A perfectly competitive market may have several distinguishing characteristics, including:
Some subset of these conditions is presented in most textbooks as defining perfect competition. More advanced textbooks (e.g. Mas-Colell et al. 1995 p. 315) try to reconcile these conditions with the definition of perfect competition as equilibrium price taking; that is whether or not firms treat price as a parameter or a choice variable. It is this distinction which differentiates perfectly competitive markets from imperfectly competitive ones. It should be noted that a general rigorous proof that the above conditions indeed suffice to guarantee price taking is still lacking (Kreps 1990, p. 265). The importance of perfect competition derives from the fact that price taking by the firm guarantees that when firms maximize profits (by choosing quantity they wish to produce, and the combination of Factors of production to produce it with) the market price will be equal to marginal cost. An implication of this is that a factor's price (wage, rent, etc.) equals the factor's marginal revenue product. This allows for derivation of the supply curve on which the neoclassical approach is based (note that this is also the reason why "a monopoly does not have a supply curve"). The abandonment of price taking creates considerable difficulties to the demonstration of existence of a general equilibrium (Roberts and Sonnenschein 1977) except under other, very specific conditions such as that of monopolistic competition .
The notion of perfect competition derives from the earlier classical notion of free competition. A freely competitive market was defined by Adam Smith as one in which there is no impediment to free contracting and free entry and exit of productive resources; the implication was that all units of the same type of productive resource (be it a kind of labour, or a kind of land, or capital) would tend to earn the same reward as in other freely competitive markets, and as a result product prices would tend toward their natural levels i.e. toward the levels allowing paying labour, land and capital their natural rates (determined by the classical theory of income distribution). This is a long-period theory of product price, in which what is essential is free mobility of productive resources, in particular free entry of firms, and competition is viewed as a dynamical process akin to competition in a race (McNulty 1967). In such a conception, competition in a product market is the stronger, the closer the price of the product gets to its natural level (later called long-period normal price by Alfred Marshall).
Historically, in neoclassical economics there have been two strands of looking at what perfect competition is.
In the first the emphasis is on the inability of any one agent to affect prices. This is usually justified by the fact that any one firm or consumer is so small relative to the whole market that their presence or absence leaves the equilibrium price very nearly unaffected. This assumption of negligible impact of each agent on the equilibrium price has been formalized by Aumann (1964) by postulating a continuum of infinitesimal agents. The difference between Aumann’s approach and that found in undergraduate textbooks is that in the first, agents have the power to choose their own prices but do not individually affect the market price, while in the second it is simply assumed that agents treat prices as parameters. Both approaches lead to the same result.
The second view of perfect competition conceives of it in terms of agents taking advantage of – and hence, eliminating – profitable exchange opportunities. The faster this arbitrage takes place, the more competitive a market is. The implication is that the more competitive a market is under this definition, the faster the average market price will adjust so as to equate supply and demand (and also equate price to marginal costs). In this view “perfect” competition means that this adjustment takes place instantaneously. This is usually modeled via the use of the Walrasian auctioneer (see article for more information). The widespread recourse to the auctioneer tale appears to have favored an interpretation of perfect competition as meaning price taking always, i.e. also at non-equilibrium prices; but this is rejected e.g. by Arrow (1959) or Mas-Colell et al. (1995, p. 315)
Thus nowadays the dominant intuitive idea of the conditions justifying price taking and thus rendering a market perfectly competitive is an amalgam of several different notions, not all present, nor given equal weight, in all treatments. Besides product homogeneity and absence of collusion, the notion more generally associated with perfect competition is the negligibility of the size of agents, which makes them believe that they can sell as much of the good as they wish at the equilibrium price but nothing at a higher price (in particular, firms are described as each one of them facing a horizontal demand curve). However, also widely accepted as part of the notion of perfectly competitive market are perfect information about price distribution and very quick adjustments (whose joint operation establish the law of one price), to the point sometimes of identifying perfect competition with an essentially instantaneous reaching of equilibrium between supply and demand. Finally, the idea of free entry with free access to technology is also often listed as a characteristic of perfectly competitive markets, probably owing to a difficulty with abandoning completely the older conception of free competition. In recent decades it has been rediscovered that free entry can be a foundation of absence of market power, alternative to negligibility of agents (Novshek and Sonnenschein 1987.)
Free entry also makes it easier to justify the absence of collusion: any collusion by existing firms can be undermined by entry of new firms. The necessarily long-period nature of the analysis (entry requires time!) also allows a reconciliation of the horizontal demand curve facing each firm according to the theory, with the feeling of businessmen that "contrary to economic theory, sales are by no means unlimited at the current market price" (Arrow 1959 p. 49).
As mentioned above, the perfect competition model, if interpreted as applying also to short-period or very-short-period behaviour, is approximated only by markets of homogeneous products produced and purchased by very many sellers and buyers, usually organized markets for agricultural products or raw materials. In real-world markets, assumptions such as perfect information cannot be verified and are only approximated in organized double-auction markets where most agents wait and observe the behaviour of prices before deciding to exchange (but in the long-period interpretation perfect information is not necessary, the analysis only aims at determining the average around which market prices gravitate, and for gravitation to operate one does not need perfect information).
In the absence of externalities and public goods, perfectly competitive equilibria are Pareto-efficient, i.e. no improvement in the utility of a consumer is possible without a worsening of the utility of some other consumer. This is called the First Theorem of Welfare Economics. The basic reason is that no productive factor with a non-zero marginal product is left unutilized, and the units of each factor are so allocated as to yield the same indirect marginal utility in all uses, a basic efficiency condition (if this indirect marginal utility were higher in one use than in other ones, a Pareto improvement could be achieved by transferring a small amount of the factor to the use where it yields a higher marginal utility). A simple proof assuming differentiable utility functions and production functions is the following. Let wj be the 'price' (the rental) of a certain factor j, let MPj1 and MPj2 be its marginal product in the production of goods 1 and 2, and let p1 and p2 be these goods' prices. In equilibrium these prices must equal the respective marginal costs MC1 and MC2; remember that marginal cost equals factor 'price' divided by factor marginal productivity (because increasing the production of good i by one very small unit through increase of the employment of factor j requires increasing the factor employment by 1/MPji and thus increasing the cost by wj/MPji, and through the condition of cost minimization that marginal products must be proportional to factor 'prices' it can be shown that the cost increase is the same if the output increase is obtained by optimally varying all factors). Optimal factor employment by a price-taking firm requires equality of factor rental and factor marginal revenue product, wj=piMPji, so we obtain p1=MC1=wj/MPj1, p2=MCj2=wj/MPj2. Now choose any consumer purchasing both goods, and measure his utility in such units that in equilibrium his marginal utility of money (the increase in utility due to the last unit of money spent on each good), MU1/p1=MU2/p2, is 1. Then p1=MU1, p2=MU2. The indirect marginal utility of the factor is the increase in the utility of our consumer achieved by an increase in the employment of the factor by one (very small) unit; this increase in utility through allocating the small increase in factor utilization to good 1 is MPj1MU1=MPj1p1=wj, and through allocating it to good 2 it is MPj2MU2=MPj2p2=wj again. With our choice of units the marginal utility of the amount of the factor consumed directly by the optimizing consumer is again w, so the amount supplied of the factor too satisfies the condition of optimal allocation. Monopoly violates this optimal allocation condition, because in a monopolized industry market price is above marginal cost, and this means that factors are underutilized in the monopolized industry, they have a higher indirect marginal utility than in their uses in competitive industries. Of course this theorem is considered irrelevant by economists who do not believe that general equilibrium theory correctly predicts the functioning of market economies; but it is given great importance by neoclassical economists and it is the theoretical reason given by them for combating monopolies and for antitrust legislation.
It is important to note that perfect competition is a sufficient condition for allocative and productive efficiency, but it is not a necessary condition. Laboratory experiments in which participants have significant price setting power and little or no information about their counterparts consistently produce efficient results given the proper trading institutions (Smith, 1987, p. 245).
When a firm is making a loss, it will have to decide whether to continue production or not. This decision will, in fact, depend on the different total costs levels and whether the firm is operating in the short run or in the long run.
If the firm is in the short run, and is making a loss whereby:
it is advisable for the firm to continue production. If it fails to achieve these conditions, it is advised to close down so that the only costs the firm will have to pay will be the fixed costs.
Even if the firm stops producing, it will have to continue to meet the level of fixed costs. Since whether the firm produces or not, it will have to pay fixed costs, it is better for it to continue production in an attempt to decrease total costs and increase total revenue, thus making profits. This can be done by:
In the long run, the condition to continue producing requires the price P to be higher than the ATC, i.e. the line representing market price should be above the minimum point of the ATC curve.
If P is equal to ATC, the firm is indifferent between shutting down and continuing to produce. This case is different from the short run shut down case because in long run there's no longer a fixed cost (everything is variable).
Perhaps the closest thing to a perfectly competitive market would be a large auction of identical goods with all potential buyers and sellers present. By design, a stock exchange resembles this, not as a complete description (for no markets may satisfy all requirements of the model) but as an approximation. The flaw in considering the stock exchange as an example of Perfect Competition is the fact that large institutional investors (e.g. investment banks) may solely influence the market price. This, of course, violates the condition that "no one seller can influence market price".
eBay auctions can be often be seen as perfectly competitive. There are very low barriers to entry (anyone can sell a product, provided they have some knowledge of computers and the Internet), many sellers of common products and many potential buyers.
In the eBay market competitive advertising does not occur, because the products are homogeneous and this would be redundant. However, generic advertising (advertising which benefits the industry as a whole and does not mention any brand names) may occur.
Free software works along lines that approximate perfect competition. Anyone is free to enter and leave the market at no cost. All code is freely accessible and modifiable, and individuals are free to behave independently. Free software may be bought or sold at whatever price that the market may allow.
The use of the assumption of perfect competition as the foundation of price theory for product markets is often criticized as representing all agents as passive, thus removing the active attempts to increase one's welfare or profits by price undercutting, product design, advertising, innovation, activities that - the critics argue - characterize most industries and markets. These criticisms point to the frequent lack of realism of the assumptions of product homogeneity and impossibility to differentiate it, but apart from this the accusation of passivity appears correct only for short-period or very-short-period analyses, in long-period analyses the inability of price to diverge from the natural or long-period price is due to active reactions of entry or exit. A frequent criticism is that it is often not true that in the short run differences between supply and demand cause changes in price; especially in manufacturing, the more common behaviour is alteration of production without nearly any alteration of price (Lee 1998). Anyway, the critics of the assumption of perfect competition in product markets seldom question the basic neoclassical view of the working of market economies for this reason. The Neo-Austrian school insists strongly on this criticism, and yet the neoclassical view of the working of market economies as fundamentally efficient, reflecting consumer choices and assigning to each agent his/her contribution to social welfare, is esteemed to be fundamentally correct (Kirzner 1981). Some non-neoclassical schools, like Post-Keynesians, reject the neoclassical approach to value and distribution, but not because of their rejection of perfect competition as a reasonable approximation to the working of most product markets; the reasons for rejection of the neoclassical 'vision' are different views of the determinants of income distribution and of aggregate demand (Petri 2004). In particular, the rejection of perfect competition does not generally entail the rejection of free competition as characterizing most product markets; indeed it has been argued (Clifton 1977) that competition is stronger nowadays than in 19th century capitalism, owing to the increasing capacity of big conglomerate firms to enter any industry: therefore the classical idea of a tendency toward a uniform rate of return on investment in all industries owing to free entry is even more valid to-day; and the reason why General Motors, Exxon or Nestle do not enter the computers or pharmaceutical industries is not insurmountable barriers to entry but rather that the rate of return in the latter industries is already sufficiently in line with the average rate of return elsewhere as not to justify entry. On this few economists, it would seem, would disagree, even among the neoclassical ones. Thus when the issue is normal, or long-period, product prices, differences on the validity of the perfect competition assumption do not appear to imply important differences on the existence or not of a tendency of rates of return toward uniformity as long as entry is possible, and what is found fundamentally lacking in the perfect competition model is the absence of marketing expenses and innovation as causes of costs that do enter normal average cost.
The issue is different with respect to factor markets. Here the acceptance or denial of perfect competition in labour markets does make a big difference to the view of the working of market economies. One must distinguish neoclassical from non-neoclassical economists. For the former, absence of perfect competition in labour markets, e.g. due to the existence of trade unions, impedes the smooth working of competition, which if left free to operate would cause a decrease of wages as long as there were unemployment, and would finally ensure the full employment of labour: labour unemployment is due to absence of perfect competition in labour markets. Most non-neoclassical economists deny that a full flexibility of wages would ensure the full employment of labour and find a stickiness of wages an indispensable component of a market economy, without which the economy would lack the regularity and persistence indispensable to its smooth working. This was, for example, Keynes's opinion. Particularly radical is the view of the Sraffian school on this issue: the labour demand curve cannot be determined hence a level of wages ensuring the equality between supply and demand for labour does not exist, and economics should resume the viewpoint of the classical economists, according to whom competition in labour markets does not and cannot mean indefinite price flexibility as long as supply and demand are unequal, it only means a tendency to equality of wages for similar work, but the level of wages is necessarily determined by complex sociopolitical elements; custom, feelings of justice, informal allegiaces to classes, as well as overt coalitions such as trade unions, far from being impediments to a smooth working of labour markets that would be able to determine wages even without these elements, are on the contrary indispensable because without them there would be no way to determine wages (Garegnani 1990).