However, for the perfectly compet.i.tive firm, since price equals marginal revenue, the amount the firm will produce corresponds in every case to its marginal cost curve. The supply curve of a single firm in a perfectly compet.i.tive market is thus its marginal cost curve.

4.5 A supply curve can be derived for a compet.i.tive firm, but not for a monopoly The supply curve for a perfectly compet.i.tive industry is constructed simply by adding up the amounts that each firm is willing to supply at a given price. This amounts to summing up their marginal cost curves, so that the supply curve for the industry represents the marginal cost of producing output. Since demand equals supply in equilibrium, the marginal benefit for the last unit consumed equals its marginal cost of production, and social utility is maximized. This results in both a higher level of output and a lower price than would occur if the industry were a monopoly.

4.6 A compet.i.tive industry produces a higher output at a lower cost than a monopoly Checking our sums.

This argument normally convinces economics students, and it explains much of the hostility economists in general have towards monopolies, or any market in which firms have some market power by virtue of their size. This 'social radicalism' is unusual for a profession which is normally perceived as socially conservative. It is also curiously at odds with the real world, where it's fairly obvious that industries have a clear tendency to end up being dominated by a few large firms why fight the real world?

Economists argue that their opposition to large firms, and their allegedly uncharacteristic radicalism on this issue, is based on sound a.n.a.lysis. But is it? Let's check, after first seeing how moving from a monopoly to a perfectly compet.i.tive industry would benefit society in my example.

Table 4.4 adds up the costs and revenues of all the compet.i.tive firms, to show the aggregate outcome for a compet.i.tive industry with 100 firms. Note that the output of this industry (in the rows shown in italic) is higher than the monopoly's output roughly 13,456 units, versus 8,973 and its price is lower roughly $327.25 per unit, versus $551.40 for the monopoly.

Economists therefore put forward three reasons to prefer a compet.i.tive industry to a monopoly: the compet.i.tive industry produces where marginal cost equals price, thus maximizing social welfare; it produces a higher level of output than a monopoly; and it sells this higher output at a lower price.

However, the key reason why neocla.s.sical economists themselves prefer perfect compet.i.tion to monopoly is that perfect compet.i.tion is the only market structure in which price and quant.i.ty are set by the intersection of the supply curve and the demand curve.

4.7 The standard 'supply and demand' explanation for price determination is valid only in perfect compet.i.tion Well, that's the theory. Now we will consider a subtle but profound set of problems which invalidate this entire a.n.a.lysis.

Calculus 101 for economists: infinitesmals ain't zero Throughout the economic a.n.a.lysis of perfect compet.i.tion, the a.s.sumption is made that the perfectly compet.i.tive firm is so small, relative to the overall market, that its impact on the market can be treated as zero. As I intimated earlier in this chapter, this kind of logic is OK when you are dealing with local approximations such as whether you can regard the ground on which you stand as either flat or curved but it will not do when those local approximations are aggregated together. When we insist that infinitesimally small amounts are not in fact zero, the apparently watertight logic behind the comparison of monopoly and perfect compet.i.tion falls apart.

TABLE 4.4 Cost and revenue for a 'perfectly compet.i.tive' industry identical in scale to hypothetical monopoly Too small to matter? An essential part of the argument for perfect compet.i.tion is that each firm is so small that it can't affect the market price which it therefore takes as given. Consequently the demand curve, as perceived by each firm, is effectively horizontal at the market price. The firms are also so small that they do not react to any changes in behavior by other firms: in the language of economic theory, their 'conjectural variation' how much all other firms change their output in response to a change in output by one firm is zero.

These two a.s.sumptions are alleged to mean that the slope of the individual firm's demand curve is zero: both the firm's price and the market price do not change when a single firm changes its output. However, they also mean that, if a single firm increases its output by one unit, then total industry output should also increase by one unit since other firms won't react to the change in output by a single firm.

However, there is a problem: these two a.s.sumptions are inconsistent.

If the market demand curve is downward sloping, then an increase in total market output must mean a fall in the market price regardless of how small a fall it might be. Since the theory a.s.sumes that other firms don't react to an increase in production by one firm, total market output must increase. Since the market demand curve is downward sloping, and supply has increased the supply curve has shifted outwards market price must fall.

Therefore market price does change because of the actions of a single firm. The only way market price could not react would be if all other firms reduced their output by as much as the single firm increased it: then the market supply curve would not shift, and the price would remain constant. But the theory a.s.sumes that firms don't react to each other's behavior.

So the market price will be affected by the actions of a single firm, in which case the demand curve facing a single firm will be downward sloping however slight the slope may be.

Putting this critique another way, the economic argument is that if you break a large downward-sloping line (the market demand curve) into lots of very small lines (the demand curves perceived by each firm), then you will have a huge number of perfectly flat lines. Then if you add all these perfectly flat lines together again, you will get one downward-sloping line.

This is mathematically impossible. If you add up a huge number of flat lines, you will get one very long flat line. If you break one downward-sloping line into many small lines, you will have many downward-sloping lines. The economic concept of perfect compet.i.tion is based on a mathematical error of confusing a very small quant.i.ty with zero.

The market matters: marginal revenue is marginal revenue A second problem with this economic model is the nature of the marginal revenue function. Economists unconsciously reason as if the marginal revenue curve at the market level is a function of the number of firms that produce the industry's output: it exists if there is only one firm, but if there are a large number of firms, it disappears. They then show that a monopoly sets its price where marginal revenue equals marginal cost, which is consistent with their theory. However, they show a compet.i.tive industry setting price where the supply curve and the demand curve intersect, with no pesky marginal revenue curve getting in the way.

Unfortunately, marginal revenue exists independently of the number of firms in the industry. If the market demand curve is downward sloping, then so is the market marginal revenue curve, and they diverge right from the very first unit sold (as you can see in the example).

So if a compet.i.tive industry did result in output being set by the intersection of the demand curve and the supply curve, then at the collective level the compet.i.tive industry must be producing where marginal cost exceeds marginal revenue. Rather than maximizing profits, as economists argue firms do, the additional output that produced past the point where marginal revenue equals marginal cost at the industry level must be produced at a loss. This paradox means that the individual firm and the market level aspects of the model of perfect compet.i.tion are inconsistent.

Creative accounting For the a.s.sertion that perfect compet.i.tion results in a higher level of output at a lower price than monopoly to be correct, then in the aggregate, the individually rational profit-maximizing behavior of perfectly compet.i.tive firms must lead to a collectively irrational outcome. This would be OK if the theory actually admitted this as do the theories of Cournot and Bertrand compet.i.tion13 but the Marshallian model taught to undergraduates claims instead that equating marginal cost and marginal revenue maximizes profits for the compet.i.tive firm.

According to the theory, the monopoly firm produces only to the point at which marginal cost equals marginal revenue, because this output maximizes its profit. Each perfectly compet.i.tive firm likewise produces to a point at which its marginal cost equals its marginal revenue, and for the same reason because this level of output maximizes its profit.

But at the market level, compet.i.tive firms produce to a point at which the collective marginal cost exceeds marginal revenue. The perfectly compet.i.tive industry produces where marginal cost equals price but exceeds marginal revenue; yet all firms in it are supposed to be producing where marginal cost equals marginal revenue.

The monopoly sets price where marginal revenue equals marginal cost, while the compet.i.tive industry sets price where the supply curve (which is the sum of all the individual firms' marginal cost curves) intersects the demand curve: this is supposed to be the result of setting marginal cost equal to marginal revenue at the firm level, which means each firm makes the maximum profit that it can. Yet at the aggregate level, while the monopoly has produced where profit is maximized, the compet.i.tive firms have produced beyond this point, so that the industry's output past the point of monopoly output has been produced at a loss which is why the profit level for the compet.i.tive firm is lower than that for the monopoly, even though all its firms are supposed to be profit maximizers.

Where did this loss come from? It certainly can't be seen in the standard graph economists draw for perfect compet.i.tion, which shows the individual compet.i.tive firm making profits all the way out to the last item produced.

Instead, this 'loss out of nowhere' is hidden in the detail that economists lose by treating infinitesimally small quant.i.ties as zeros. If perfectly compet.i.tive firms were to produce where marginal cost equals price, then they would be producing part of their output past the point at which marginal revenue equals marginal cost. They would therefore make a loss on these additional units of output.

As I argued above, the demand curve for a single firm cannot be horizontal it must slope downwards, because if it doesn't, then the market demand curve has to be horizontal. Therefore, marginal revenue will be less than price for the individual firm. However, by arguing that an infinitesimal segment of the market demand is effectively horizontal, economists have treated this loss as zero. Summing zero losses over all firms means zero losses in the aggregate. But this is not consistent with their vision of the output and price levels of the perfectly compet.i.tive industry.

The higher level of output must mean losses are incurred by the industry, relative to the profit-maximizing level chosen by the monopoly. Losses at the market level must mean losses at the individual firm level yet these are presumed to be zero by economic a.n.a.lysis, because it erroneously a.s.sumes that the perfectly compet.i.tive firm faces a horizontal demand curve.

Perfect compet.i.tion equals monopoly The above critique raises an interesting question: what will the price and output of a perfectly compet.i.tive industry be, if we drop the invalid a.s.sumption that the output of a single firm has no effect on the market price? The answer is: the price and output levels of a compet.i.tive industry will be exactly the same as for the monopolist (if the aggregate marginal cost curve of the compet.i.tive firms is identical to the marginal cost of the monopoly, which economic theory a.s.sumes it is).

Economic explanations of price-setting in a compet.i.tive market normally start from the level of the market, where they show that the intersection of supply and demand sets both price and quant.i.ty. They then argue that the price set by this intersection of supply and demand is taken as given by each compet.i.tive firm, so that the supply curve for the individual firm is its marginal cost curve. Then they notionally add all these marginal cost curves up, to get the supply curve for the industry and its point of intersection with the demand curve determines the market price.

But there is a 'chicken and egg' problem here. Which comes first price being set by the intersection of supply and demand, or individual firms equating marginal cost to price? And why should a level of output which involves making a loss on part of output (the part past where market marginal revenue equals marginal cost) determine where each individual firm perceives price as being set?

Economists have been bewitched by their own totem. They draw a downward-sloping market demand curve, and an upward-sloping supply curve, and a.s.sume that price and quant.i.ty must be set by the intersection of the two curves. But the 'supply curve' is really only the aggregate of the marginal cost curves of all the compet.i.tive firms. It isn't a supply curve unless they can prove that, whatever the market demand curve looks like, the industry will supply the quant.i.ty given by the intersection of the demand curve and the aggregate marginal cost curve.

This isn't the case in their own model of monopoly. The intersection of marginal cost and marginal revenue determines the quant.i.ty produced, while the price charged is set by the price the demand curve gives for that quant.i.ty: price and quant.i.ty are not determined by the intersection of the demand curve and the marginal cost curve.

Economists claim that price and quant.i.ty are set by the intersection of the demand curve and the aggregate marginal cost curve in the case of perfect compet.i.tion, but their 'proof' relies on the erroneous proposition that the demand curve perceived by each individual firm is, you guessed it, horizontal.

Once this spurious proposition is removed, the price that the compet.i.tive firm takes as given is the price determined by the intersection of the market demand curve and the aggregate marginal cost curve which is precisely the same price as a monopoly would charge. To argue otherwise is to argue for either irrational behavior at the level of the individual firm so that part of output is produced at a loss or that, somehow, individually rational behavior (maximizing profit) leads to collectively irrational behavior so that profit-maximizing behavior by each individual firm leads to the industry somehow producing part of its output at a loss. However, the essence of the neocla.s.sical vision is that individually rational behavior leads to collectively rational behavior.

Therefore, the price that the perfectly compet.i.tive firm will take as given when it adjusts its output is not a market price set by equating price to marginal cost, but a market price set by equating marginal revenue to marginal cost. The quant.i.ty produced at this price will be equivalent, when summed, to the output of a single monopolist. On the grounds of properly amended economic theory, monopoly and perfect compet.i.tion are identical.

Returns to scale and the durability of perfect compet.i.tion To date I have accepted the a.s.sumption that a monopoly has no scale advantages over a perfectly compet.i.tive firm, so that it is possible to sum the cost functions of numerous small firms and come up with aggregate costs similar to those of a large firm.

In general, this a.s.sumption of scale-invariant costs will be invalid. If we are simply considering the costs of producing a h.o.m.ogeneous product, then it is likely that a very large firm will have scale advantages over a very small one. In the vernacular of economics, large firms benefit from returns to scale.

Returns to scale occur when the cost of production rises less rapidly than the output as the scale of production increases. A simple example is in farming, where farms need to be separated from each other by fences. The amount of fencing required depends on the perimeter of the farm. If we consider a square block of land, fencing will depend on the length of the four sides of the square. Cost is thus the cost of fencing per mile, times four times the length of a side. But the area enclosed by the fence depends on the length of a side squared. The output of a farm is related to its area, so that output is a function of the length of a side squared. Doubling the perimeter of a farm thus doubles its fencing costs, but increases its output fourfold.

As a result, large farms have a scale advantage over smaller farms. A farm with a square mile of land requires four miles of perimeter fencing to each square mile, while a farm with four square miles of land requires eight miles of perimeter fencing or just two miles of perimeter fencing to each square mile of land.

4.8 Double the size, double the costs, but four times the output The same concept applies in numerous ways. For a substantial range of output, a large blast furnace will be more cost effective than a smaller one, a large ship than a smaller one, a large car factory than a smaller one.

If large firms have cost advantages over small ones, then given open compet.i.tion, the large firms will drive the small ones out of business (though marketing and debt problems will limit the process, as Sraffa notes). Hence increasing returns to scale mean that the perfectly compet.i.tive market is unstable: it will, in time, break down to a situation of either oligopoly (several large firms) or monopoly (one large firm).

Economists have been well aware of this dilemma since Marshall at the end of the nineteenth century, and the fiction that has been invented to cope with it is the concept of the long run average cost curve.14 The curve is 'u-shaped,' which a.s.serts that there is some ideal scale of output at which the cost of production is minimized. In the long run, all inputs can be varied, so this shape is supposed to represent increasing returns to scale up to the point of minimum cost, beyond which decreasing returns to scale start to occur, so that the cost of production rises.

A compet.i.tive industry is supposed to converge to this ideal scale of output over time, in which case its many extremely big firms are safe from the predations of any much larger firm, since such a compet.i.tor would necessarily have higher costs.

This defence is specious on several counts.

First, the question of whether perfect compet.i.tion can exist in a particular industry becomes an empirical one: what is the ideal scale of output, and how many firms could then occupy a particular industry at a particular time?

For some industries, the answer might well be 'very many' the ever popular wheat farm comes to mind. However, for some other industries, the answer might well be 'very few.' It seems, for example, that the worldwide market for large intercontinental pa.s.senger airplanes can support at most three firms.

The argument that, 'in the long run,' this industry could be perfectly compet.i.tive because it could grow big enough to support hundreds or thousands of compet.i.tors is ludicrous. By the time the world was large enough to support hundreds of Boeings and Airbuses, it is highly likely that some entirely different form of transport would have superseded the airplane.

Secondly, the long run supply curve is actually constructed under the a.s.sumption of constant technology: in other words, it is not really a concept in time at all. The scale economies are supposedly there all the time, ready to be exploited.

If so, then unless an industry is already big enough to support the enormous number of firms surmised by the model of perfect compet.i.tion all operating at the ideal scale large firms can immediately out-compete small firms. In other words, the only way compet.i.tive firms can survive is if the industry is already so large that it can support an enormous number of firms of the ideal scale.

The theoretical response of economists to this dilemma has been to presume constant returns to scale. With constant returns, 'size does not matter': a small firm will be just as cost efficient as a large one.

Unfortunately, size does matter. Economies of scale are an important part of the reason that most industries are dominated by a small number of very large firms. We do need an adequate a.n.a.lysis of how such an industry functions, but neocla.s.sical economics does not provide it.

Addendum: the war over perfect compet.i.tion.

As noted, my plan to start work on Finance and Economic Breakdown (a book-length treatment of Minsky's 'Financial Instability Hypothesis') when I finished Debunking Economics was derailed for the next four years as I found myself embroiled in disputes with neocla.s.sical economists via email, in web forums, in public and in referee comments on my papers about this argument. The end result was a substantial strengthening of the critique, the most important component of which was a proof that equating marginal cost and marginal revenue does not maximize profits.

I developed this proof after realizing that the key result in this chapter that the demand curve for a compet.i.tive firm cannot be horizontal was discovered by the neocla.s.sical economist George Stigler over half a century ago. Why, I wondered, did he nonetheless continue to subscribe to and defend neocla.s.sical theory?

Apart from the usual psychological explanation that when you've committed yourself to a particular belief system and made your reputation in it, it is extraordinarily hard to accept that it might be false there is a technical reason in the same paper. Though he proved that the individual firm's demand curve had the same negative slope as the market demand curve, Stigler also proved that, if firms produced where marginal cost equaled marginal revenue, then the more firms there were in an industry, the closer industry output would be to where price equaled marginal cost.

It is intuitively plausible that with infinite numbers all monopoly power (and indeterminacy) will vanish [...] But a simple demonstration, in case of sellers of equal size, would amount only to showing that Marginal revenue = Price + Price/Number of sellers [times] Market elasticity, and that this last term goes to zero as the number of sellers increases indefinitely. (Stigler 1957: 8) Stigler thus believed that he had neutralized his finding in the same paper. Yes, the conventional neocla.s.sical belief that the individual compet.i.tive firm faces a horizontal demand curve is false, but if there are a large number of firms in an industry, then marginal revenue for the individual firm will be very close to the market price. Therefore the collective effect is the same: price will be set where supply equals demand. The key result of compet.i.tion is restored.

From this point on, the standard failings of neocla.s.sical research and pedagogy took over. Only a minority of economists read the paper; textbooks continued to teach the concept that Stigler had disproved; and the minority of economists who were aware of Stigler's paper defended the failure to take his result seriously because, in the end, the outcome was alleged to be the same: supply will equal demand in a compet.i.tive market.

I instead saw a logical error: Stigler's proof that marginal revenue for the individual firm would converge to market price as the number of firms increased was correct, if those firms all set marginal revenue equal to marginal price. But all the problems that I had identified in this chapter still remained: in particular, producing where supply equaled demand required 'profit-maximizing' firms to actually make losses on all goods sold past the point at which industry-level marginal revenue equaled marginal cost.

There was only one explanation: equating marginal cost and marginal revenue couldn't be profit-maximizing behavior.

I followed the logic forward and proved that the true profit-maximizing formula was quite different. If compet.i.tive firms did actually profit-maximize, they would produce an output much lower than the level where marginal cost equaled marginal revenue. The market outcome was that a compet.i.tive industry would produce the same amount as a monopoly, and market price would exceed marginal cost.

Equating marginal cost and marginal revenue does not maximize profits.

The logic is fairly simple to follow if you imagine that you are running a compet.i.tive firm, and ask yourself this question: 'Is my level of output the only factor that can affect my profits?' The answer is, of course not: your profit depends not just on how much you produce, but also how much all the other firms in the industry produce. This is true even if you can't control what other firms do, and even if you don't try to react to what you think they might do. You work in a multi-firm industry, and the actions of all other firms impinge upon your own profits.

However, the neocla.s.sical 'profit-maximizing' formula implies that your output is the only factor determining your profits: it uses simple calculus to advise you to produce where the change in your profits relative to your own output is zero. What you really need to do if you're going to try to use calculus to work out what to do is to work out where the change in your profits relative to total industry output is zero.

Intuitively, this is likely to mean that the actual amount you produce which is something you can control should be less than the amount the neocla.s.sical formula recommends. This is because it's highly likely that the impact on your profit of changes in output by other firms which you can't control will be negative: if other firms increase their output, your profit is likely to fall. So when you work out the impact that changes in output by other firms has on your profits, the sign of this change is likely to be negative.

Therefore, to find the point at which your profit is at a maximum with respect to total industry output, you're likely to want the sign for the impact of your changes in output on your profit to be positive. This will mean that your output level will be less than the level at which your marginal cost equals your marginal revenue.

The best way to solve this problem precisely is to work out when what is known as the 'total differential' of the firm's profit is equal to zero (to avoid using symbolic terms like 'n' for the number of firms in the industry, I'll work with a hypothetical industry with 1,000 firms in it, but the logic applies independently of the number of firms in the industry).

The profit of your firm will be its revenue which will equal your firm's output times the market price minus costs. What we have to do is work out how these two aspects of profit are influenced by the changes in output by all the firms in the industry, including your own.

Using a calculus procedure known as the Product Rule, the change in the revenue side of this calculation can be broken down into two bits: your output, times how much a given firm's change in output changes market price; plus market price, times how much a given firm's change in output causes you to alter your own output.

Thanks to Stigler's accurate calculus from 1957, we know that we can subst.i.tute the slope of the market demand curve for 'how much a given firm's change in output changes market price,' so the first term in the change in revenue calculation becomes your firm's output, times the slope of the market demand curve. With 1,000 firms in the industry, we get 1,000 copies of this term, which is your firm's output, multiplied by the slope of the market demand curve.

The second term in the change in revenue is the market price, times the amount that your output changes owing to a change in output by a given firm. Since we're working with the Marshallian model, which a.s.sumes that firms don't react strategically to what other firms do, then 999 times out of 1,000 this term will be the market price times zero. But once, it will be how much your output changes, given a change in your output. The ratio of the change in your output to the change in your output is one, so once and only once this calculation will return the market price.

Finally, we have to consider the cost side of the calculation: this will be how much your total costs change, given a change in output by a given firm. As with the last calculation for revenue, 999 times out of 1,000 this will be zero because your costs don't change when the output of another firm changes. But once, and only once, it will be how much your total costs change, given a change in your output. This is your firm's marginal cost.

That gives you three terms, and when the output level you choose causes the sum of these three to be zero, you have identified the output level for your firm that will maximize your profits. These three terms are: the market price (a positive number); plus the slope of the market demand curve multiplied by 1,000 times your output (a negative number, since the slope of the market demand curve is negative); minus your marginal cost.

The difference between this formula and the neocla.s.sical formula is subtle and the size of an elephant at the same time. The neocla.s.sical formula tells you that you maximize your profits when these three terms are equal: the market price; plus the slope of the market demand curve multiplied by your output; minus your marginal cost.

Whoops! The neocla.s.sical formula has erroneously omitted 999 times your output times the slope of the market demand curve a very large negative term (since the slope of the market demand curve is negative). It therefore takes a larger marginal cost to reduce the whole neocla.s.sical expression to zero, which you can only achieve by producing a higher output. All of this additional output will be sold at a loss: the increase in revenue you get from selling those additional units will be less than the increase in costs the additional production causes.

The neocla.s.sical formula is thus categorically wrong about the level of output by the individual firm that will maximize its profits except in the one case of a monopoly, where the two formulas coincide.

If compet.i.tive firms are truly profit maximizers, then they will produce substantially less output each than neocla.s.sical theory says they will (roughly half as much), and the sum of this output will if they face identical costs of production be the same as would be produced by a monopoly.

It could be argued that the accurate formula derived above requires the firm to know something that it can't possibly know which is how many firms there are in the industry. In fact, this is less of a problem than it seems, because it's possible to reorganize this formula into a form in which the number of firms in the industry isn't that important.15 But a more important point is that in reality firms don't 'do calculus.' They are far more likely to work out the answer to this and other questions by trial and error.

Calculus schmalculus.

What firms would actually do is work out the ideal amount to produce to maximize profits by choosing some output level at random, and then vary this amount to see what happens to their profits. If a firm's profit rose, then it would continue altering its output in the same direction; but if its profit fell, then it would reverse direction.

Unfortunately we can't test this using empirical data, because, as I argue later, the a.s.sumptions of the neocla.s.sical model (a falling demand curve, a rising supply curve, and a static setting in which to maximize profits) don't even come close to describing reality. But one can today create an artificial market using a computer model that does fit the neocla.s.sical a.s.sumptions, and then see what happens.

The next few graphs show the results of this simulation: firms choose an initial output level at random; the initial market price is determined by the sum of these randomly chosen outputs; each firm then chooses a random amount to vary its output by, and changes its initial output by this amount; a new market price is calculated; if a firm's profit has risen as a result of the change in output, it continues changing its output in the same direction; otherwise it reverses direction.

At the extremes considered here, of a monopoly and a 100-firm industry, neocla.s.sical theory is correct for a monopoly, but very wrong for the 100-firm industry. It predicts that such an industry will produce effectively where marginal cost equals price where the 'supply curve' intersects the demand curve but in practice the 100-firm industry produces an output that is almost the same as the monopoly's.

4.9 Predictions of the models and results at the market level Neocla.s.sical theory also predicts that industry output will converge to the compet.i.tive ideal as the number of firms in the industry rises. Simulations with between 1 and 100 firms in the industry show no pattern, though in general the output level is well below that predicted by neocla.s.sical theory but close to the prediction of my equation (see Figure 4.12).

4.10 Output behavior of three randomly selected firms This market outcome is not caused by collusion, but is simply the result of profit-maximizing behavior. Firms also follow very different paths in their output, even though the basic 'strategy' is the same for each firm: vary output and try to find the output level that generates the largest profit.

4.11 Profit outcomes for three randomly selected firms Firms have very different outcomes with respect to profit as well, though in general most make far more profit from this 'suck it and see' algorithm than they would make if they followed the neocla.s.sical formula.

Many different outcomes are possible with different a.s.sumptions in particular, the introduction of some irrationality by firms (continuing to increase output when the last increase in output reduced profit, for example), or a greater dispersal in the size of the changes in output by firms causes the aggregate result to move in the direction of the neocla.s.sical formula (Keen and Standish 2010: 6974). But the neocla.s.sical proposition that strictly rational behavior leads to a compet.i.tive industry producing where individual marginal revenue equals marginal cost is strictly false.

4.12 Output levels for between 1- and 100-firm industries.

Dialogue with the deaf.

There are other theories of compet.i.tion (the Cournot-Nash and Bertrand models of game-theoretic behavior, where firms do react strategically to what other firms might do), where a 'perfectly compet.i.tive' outcome can occur from non-profit-maximizing behavior. But the standard Marshallian model of the firm is categorically false: the demand curve for a compet.i.tive firm is not horizontal, equating marginal cost and marginal revenue does not maximize profits, and a compet.i.tive industry will produce the same amount of output as a monopoly and sell it for the same price. The 'Marshallian' model of the compet.i.tive firm is dead.

Or it should be. Instead, given the resistance of neocla.s.sical economics to criticism, this false model is likely to live on for decades. Though this critique has been published in a range of journals including one edited by physicists, whose mathematical capabilities far exceed those of economists I have been unable to get it published in neocla.s.sical economics journals. The odds that this critique will ever be recognized by economics textbooks writers are therefore effectively zero.

Every manner of excuse has been offered to avoid confronting these uncomfortable but mathematically unimpeachable results. The most remarkable excuses came from referees for the Economic Journal and the Journal of Economics Education.16 A referee for the former journal admitted that this result was significant, but argued that it did not matter because, he alleged, the conventional theory a.s.sumed that firms attempted to maximize their profits while a.s.suming that the output of other firms was fixed. This alleged a.s.sumption cannot be found in any textbook on perfect compet.i.tion, and amounts to an a.s.sumption of irrational behavior on behalf of firms: 'Needless to say, this result is worthy of publication on Economic Journal if it is correct. However, after reading the paper, I am not convinced that this is the case. On the contrary I think the result is due to authors' confusion about an individual firm's rationality: maximizing its profit given others' outputs fixed [...]' (Referee, Economic Journal).

Though neocla.s.sical economics has always insisted that it is a mathematically based theory, a referee for the Journal of Economics Education refused to consider that one of the most basic procedures in calculus the Chain Rule could be applied in microeconomics: 'Stigler's many attempts to save neocla.s.sical theory have always caused more problems than they have solved. His version of the chain rule is contrary to the partial equilibrium method and thus is irrelevant' (Referee, Journal of Economics Education).

These and many other frankly irrational responses by editors and referees for other neocla.s.sical journals emphasize a frequent refrain in this book, that neocla.s.sical economics is far more a belief system than it is a science.

So what?

The main consequence of this critique for neocla.s.sical economics is that it removes one of the two essential pillars of their approach to modeling the economy. Unless perfect compet.i.tion rules, there is no supply curve.

This fact goes a long way to explaining why neocla.s.sical economists cling to using a notion that is so unrealistic, and so unlike any industry in the real world: because without it, their preferred method of modeling becomes impossible.

Economics has championed the notion that the best guarantee of social welfare is compet.i.tion, and perfect compet.i.tion has always been its ideal. The critiques in this chapter show that economic theory has no grounds whatsoever for preferring perfect compet.i.tion over monopoly. Both fail the economist's test of welfare, that marginal cost should be equated to price.

Worse, the goal of setting marginal cost equal to price is as elusive and unattainable as the Holy Grail. For this to apply at the market level, part of the output of firms must be produced at a loss. The social welfare ideal thus requires individual irrationality. This would not be a problem for some schools of economics, but it is for the neocla.s.sical school, which has always argued that the pursuit of individual self-interest would lead to the best, most rational outcome for all of society.