Box 42 Further reasons for economies of scale the learning curve
Fixed costs and increasing returns to scale are not the only reasons why average costs of production fall as output rises. Another important factor in some industries has been the learning curve, which relates the firm's average cost of production to its cumulative output. An example of the way we might express such economies is that every time a company doubles its output, costs per unit fall by 25 percent. Such reductions in cost may occur due to better organization and scheduling of complex production processes, such as the assembly of aircraft. In the production of semiconductors they result from the ability to eliminate flaws in the production process. Initial production runs may yield as few as five usable chips out of 100 produced; after more experience is gained, the yield of usable chips may rise as high as 95 percent.
An important aspect of learning is whether it can be transferred from one plant to another within a company or whether it easily spills over to other firms in the same country or even to other countries. A steep learning curve where costs fall rapidly as output expands is likely to result in an industry with fewer firms, because learning represents a barrier to entry similar to fixed costs or increasing returns. Learning is less of a barrier to entry if it easily spills over to domestic competitors. In fact, that possibility is what creates external economies of scale in an industry. If the learning of one firm spills over to another, and vice versa, then expansion of industry output allows all firms to produce more cheaply. Correspondingly, if learning spills over internationally to firms in other countries, then external economies do not create a competitive advantage for producers of just one nation.
A study by Douglas Irwin and Peter Klenow of the worldwide semiconductor industry provides empirical evidence on several of the points raised above.8 Based on analysis of seven successive generations of dynamic random-access memory chips (DRAMs) from 1972 to 1992, they report an average learning rate of 20 percent. This figure holds for both US and Japanese firms. With respect to spillovers within the industry, they find that firms learn three times more from an additional unit of their own cumulative output than from another firm's cumulative output. Thus, firms appear able to appropriate a large share of the benefits from their learning, but because world output is far more than three times the output of any one firm, spillovers play a major role in allowing firm production costs to fall. Spillovers that do occur are just as large across firms in different countries as they are across firms in the same country, and therefore policies to promote national production end up providing a benefit to others. Also, spillovers across different generations of chips generally are not observed, specifically not in the two most recent generations. Thus, fears that government measures will create successful firms in one generation and thereby develop a competitive advantage over other firms in subsequent generations do not appear well founded.
monopolistic competition is that there are few enough firms in the industry that the action of one will not be ignored by the others.
There is even more diversity among models that economists have applied to represent the variety of circumstances that may apply. One extreme is the case where a single domestic producer would not find it attractive to produce for the domestic market alone, but the opportunity to trade and serve the larger world market would warrant the entry of one firm.
High research and development costs to develop a drug that very few people in any one country ever require represents such a case. In the absence of trade, the drug simply would not exist, a clear loss of world welfare. Similarly, the high cost of developing a wide-bodied long-range aircraft to seat 600 passengers would never be warranted if sales were limited to airlines based in a single-country market, and even with access to the world market, no more than one producer appears likely to produce such a plane.
Consider a less extreme case where two firms producing an identical product do exist to serve the world market. We begin by applying a duopoly model that shows how one firm alters its output in response to output decisions of the other firm.9 Such a model, developed by Augustin Cournot,10 can be summarized in two reaction curves as shown in Figure 4.7. Let the two curves correspond to a Dutch firm and to an English firm. If the Dutch firm held a monopoly it would produce at point DM along the vertical axis; if the English firm held a monopoly it would produce at point EM along the horizontal axis. The English firm's reaction function shows that as Dutch output rises, English production will fall. Because two firms find it profitable to operate in this industry, the English firm will not be able to operate as a monopolist at point EM. If English output initially were at that level, the Dutch response would be to produce at D1, as given by the Dutch reaction function. At that level of output, the English firm would then choose to produce E1. In turn, the Dutch firm would respond by producing D2. This process converges to the equilibrium shown at Z where the two reaction curves intersect. Point Z does not lie along a straight line connecting DM and EM, and therefore this solution shows that more total output will be produced than when a monopoly controls the market. Because more output is sold, a lower price must be charged. Thus, gains from competition are possible in a duopoly setting.
Douglas Irwin applied this duopoly framework to explain the rivalry between the English East India Company and the Dutch United East India Company for the spice trade with Southeast Asia from 1600 to 1630.11 Because land transportation was such an expensive alternative, competition between sea-faring traders provided the main check on the market power of any one firm. Furthermore, Queen Elizabeth I granted a 15-year exclusive monopoly
English output
Figure 4.7 Reaction curves and duopoly trade. An English monopolist chooses to produce EM. If a Dutch firm enters the market, it offers the quantity D1 as indicated by its reaction curve. The English firm reacts by producing E1, as indicated by its reaction curve, which results in a further Dutch response to offer D2. This sequential adjustment leads to equilibrium at point Z.
English output
Figure 4.7 Reaction curves and duopoly trade. An English monopolist chooses to produce EM. If a Dutch firm enters the market, it offers the quantity D1 as indicated by its reaction curve. The English firm reacts by producing E1, as indicated by its reaction curve, which results in a further Dutch response to offer D2. This sequential adjustment leads to equilibrium at point Z.
to the English East India Company, and the Dutch similarly granted the Dutch United East India Company monopoly rights to trade with Asia. No other country had comparable maritime power, and thus, a duopoly setting describes this trading situation quite accurately.
The Cournot model implies that the basic decision each firm must make is how large a quantity of goods to bring to market, which is an appropriate description of the spice trade. Each trading company determined the number of ships to send to Asia and then auctioned off the pepper brought back to Europe. The symmetric diagram shown in Figure 4.7 also appears appropriate because the Dutch and English each sold pepper in the same European market, they both had access to the Asian markets to acquire pepper, and they had comparable costs to transport it back to Europe. We would expect each firm to gain half of the market.
That outcome, however, did not emerge. The Dutch accounted for nearly 60 percent of the market. Irwin suggests that the Dutch East India Company followed a strategy other than the profit maximization assumed in the Cournot model. Stockholders could not check the actions of company agents in the field, whose remuneration depended upon total turnover and growth. Such agents had no incentive to cut back their efforts when British sales expanded, and the Dutch produced more than called for by the Cournot model. Nevertheless, this strategy was beneficial to the Dutch, giving them 20 percent higher profits than in the Cournot case, because it in effect implemented a leadership strategy later identified by Heinrich von Stackelberg.12 The success of the strategy arises due to the reduction in the competitor's (British) output, given the leader's (Dutch) decision to expand so much. The outcome is comparable to Dutch maximization of profits assuming it could count on a subsequent British reduction in output. In terms of Figure 4.7, the strategy represents a point such as W, where total industry output (British plus Dutch) is greater than at Z, and prices are lower. Dutch profits are greater due to their larger share of this expanded market. Even though prices are lower, they still exceed the cost of production and contribute to higher profits when sales expand sufficiently.
In Chapter 6 we return to this topic because it has arisen in current debates over strategic trade policy. The Dutch gain was not the result of a carefully implemented government strategy, and Irwin demonstrates that an even larger gain was possible. Could modern-day governments achieve similar gains with more purposeful intervention? Although any historical example is subject to multiple interpretations, Irwin raises the cautionary note that aggressive Dutch expansion in the Indonesian spice trade relegated Britain to greater trade with India. The subsequent British opportunity to develop trade in cotton and cotton textiles is viewed by some economic historians as an important ingredient in the birth of the Industrial Revolution.13
The model presented above applies when two firms compete to serve a single market as in the case of the seventeenth-century pepper trade. An advantage of that situation is that drawing any conclusions about the welfare of the two supplying countries is more straightforward. When the consumption primarily occurs in some third-country market, only the change in profits earned by the supplying firms must be examined. However, we can also apply this framework to consider two identical countries that initially are each served by a domestic monopoly. If trade becomes possible and the two firms compete as Cournot oligopolists, with the same cost of serving either market, the solution in Figure 4.7 applies to any one country's market. The English producer, for example, no longer holds a monopoly in the English market. Competition with the Dutch firm leads to the solution at point Z, where more of the product is sold to consumers at a lower price. In the Dutch market, the Dutch monopolist likewise must compete with the English firm, which results in a greater quantity and a lower price being charged. The possibility of trade has a pro-competitive effect that benefits each country, as the market price comes closer to marginal cost, the optimal condition from a competitive market. Although monopoly profits fall, that represents a benefit to consumers, and in the symmetric case assumed here, any loss in English (Dutch) profits is more than offset by gains to English (Dutch) consumers.
Post a comment