why there is no supply curve.

The image of one downward-sloping line intersecting with another upward-sloping one to determine an equilibrium is so iconic to neocla.s.sical economics that a renowned wit once described it as the 'Totem' of economics. In a wonderful satire ent.i.tled 'Life among the Econ,' Swedish economist Axel Leijonhufvud imagined himself as an anthropologist investigating academic economists, whom he portrayed as a tribe living in the cold and arid Arctic: 'The Econ tribe occupies a vast territory in the far North. Their land appears bleak and dismal to the outsider, and travelling through it makes for rough sledding; but the Econ, through a long period of adaptation, have learned to wrest a living of sorts from it' (Leijonhufvud 1973: 327).

The Econ, he noted, were xenophobic towards the neighboring PolScis and the Sociogs tribes, obsessed with the building of 'modls,' and sharply divided into castes, the most numerous of which were the Micro and the Macro. The castes distinguished themselves from each other using Totems that were, to the outsider, remarkably similar. The 'Totem of the Micro' was a pair of lines labeled 'S' and 'D,' while (when Leijonhufvud wrote the paper in 1973) the totem of the Macro was a pair of intersecting lines labeled 'IS' and 'LM': The Totems are easily drawn, but deriving them logically from the underlying theory is another matter altogether. As we saw in the previous chapter, a demand curve derived in accordance with the underlying theory can have any shape at all it will more often look like a snake in a hurry than the simple downward-sloping line drawn here.

The supply curve suffers an even worse fate: it doesn't exist.

Economists attempt to derive the supply curve from their theory of how profit-maximizing firms decide how much output to produce. One essential step in this derivation is that firms must produce so that the price they are paid for their output equals what is known as the 'marginal cost' of production the additional expense incurred in producing one more unit of output. Unless this condition is met, a supply curve cannot be drawn.

4.1 Leijonhufvud's 'Totems' of the Econ tribe.

This explains the extreme hostility that neocla.s.sical economists have towards monopolies. It's not only because they can abuse the power that being a monopoly can confer: it's also because, according to neocla.s.sical theory, a monopoly will set its price above the marginal cost of production. If monopolies were the rule, then there could be no supply curve, and standard neocla.s.sical microeconomic a.n.a.lysis would be impossible.

Conversely, neocla.s.sical economists love the market structure they call 'perfect compet.i.tion,' because it guarantees that profit-maximizing behavior will cause firms to produce an output at which marginal cost equals price.

Only it won't. The manner in which neocla.s.sical economics derives the result that profit-maximizing behavior by compet.i.tive firms means that they will produce where marginal cost equals price commits one of the simplest mathematical mistakes possible: it confuses a very small quant.i.ty an 'infinitesimal,' as mathematicians describe it with zero.

When that error is corrected, it is easily shown that a compet.i.tive market will also set price above marginal cost, and therefore a supply curve that is independent of the demand curve can't be drawn. The other half of the 'Totem of the Micro' disappears.

The kernel.

Try this party trick: convince someone that the world is flat, starting from the premise that it is a sphere.

The argument is simple. If you take a small enough segment of the world say, the two feet your victim is standing on then the curvature of that segment is so small that it is, to all intents and purposes, flat. Then consider the segment you're standing on it is also so small that it is effectively flat.

Next, consider the angle between the two segments: it too will be so small that it is effectively zero. So these two small segments are effectively flat.

Finally, extrapolate your argument from these two tiny segments and the angle between them up to the level of the entire globe. If you consider the segment your victim occupies and the segment behind him, that pair is also effectively flat. Keep on going, and the entire world is flat.

The fallacy in the argument, clearly, is that while it will do as an approximation to treat your immediate surroundings as effectively flat, it will not do to ignore those imperceptible but non-zero angles if you move from the scale of one or two segments to the entire globe.

Yet this fallacy lies at the heart of the economic preference for small, compet.i.tive firms over large monopolistic ones. At crucial stages of the economic argument, an imperceptibly small quant.i.ty is treated as zero, and then all these zeros are added up to yield zero at the scale of an entire market. This is intellectually and mathematically unsound. When the correct position is imposed that something which is extremely small is nonetheless not zero the economic argument against monopolies and in favor of small compet.i.tive firms collapses.

Oh, and if your party trick convinces your victim? Then he is either stoned, or an economist.

Prelude: the War over Perfect Compet.i.tion.

Most of this book explains flaws in neocla.s.sical economic theory that have been known for decades, but have been ignored by neocla.s.sical economists. When I first wrote Debunking Economics, I thought that the argument presented in this chapter was a new critique.

As I found out shortly after the book was published in 2001, it wasn't: the same key point had been made forty-four years earlier, and not by a critic of neocla.s.sical economics but by one of the most strident defenders, George Stigler. In his paper 'Perfect compet.i.tion, historically contemplated' (Stigler 1957: 8, n. 31), Stigler applied one of the most basic rules of mathematics, the 'Chain Rule,' to show that the slope of the demand curve facing the compet.i.tive firm was exactly the same as the slope of the market demand curve see Figure 4.2.

If you haven't yet studied economics, then the importance of that result won't yet be obvious to you. But if you have, this should shock you: a central tenet of your introductory 'education' in economics is obviously false, and has been known to be so since at least 1957.

Stigler's mathematics deconstructed the demand curve for the individual firm into two components: the slope of the market demand curve; multiplied by how much market output changes given a change in the output of a single firm.

Neocla.s.sical theory a.s.sumes that the slope of the market demand curve is negative: a fall in price will cause demand to increase. So the demand curve for the individual firm can only be zero if the second component is zero: the amount that industry output changes given a change in output by a single firm.

4.2 Stigler's proof that the horizontal firm demand curve is a fallacy However, Stigler very correctly stated that this second component is not zero, but instead equals one. In the basic 'Marshallian' theory of the firm that is taught to undergraduates, individual firms are a.s.sumed not to react strategically to what other firms do or might do. Therefore if one firm changes its output by ten units, there is no instantaneous reaction to this by the other firms, so that industry output also changes by ten units (though it might alter afterwards as other firms adjust to the new market price). The ratio of the change in industry output to the change in output by a single firm is therefore 1.

As a consequence, the slope of the demand curve for the individual compet.i.tive firm equals the slope of the market demand curve. Far from the individual firm's demand curve being horizontal, it has the same negative slope as the market demand curve.

I was stunned. This had been known for over four decades, and yet economic textbooks everywhere continued to mouth the fallacy that the individual compet.i.tive firm had a horizontal demand curve?

Even by the standards of mendacity that I had come to expect of economic textbooks, this surprised me. Most critiques of neocla.s.sical theory involve complicated concepts like the critique of the 'Law of Demand' outlined in the previous chapter, or the disputes over the nature of capital in Chapter 7. Frequently, when I have criticized textbooks for not discussing these issues, I have been hit with the rejoinder that this material is just too complicated for undergraduates to understand: better leave it for more advanced courses.1 But this error in the theory is so simple that it can be explained in a few lines of English (and one line of calculus).

Neocla.s.sical economists ignored most of this book, but vigorously attacked this chapter. As I responded to their attacks, the critique grew in depth and complexity. Attempts to get it into neocla.s.sical journals failed, but it was published in a range of non-neocla.s.sical outlets, including the journal of interdisciplinary physics Physica A (Keen and Standish 2006), A Guide to What's Wrong with Economics (Keen and Fullbrook 2004), the Handbook of Pluralist Economics Education (Keen 2009a), and the Real-World Economics Review (Keen and Standish 2010).2 Though nothing in that war with neocla.s.sical economists challenged the accuracy of the case I first made in 2000, I have made extensive changes to this chapter to focus on the key challenge it makes to neocla.s.sical orthodoxy: that a 'supply curve' cannot be drawn. I have also added new material, including the key advance over the case made in 2000: a proof that the alleged profit-maximizing formula ('set marginal cost and marginal revenue equal to maximize profits') does not maximize profits. I derive another formula that does maximize profits, given the a.s.sumptions of neocla.s.sical theory.

The roadmap.

In this chapter I outline the neocla.s.sical a.n.a.lysis of monopolies on the one hand, and 'perfect compet.i.tion' on the other, and point out that the sole difference between them is that a monopolist is shown to face falling marginal revenue, whereas the compet.i.tive firm faces constant marginal revenue which is equal to the market price. From this proposition alone flows the crucial result, for the neocla.s.sical approach to economics, that a supply curve can be derived that is independent of the demand curve.

I then show that this proposition leads to logical fallacies: a quant.i.ty that economists a.s.sume is zero actually has to be minus one; firms that are allegedly profit maximizers must produce more than the amount which maximizes profits; zero amounts at the individual level must somehow aggregate to negative amounts at the aggregate.

A careful a.n.a.lysis of what is implied by this proposition that marginal revenue equals price for compet.i.tive firms shows that it is based on a simple mathematical error. Once this is corrected, it is obvious that a compet.i.tive market with profit-maximizing firms that faces the same cost conditions as a monopoly will produce the same amount at the same price.

It follows that the amount supplied by a compet.i.tive industry is not determined by the aggregate marginal cost curve alone, but instead depends on conditions of demand as well, as with a monopoly. A supply curve that is independent of the demand curve therefore cannot be derived.

Economic perfection.

Pejorative expressions abound in economics, despite its claim to be a value-free science, and 'perfect compet.i.tion' is possibly the most value-laden of all. To economists, however, the word 'perfect' has a very precise meaning: it is a market in which the compet.i.tively set price equals the marginal cost of production.

This is 'perfect' because, according to economic theory, it achieves the maximum possible gap between community welfare and the cost of providing it. Community welfare is maximized when the gap between total benefit to society from consuming a given product and the total cost of providing that benefit is as big as it can be. Given the shape that economists a.s.sume that these benefits and costs take the benefit of consumption rising but at a decreasing rate, the cost of production rising at an increasing rate the gap between the two is highest when the rate of change of total benefit equals the rate of change of total cost.

The demand curve (which we deconstructed in the last chapter) represents the rate of change of the total benefit, while the supply curve represents the rate of change of total cost. Therefore the benefit to society is maximized where these two rates of change one rising, the other falling are equal.

Producers are trying to maximize the benefit to them their profits not society's benefits. These two interests consumers aiming to get the maximum benefit out of consumption, producers trying to get the maximum profit out of production only coincide if the price equals the change in revenue that producers get from selling an extra unit, which economists call 'marginal revenue.' This is because the price the amount that consumers are willing to pay tells you the 'marginal utility' they get from the last item consumed. Only if this also equals the 'marginal revenue' that the producer gets from selling this very last unit of output will the benefits to society also equal the individual gain for the producer who sells it. This can only occur if the 'marginal revenue' for producing this last item sold equals its price.

Only perfect compet.i.tion guarantees this outcome, because, economists believe, only then does marginal revenue always equal price.

Perfect compet.i.tion is also 'perfect' because a supply curve exists if, and only if, price equals marginal cost. Without perfect compet.i.tion, though a marginal cost curve can still be drawn, this will not be the supply curve, and as we shall see, the amount supplied to the market will be less than the amount that will maximize social welfare.

This concept of economic perfection relies upon downward-sloping market demand curves, which we already know is invalid. However, even if we accept, for the sake of argument, that the market demand curve is smoothly downward sloping and represents community welfare, the neocla.s.sical argument for the superiority of the perfectly compet.i.tive market over the monopoly firm is still internally flawed. To establish this, we'll first consider the market form least favored by economics: monopoly.3 Monopoly.

A monopoly has the entire market demand curve to itself. If the market demand curve is smoothly downward sloping, the price at which its output can be sold decreases as the quant.i.ty it tries to sell increases. In this chapter I'll work with a hypothetical example in which the market price is a.s.sumed to start at $1,000 for the first unit sold, and then to drop by five cents for every additional unit (see Table 4.1).

TABLE 4.1 Demand schedule for a hypothetical monopoly.

This may seem silly if you've never read an economics textbook before why not simply use some real data on a real firm instead? and you are right! The reason, as I explain in Chapter 5, is that there are no such data: the revenue and costs of real firms are nothing like those a.s.sumed by neocla.s.sical economists. As a result, they always use made-up number in their examples. To critique their theory, I have to do the same. So here we go ...

Since the firm can sell one unit of output for $1,000.00, its total revenue is $1,000.00, and its 'marginal revenue' the change in total revenue from zero dollars for zero units sold, to $1,000.00 for one unit sold is also $1,000.00. So price equals marginal revenue at this level; but as soon as another unit is sold, price and marginal revenue diverge. Two units can only be sold if the firm drops the price for all units by 5 cents, so the market price becomes $999.95, the total revenue is $1,999.90, and the marginal revenue for the firm the change in revenue from selling one unit to selling two is $999.90.4 The interests of the firm therefore diverge from those of society, since the marginal benefit to it (the marginal revenue) is less than the marginal benefit to society as a whole (the price).

The price consumers are willing to pay drops smoothly as the quant.i.ty supplied rises, so that for an output of 10 units, the sale price has to drop to $999.55 per unit. The total revenue for selling 10 units is $9,995.50. If 11 units were to be sold, the monopolist would have to drop the price per unit by 5 cents, to $999.50 each. Total revenue would be $10,994.50 (eleven times $999.50), and marginal revenue would be $999.00.

The same process continues indefinitely, so that if output were 2,001 units, then sale price would have to drop to $900. Total revenue would be $1,800,900, and marginal revenue the amount of additional revenue added by selling the 2,002nd unit would be $800.00.

Eventually, the point is reached at which any further increase in output requires a price cut which reduces, rather than increases, total revenue. In this example, this occurs at an output of 10,001 units, where the sale price is $500. The sale of the 10,001st unit adds nothing to total revenue, and any increase in sales past this point actually reduces total revenue marginal revenue has become negative.

That covers the revenue side of the a.n.a.lysis. The picture is completed by the a.n.a.lysis of costs, which I'll cover extensively in Chapter 5. Briefly, the firm has two types of costs: fixed costs, which apply no matter what the level of output is, and variable costs, which depend directly on how many units are produced.

Fixed costs are just that fixed so that the fixed cost per unit of output will fall as output rises. One fixed cost is the design of a product, and if this was, say, $10 million, then that component of the fixed costs per unit would be $1 million per unit when output was 10 units, and $1 per unit when output was 10 million units.

Variable costs depend on how many units are produced. One obvious variable cost is labor, and clearly you will need more labor to produce 10 million units than to produce 10. Neocla.s.sical economics also a.s.sumes that, eventually, the productivity of the variable inputs such as labor will fall as output rises (we explore this a.s.sumption in Chapter 5). Therefore the variable costs to produce the 10 millionth unit will be much higher than those for the 10th unit. In my example, fixed costs are $10,000, and variable costs are defined by an equation in which they start at just over $15 each, fall for a while but then ultimately rise (see Table 4.2).5 Variable costs fall for a while because the firm experiences 'rising marginal productivity' as the ratio of the variable factors of production to fixed factors approaches the ideal level. This means that, for a while, the additional cost involved in producing the next unit of output falls. In my example, while it cost an additional $15 to go from producing zero units of output to producing one unit, it cost only an additional $8.80 to go from producing 2,001 units to 2,002 units.

This change in the cost of production resulting from producing one more unit is a very important concept in neocla.s.sical economics, called the 'marginal cost of production.' As you can see from this example, marginal cost depends only on the change in variable costs since fixed costs are the same no matter what level of output you produce and it changes only because of changes in productivity that in turn reflect how many variable inputs are being used (workers) relative to the fixed inputs (machines).

TABLE 4.2 Costs for a hypothetical monopoly.

Common sense, and earlier theories of economics like Ricardo's theory of rent, might consider that maybe the productivity of the individual inputs changes. Ricardo, for example, a.s.sumed that the cost of producing food rose as population rose because farmers started off using the most productive land, and had to use less fertile land as population increased. Common sense might suggest that as a firm demands more workers, it affects the wage at which workers can be hired, thus driving its costs per worker higher.

But neocla.s.sical economists rule both these effects out, by a.s.suming first that all inputs are h.o.m.ogeneous, and secondly that, while the monopoly has its own market to itself, it is a small player in the labor market and can hire as many workers as it likes at the going wage. The only source of changes in marginal cost that they allow arises from changing the ratio of variable inputs to the fixed inputs.

Consider a road construction firm, whose fixed costs include a number of jackhammers say 100 of them. At a very low level of production, it will have only one worker and 100 jackhammers, so the worker will be very inefficient (please read the footnote here).6 However, as the number of workers rises the firm will approach the ideal ratio of one worker per jackhammer, at which point maximum efficiency will be reached. But once the firm hits the ideal ratio, additional workers will add to output at a diminishing rate. Marginal productivity will fall, and therefore marginal costs will rise.

Table 4.3 combines the revenue information from Table 4.1 with the cost information from Table 4.2, and indicates the role of marginal revenue and marginal cost in identifying the point of maximum profit. For a while, each additional unit sold adds much more to revenue than it causes the total cost of production to rise: marginal revenue exceeds marginal cost, and therefore the final column in the table, which shows marginal revenue minus cost, is positive. But once marginal revenue and marginal cost are equal, profit is maximized.

The precise point at which this occurs lies between 8,973 and 8,974 units in this table, but the firm can't sell a fraction of a unit, so it will produce the lower amount of 8,973 units, at which the marginal cost is $102.77 and its profit will be $3,671,679.

The second column tells us that the market is willing to pay a price of $551.40 per unit if total supply is 8,973 units so the sale price is $448.63 higher than the marginal cost of production (and $409.19 above the average cost).7 Thus to maximize its profits, the firm produces where marginal cost equals marginal revenue, and sells the output at a much higher price.

As well as substantially exceeding the average cost of production, the market price exceeds the marginal cost of producing the last unit sold. This means, in economic welfare terms, that the marginal benefit of the last unit sold exceeds the marginal cost of producing it. Society would therefore benefit from an increased level of production, since additional units of output would increase social welfare. But the monopolist has no incentive to produce more: in fact producing any more would reduce his profits. Therefore, according to economists, monopolies reduce social welfare.

4.3 Profit maximization for a monopolist: marginal cost equals marginal revenue, while price exceeds marginal cost Crucially for the way neocla.s.sical economists prefer to model the economy, a supply curve can't be derived for a monopoly. Instead, if monopolies were the rule, then three curves price, marginal revenue, and marginal cost would be needed for a complete 'Totem of the Micro.' The intersection of the marginal revenue curve with the marginal cost curve would determine the amount the firm produced, and the market price would then depend on this quant.i.ty. In place of the simple mantra that 'prices are set by supply and demand,' the minimum statement of the 'Creed of the Micro' would be 'price is set by the demand curve, given the quant.i.ty set by marginal cost and marginal revenue.'

TABLE 4.3 Sales and costs determine the level of output that maximizes profit.

It's no wonder, then, that, despite all the criticisms leveled at it, neocla.s.sical economists cling to the model of the 'perfect' compet.i.tive market. In a compet.i.tive market, since marginal revenue equals price, profit-maximizing behavior leads to an output level at which price equals marginal cost. This is the embodiment of Smith's 'invisible hand' metaphor about the capacity of market economy to reconcile private interest and public virtue, and that is the real message of the 'Totem of the Micro.'8 Perfect compet.i.tion.

The main distinguishing feature of the perfectly compet.i.tive market is the number of firms in it. Whereas a monopoly has just one firm which therefore has the entire market demand curve to itself a perfectly compet.i.tive market has many little firms, each competing for a tiny slice of total demand.

In the standard 'Marshallian' model that economists teach in undergraduate courses, these firms are a.s.sumed to be profit maximizers who behave in an 'atomistic' way: they neither know of, nor react in any way to, what other firms do or may hypothetically do they simply respond to the market price.9 In addition, it is a.s.sumed that entry into and exit from a compet.i.tive industry is 'free,' or more accurately, not subject to any barriers. Therefore firms outside the industry can move in at any time to take advantage of any above-normal profits if they exist.10 All firms are a.s.sumed to produce a product that is h.o.m.ogeneous from the consumers' point of view, so that there is no brand loyalty. All firms are therefore 'price-takers': they cannot influence the market price, but instead must take price as given.

At the market level, demand is still a negative function of price. Therefore, total market revenue will initially be a rising and then a falling function of price, and marginal revenue at the market level will be less than price (because to increase overall sales, the average price must fall).

However, economists argue that for each price-taking firm, marginal revenue and price are identical. The argument is that since they are each so small, no single firm can influence the market price. As a result, if any firm increases its price above the market equilibrium price, it will lose all its customers; while if any firm decreases its price below the market equilibrium, it will suddenly be swamped by all customers for that commodity. Therefore, the firm effectively sees a horizontal demand curve (set by the intersection of supply and demand at the level of the market).11 Given the a.s.sumption that they can sell as much as they like at the price set by the market, then as profit maximizers they will produce until the marginal cost of producing this amount equals the marginal revenue from doing so. Since price is a constant for them, marginal revenue equals price, so they produce at the point where marginal cost equals price. In a 100-firm industry whose costs are identical to the monopoly I discussed previously, this results in the representative firm producing about 135 units.12 This then results in a profit of $22,255.26 for the firm, or $2,225,526 dollars for the industry in total.

4.4 Profit maximization for a perfectly compet.i.tive firm: marginal cost equals marginal revenue, which also equals price Since the total revenue for a perfectly compet.i.tive firm is simply a constant price times the number of units it sells, increasing its sales has no effect on its price, so its marginal revenue is constant. This in turn is why a supply curve can be derived for perfect compet.i.tion, but not for monopoly.

The amount a monopolist will supply depends both on the firm's marginal cost function, and the market's demand function. Since both are needed to determine supply, and since many different demand curves can be drawn through the same point each with a different slope and therefore different marginal revenue implications it is impossible to derive a curve which shows how much a monopolist will supply at each price level (all you can do is consider specific examples of hypothetical demand curves, as I did to generate Table 1).