Investment Decisions with Economies of Scale and Learning
分析存在规模经济和学习效应时企业的动态竞争策略，比较纳什竞争与斯塔克尔伯格竞争对市场结构和效率的影响，发现更激烈的竞争可能导致更低产出。
Economists in an antitrust case have at their disposal quite a large bag of tools and truisms, but for the most part these are derived from studies of static models. In many industries the policy issues concern the implications of firm behavior on market structure and performance over time. A case in point is a recent Federal Trade Commission complaint against the DuPont Corporation. The FTC alleged that DuPont had engaged in a strategy designed to monopolize the market for titanium dioxide, better known as the coloring agent in white paint. The alleged strategy was, in essence, what some would recognize as the Boston Consulting Group story: When a firm has a lead in an industry with significant learning economies, the firm should price below competitors' costs and expand to take further advantage of learning effects and, in the process, to increase market share.' This paper summarizes the results of an analysis of dynamic competition with scale and learning effects. The research is a preliminary exploration. All results are obtained under rather special assumptions about the production technology, the effects of experience on costs, and the strategic interactions between firms. Two kinds of strategic behavior are considered. In the first case, each firm takes the production decisions of competitors as given (the Nash assumption). In the second case, firms consider pre-emptive capacity investments, taking into account that competitors will alter their future investment plans to achieve nonnegative profits. We call this a Stackelberg (CS) game. Using the maximization of net surplus as a socially optimal benchmark, neither form of competition yields an efficient outcome when new investment exhibits increasing returns to scale. In the absence of learning effects, smaller firms in a Nash competition have a greater incentive to add new capacity than do larger firms, and the equilibrium industry structure approaches equal market shares. Introducing learning effects in the Nash game causes a tendency toward increased concentration, but monopoly is not an inevitable consequence. The CS game is competitive than the Nash game in the sense that firms compete for the right to invest at each instant of time. With identical firms and no learning effects, the market structure in a CS equilibrium is indeterminate, although the sequence of industry investments is well defined. Introducing firm-specific and nonstochastic learning has a dramatic effect on the CS equilibrium. All new investment is undertaken by a single firm, even if the learning economies are small. Moreover, the level of output could be lower (and price higher) in a CS equilibrium than it would be in a Nash equilibrium. In this sense more competition can lead to a lower rate of output over time.