Industrial Organization The central research question in industrial organization is: “How can the behavior and performance of firms and markets be explained and predicted with observable data?” IO economists apply microeconomic theory and econometrics to study firms and markets – theoretical and empirical work are both important There are several implications for business strategy and public policy – public policy is often the focus of the analysis Industrial Organization • The main goal in industrial organization is to understand how firms strategically interact in well-defined markets • Ideally, we want to be able to use observable variables to predict firm behavior, profits, and social welfare at the level of particular markets • IO is particularly concerned with how government intervention influences firm behavior and market performance • Antitrust policy, regulatory policies, patent policies, etc., are all important Main Approaches • The typical approach in theoretical IO is to use a formal game theoretic model with profit maximizing firms • The firm as a strategic agent is the unit of analysis; the implications for market performance are critical outcomes • IO approaches often downplay differences in resources and capabilities at the firm level and focus on the implications of “pure” strategic interaction • This is in contrast to the strategy literature, where much of the focus is on how differences in resources and capabilities influence strategic choices Key Concepts: Defining the Market • Most theoretical models in IO focus on strategic interaction in a well defined market (in contrast to Strategy, where the choice of which industries to operate in is a key issue of interest) • In a model, it is easy to define the boundaries of a market; in reality it is more difficult to do so Market Boundaries 1. What is the relevant product? • Most large firms produce several different products • Even if we focus on a particular product it is difficult to draw industry boundaries • For example, should desktop computers be in the same industry as laptops? • In practice, we need to determine how substitutable the products are and how similar the production processes are • The boundaries are often determined by data availability Market Boundaries 2. What is the relevant geographic area? • Often data is available only within the U.S. at the national level, and it may include only domestic firms • This is problematic for several reasons: • First, in many industries the largest U.S. firms sell products all over the world – their market is not confined to the U.S. • For example, many banks argue that they need to be large in order to compete effectively in the international market for loans Second, some industries have natural regional boundaries within the U.S. • For example, how many movie theater chains operate in the Inland Empire? • Edwards, AMC, not many others, yet nationally there are many more firms Third, imports are often obtain a high share of the domestic market, as in the automobile industry • Neither business strategists nor public policy makers would analyze the domestic automobile industry without considering Honda, Toyota, and the other foreign automobile manufacturers Market Boundaries 3. What about dynamics? • Suppose we follow an industry over time • There may be substantial turnover in the leading firms, the products sold, the product processes employed, etc. • Consider video games or computers • Even in non high-tech industries the environment evolves over time Most Real-World Industries are Oligopolies It is unusual to observe an industry with no large firms or only one large firm Typically some firms are larger and more profitable than most of their competitors Firms differ along many dimensions: products, production processes, organizational forms, brand names, location, distribution networks, etc. Those that are able to sustain profits above the norm in their industry are said to have a “competitive advantage” The source of a competitive advantage is typically some resource or capability that is difficult to imitate (particular technologies, trade secrets, brand names, etc.) Industry Structure Because oligopolies vary substantially, it is important to describe the structural features of these industries Key aspects of industry structure include: The number and relative size of large firms The degree of product differentiation The barriers to entry and expansion How much production processes vary Structure-Conduct-Performance Industrial Organization emerged as a distinct field of economics in the late 1930s, beginning with case studies of industries The SCP paradigm led economists to make comparisons across industries; it continues to influence economists and policy makers The basic idea of SCP is that industry Structure determines Conduct (how the firms behave, particularly with respect to pricing), and Conduct determines Performance (measured using rates of return or price-cost mark-ups) In practice, the main link explored was between structure and performance Barriers to Entry • Barriers to entry are important in the SCP paradigm because it takes an essentially static view of an industry: perfect competition is ideal and monopoly is associated with the largest welfare losses • If entry and exit can occur easily, then it is difficult for the established firms to collude or maintain prices above marginal costs • A barrier to entry is a cost of producing which is borne by new entrants but not by established firms • Determining whether something is a barrier to entry can be tricky; assessing the welfare implications of a particular barrier is even more tricky (consider patents) Some Examples of Barriers to Entry 1. 2. 3. • Absolute Cost or Quality Advantages (trade secrets, cost advantages due to learning, location) Capital Requirements (new entrants may lack a reputation and face high costs of raising funds) Economies of Scale (new entrants may have to enter as small, high cost firms or enter large and face the difficulty of acquiring sufficient customers to sell their output) Innovation and Product Differentiation may be used to get around entry barriers Criticisms of SCP Empirical studies have as a whole been inconclusive: at best there is weak evidence of a link between market structure and performance Game theory establishes that conduct is not determined completely by structure: firm behavior must be considered independently of industry structure Further, structure and performance both change over time Causation is not just one-way: High performance encourages entry through innovation or differentiation, which affects structure Low performance encourages exit or capacity reduction Problems with Measuring Performance There are several problems with measuring performance Ideal measures of performance would be economic profits or the difference between price and marginal cost Unfortunately, measuring economic profits is extremely difficult Several adjustments to accounting data must be made to measure assets using replacement values, value investments and intangible assets, and adjust for risk Similarly, measures of marginal cost are typically unavailable, and average cost must be used instead Correlation vs. Causation Even if a statistical link between performance and concentration is discovered, it does not imply that concentration causes the performance It could be that a third variable causes both high mark-ups and high concentration The differential efficiency hypothesis is that large firms are large and profitable because they are more efficient – they have cost advantages or better products Thus, high concentration and high profits are observed together because differential efficiency causes some firms to become large and profitable Implications of Differential Efficiency If the differential efficiency hypothesis is correct, breaking up firms in highly concentrated industries is not a good idea Breaking up these firms would punish them for being efficient and discourage investments that improve efficiency Empirical tests strongly support the differential efficiency hypothesis: a firm’s profit is strongly correlated with its market share In contrast, there is typically only a weak positive association between industry profit and concentration Dynamics • The most interesting current approaches to industry analysis focus on dynamics • In a dynamic world, innovation becomes important, and providing profit incentives for innovation is critical • In a dynamic world, it is no longer the case that perfect competition is ideal; it is essential for firms to obtain profits above the norm when they innovate in order for them to have the incentive to innovate in the first place • Analyses of the links between market structure (or any observable variables) and conduct and performance become more difficult and more complex Entry and Exit • In a dynamic framework, it is critical to consider entry and exit • Dunne, Roberts and Samuelson (1988) summarize the patterns of firm entry, growth, and exit in 387 4-digit SIC Code U.S. manufacturing industries • Their data is census data, so they only observe each industry every five years: 1963, 67, 72, 77, 82 • This implies that they underestimate turnover, because if a firm enters and exits within the five year interval they never observe it • The data is constructed from plant-level data collected in the Census of Manufactures • The data has the complete list of 7-digit products produced in each plant, so even though the 4-digit categories change over time, it is possible to group goods in a consistent way (just reconstruct from the 7-digit data) Types of Entrants They distinguish between three types of entrants: 1. 2. 3. New firms Existing firms that diversify into an industry by opening a new plant Existing firms that enter by altering the mix of outputs they produce in their existing plants Most entrants are new firms, and most are small Diversifying firms are larger Key Results • On average, 93.4% of firms are single-plant firms, but they account for only 17.1% of the value of production • After deleting the smallest firms (those that together produce 1% of the industry’s output), the average entry rate varies from .31-.43; the exit rate varies from .31-.39 • On average, 38.6% of firms are new to the industry each census year • Entrants and exiting firms are small: entrants account for 15.8% of output, and the market share of exiting firms varies from .144 to .191 • An entrant produces 35.2% of what an incumbent does on average Entry by Type New firm, new plant entrants are 55.4% of the number of entrants 50.0% of entrant output 28.4% as large as incumbents Diversifying firm, new plant entrants are 8.5% of the number of entrants 14.4% of entrant output 87.1% as large as incumbents Diversifying firms who are changing their product mix in an old plant are 36.1% of the number of entrants 35.6% of entrant output 34.9% as large as incumbents Exit by Type Of entrants that exit, new firms that construct new plants are the largest group of exiters A relatively high proportion of diversifying firms who constructed new plants survive over time On the whole, entrants and exits tend to be smaller than continuing firms New firms that entered with new plants in the period 19631977 period were the majority of exits during 1963-1982 Variation • There is a lot of variation in entry and exit rates within 2-digit sectors and between 2-digit sectors: industries differ • We can characterize how industries differ. It is not random, but systematic: • High entry rates today imply high entry rates tomorrow (the same relationship holds for exit rates; high exit rates persist) • Further, entry and exit rates are positively correlated • This implies that there are high entry/exit industries and low entry/exit (turnover) industries • After correcting for industry effects by looking at deviations from mean entry/exit rates, periods with above average entry are associated with below average exit Entry Cohorts • The market share of an entry cohort is highest when it first enters; then it declines • Survivors tend to grow and failures are small • The average size of survivors relative to the average size of firms in the industry rises over time • So the trend in market share is due to exit • Exit rates for entry cohorts are quite high, with most of the exit occurring in the first 5 years after entry The Role of Small Firms • It is important to clarify the role of small firms in evolving industries • Are small firms just stragglers that are likely to fail? Are they innovators at the frontier? How do success and failure probabilities depend on strategic choices and luck? • The static perfect competition paradigm suggests that entrants are important for maximizing welfare – they depress prices and increase output • The results of Dunne, Roberts, and Samuelson (1988) suggest that while entrants might play this role, they do not often become important firms in the industry Small Firms in Evolving Industries • In Filson and Gretz (2004), we consider the role of small firms in high-tech industries • Small firms enter to compete in innovation races; if a small firm wins an innovation race it can either market the innovation or license/sell it to an existing firm • Thus, in this model, small firms can grow (if they market their innovation successfully), profit through selling technology, or simply fail • The nature of the innovation and the nature of environmental shocks are important for determining which path occurs Alliances • Alliances are an important phenomena in modern high-tech industries and in other industries where globalization is important • In the biotechnology industry there are complex webs of alliances • So far, few authors have attempted to study these overlapping networks of alliances • For example, research on biotechnology alliances could clarify the roles that small biotechnology firms and large pharmaceutical firms play in these networks Additional Topics • IO is too broad to summarize the various topics, approaches and results in a simple way; corporate finance, strategy, and the theory of the firm are relatively narrow pursuits by comparison • Instead, I will briefly review some of the main topics and some interesting open questions; the list is by no means exhaustive • We have already talked about market structures and the importance of understanding dynamic competition; I will emphasize other topics Pricing • IO is concerned with pricing behavior; much of the analysis is based on microeconomic price theory • Price discrimination, nonlinear pricing, tie-in sales, pricing durable goods, predatory pricing, and limit pricing have all received attention in the literature • Much of this work is quite mature and dates back to the 1950s and 1960s • Recent research on pricing has focused on new environments like the Internet that raise new issues: bundling information goods (like software or music), designing subscription models, considering micropayments, etc. Information and Advertising • For those interested in behavioral economics, advertising would make a useful application • Traditional neoclassical approaches are poorly suited for understanding advertising • In standard microeconomic analysis, the utility function is taken as given – you can inform a consumer about a good but you cannot change his taste for it • In reality, most advertising is persuasive, not informative: it is designed to change consumer tastes • We lack a good understanding of how this occurs A Structural Model of Advertising • Economists who study advertising typically assume it shifts the consumer demand curve in some fashion, but this is a “reduced form” model • A structural model would allow a researcher to assess how advertising works at the micro level (in the brain, perhaps) • Such a model might allow firms to better optimize their advertising strategies, and it would inform policy makers about the potential impacts of alternative advertising policies Technological Change • Much of modern IO is particularly concerned with technological change and policies that influence technological change • Patents, copyrights, and trademarks are important objects of analysis • Joint ventures and alliances • The effects of regulatory policies on technological progress Antitrust Policy and Regulation • Antitrust policy in the U.S. is based on Acts passed in the late 1800s and early 1900s • Since then we have learned a great deal about how industries evolve, and many new devices that affect competitive behavior (such as overlapping networks of alliances) did not exist before • Exploring what modern dynamic analysis can teach us about appropriate antitrust and regulatory policies is likely to be a fruitful research agenda Theory vs. Empirical Work • The ratio of theory to empirical work in industrial organization is generally too high • Small changes in assumptions in game theoretic models often produce substantial changes in results (for example, firms may do too much of x under one set of assumptions and too little of x under another, where x can be just about any strategic choice, such as product differentiation, advertising, etc.) • This would be ok if we could measure and quantify to assess which set of assumptions is valid, but often this is impossible because the models are too abstract (for example, Cournot vs. Stackelberg? Neither is the truth) Reaction • One reaction to this problem is to simply abandon the study of strategic interaction, but this is too extreme • A useful approach is to combine theory with empirical work • Avoid theories that depend heavily on specific assumptions about intricate strategic thinking and concentrate on those that can be tied to observable data • It is also important to try to verify assumptions as much as possible – talk to industry participants if possible • Few economists have produced useful insights in the absence of data to guide them Problems with Talking to Industry Participants There are several problems with talking to industry participants that should be noted: 1. People may distort the facts to protect their interests 2. Practices and strategies often evolve over time; current participants are often not deep thinkers or theorists who continually evaluate standard industry practices, and they may not be able to articulate why they do what they do (Friedman’s example of the pool player is relevant) 3. Industry participants often use terms such as “risk” “variable cost” and so on in different ways than economists – it can be difficult to communicate Problems with Industry Participants 4. Economists often rely on historical data, and it may be difficult to find participants who recall industry events with a high degree of accuracy 5. Participants may not want to talk to you at all; it is often easier to get information from a disinterested party such as a retired executive or consultant, but they may not have the most relevant and up-to-date information Trade journals and historical studies are alternative information sources for learning about firms and markets