Incumbent Pricing Responses To Entry

Strategic Management Journal Strat. Mgmt. J., 26: 1229–1248 (2005) Published online in Wiley InterScience (www.interscience.wiley.com). DOI: 10.1002/smj.502

INCUMBENT PRICING RESPONSES TO ENTRY DANIEL SIMON* Department of Applied Economics and Management, Cornell University, Ithaca, New York, U.S.A.

Empirical research on incumbent pricing responses to new entry has yielded mixed results. Some studies find that incumbents cut prices post entry, while others find that incumbents accommodate entry by leaving prices unchanged (or even raising prices). To better understand these findings, this study explores the conditions under which incumbents are more likely to cut prices when faced with entry. I argue that incumbents vary in their incentives to cut prices; incumbents with greater incentives are more likely to respond aggressively to entry. Using data on magazine subscription prices, I find that, on average, incumbent magazines do not reduce prices following entry. But, the results suggest that newer incumbents cut prices more than older incumbents, while incumbents that compete in fewer and in more competitive markets cut prices less following entry. These results help to explain inconsistent empirical results in the literature, and support a more general explanation of when firms respond aggressively to entry: incumbents respond to entry more aggressively when their incentives to do so are greater. Copyright  2005 John Wiley & Sons, Ltd.

INTRODUCTION The entry of new firms has important effects on incumbent firms and consumers. New entrants increase competition, reducing market share and profits of incumbent firms. New entrants also introduce new products and processes, forcing incumbents to become more efficient and innovative (Geroski, 1995). As a result, incumbents have strong incentives to deter entry, while consumers welcome new entrants. When facing new entry, incumbent firms must decide how to respond. They may reduce prices before or after entry, either to deter potential entrants or simply to maximize current profits in Keywords: entry; incumbent response; pricing; magazines; competitive dynamics *Correspondence to: Daniel Simon, Department of Applied Economics and Management, Cornell University, Ithaca, NY 14853, U.S.A. E-mail: [email protected]

Copyright  2005 John Wiley & Sons, Ltd.

the face of increased competition. Alternatively, incumbents may accommodate entry, leaving prices unchanged (or raising prices). This study examines incumbents’ pricing responses to new entry in the consumer magazine industry. Theories of entry deterrence and incumbent response to entry have evolved, with a large number of studies generating a variety of predictions. Early limit pricing studies predicted that incumbents would reduce prices before entry, but leave them unchanged after entry. More recent gametheoretic studies suggest that incumbents may reduce prices post entry as a way to drive out entrants and deter future entrants. These evolving theoretical predictions have prompted many empirical studies of incumbent responses to entry. The results of these studies are very inconsistent: some find that entry has a negative effect on incumbent prices; others find that entry has no effect on incumbent

Received 24 September 2002 Final revision received 4 June 2005

1230

D. Simon

pricing. Even within the same industry, incumbent responses to entry may vary (Frank and Salkever, 1997; Yamawacki, 2002). These inconsistent findings may indicate that incumbents only respond aggressively to entry under certain conditions. This study argues that incumbents’ heterogeneous responses reflect varying incentives to respond to entry; incumbents with stronger incentives are more likely to respond. To consider the inconsistent findings regarding incumbent pricing responses to entry, I examine whether incumbent and market characteristics may influence the incumbents’ pricing response. That is, this study seeks to better understand the heterogeneity in incumbent responses to entry by examining the conditions under which incumbent firms have a greater incentive to cut prices in response to entry. Consistent with prior empirical work, this study finds substantial heterogeneity in incumbents’ pricing responses. The results indicate that incumbent and market characteristics influence the incumbent’s incentive to respond to entry; newer incumbents cut prices more than older incumbents, while incumbents that compete in fewer and in more competitive markets cut prices less following entry. These results suggest that some of the inconsistent findings may reflect variation in incumbent and market characteristics. While this study focuses on pricing responses to entry, it should be pointed out that incumbents are not restricted to reducing prices as a way to respond to the threat of entry. As discussed below, firms can use various non-price weapons to deter entry (Geroski, 1995). This study examines pricing behavior because theory offers more testable predictions regarding heterogeneous pricing responses by incumbents, and because the inconsistent empirical results regarding incumbent pricing responses pose a mystery to researchers. Moreover, prices are observable and measurable, while many non-price responses are unobserved or difficult to measure.

LITERATURE REVIEW A great deal of research has examined entry and incumbent responses to entry. But as Geroski (1995) notes, empirical research on entry is much less conclusive than theoretical research. Theoretical models suggest that firms can use price as an Copyright  2005 John Wiley & Sons, Ltd.

entry deterrent before or after entry has occurred. Limit pricing models emphasize deterring entry before it occurs. Predation models focus on cutting prices after entry as a way to drive out entrants and deter future entry. But empirical research has provided little support for any theory of price as an entry deterrent. It appears that in an effort to be parsimonious and tractable, theoretical models have omitted important contingency factors that influence the incumbent’s incentive to respond to entry. Early theoretical research, notably by Bain (1956), Modigliani (1958), and Sylos-Labini (1962), emphasizes limit pricing: an incumbent firm setting its pre-entry price low enough to make entry appear unprofitable. Implicit in the limit pricing models is the belief that potential entrants use current industry profits as an indicator of future profits (Masson and Shaanan, 1982). Such a strategy makes sense for incumbents if the foregone pre-entry profits are less then the additional profits earned by deterring entry and subsequently raising prices back to monopoly levels (Geroski, 1995). Empirical evidence provides little support for limit pricing strategies. A survey of U.S. firms by Smiley (1988), and a survey of U.K. firms by Singh, Utton, and Waterson (1997) both find that firms report rarely using prices to deter entry. Dynamic limit pricing models, which suggest that incumbents use their prices to regulate entry behavior continuously, also find little support empirically. Game-theoretic models emphasize that under conditions of complete information limit pricing is not a credible deterrent. Limit pricing models traditionally assume that potential entrants believe that incumbents will not change their price or output after entry. But game-theoretic models demonstrate that under conditions of complete information, regardless of its pre-entry behavior, maintaining the limit price after entry is not profit maximizing; once entry has occurred, the incumbent has an incentive to raise the price from the limit price to the post-entry equilibrium level. In other words, with complete information, pre-entry prices or profits do not predict post-entry profitability (Milgrom and Roberts, 1982a). Rational potential entrants will not be affected by low pre-entry prices, and thus incumbents only squander profits by limit pricing (Milgrom and Roberts, 1982a). These game-theoretic models emphasize the need Strat. Mgmt. J., 26: 1229–1248 (2005)

Incumbent Pricing Responses to Entry for incumbents to credibly commit to low postentry prices in order to deter entry. Game-theoretic research has considered a variety of strategies and conditions that might provide credible deterrents to entry. Spence (1977) and Dixit (1980) develop a model in which an incumbent invests in excess capacity as a way to deter entry. The excess capacity allows the incumbent to commit to increased post-entry output and a lower post-entry price. This commitment to price reduction makes entry less attractive. But empirical research provides little support for the use of excess capacity as an entry deterrent. Masson and Shaanan (1986) and Lieberman (1987a) find no evidence that incumbents invest in excess capacity to deter entry, although Lieberman (1987b) does find that in more concentrated industries firms build capacity after entry. In a survey of U.S. firms, Smiley (1988) finds that almost none of the respondents used excess capacity to deter entry. Singh et al. (1997) received similar responses from U.K. firms. Another game-theoretic approach uses asymmetric information to make low prices a credible threat. Milgrom and Roberts (1982a) develop a signaling model in which the entrant is uncertain about the incumbent’s costs. In this model, the incumbent tries to signal that it has low costs by setting a low price pre-entry, as a way to deter entry. Kreps and Wilson (1982) and Milgrom and Roberts (1982b) show that when the entrant is uncertain about the incumbent’s payoffs the incumbent may have an incentive to cut prices after entry as a way to build a reputation for Table 1.

1231

fighting entry. A tough reputation may deter future entrants by reducing the expected profitability of entry. Few studies have explicitly tested theories of predatory pricing, largely because it is difficult to identify predatory pricing. But several papers have examined incumbent pricing responses to entry, yielding inconsistent results. Some find that incumbents cut prices post entry while others find no response, or even a positive response. In the airline industry, Joskow, Werden, and Johnson (1994) find that incumbents cut prices following entry, while Windle and Dresner (1995) report that entry by low-cost carriers induces price cuts from incumbents. Looking at retail grocery stores, Marion (1998) finds that prices are negatively affected by entry by warehouse stores. In the retail tire industry, Bresnahan and Reiss (1991) find that prices fall with entry. On the other hand, Thomas (1999) finds that incumbents in the readyto-eat breakfast cereal industry do not cut prices after entry (Thomas, 1999). Finally, two studies find that incumbent pricing responses vary within the same industry. Frank and Salkever (1997) find that brand-name prescription drug prices increase after generic entry, but generic prices fall with generic entry. Yamawacki (2002) finds that some car manufacturers cut prices in response to entry while others do not. Table 1 summarizes the empirical studies of incumbent pricing responses to entry. While empirical results are mixed, little is known about why some incumbents reduce price post entry and others do not. The literature offers little

Summary of empirical studies of incumbent pricing responses to entry

Authors Joskow, Werden, and Johnson (1994) Windle and Dresner (1995) Marion (1998) Thomas (1999) Frank and Salkever (1997) Bresnahan and Reiss (1991) Yamawacki (2002)

Industry

Key result

Airlines

Incumbents cut prices post entry

Airlines

Incumbents cut prices following entry by low-cost carriers Incumbent supermarkets reduce prices following entry by warehouse stores Incumbents accommodate on price

Grocery stores RTE breakfast cereal Pharmaceuticals Local retail markets Automobile segments

Copyright  2005 John Wiley & Sons, Ltd.

Incumbent producers of prescription drugs raise prices after generic entry, while generic incumbents reduce price after generic entry Incumbent tire retailers reduce prices following entry in local retail markets Firm-specific and group-specific factors determine which incumbent auto manufacturers respond to entry Strat. Mgmt. J., 26: 1229–1248 (2005)

1232

D. Simon

explanation for these inconsistent findings. Little empirical research has considered the conditions under which firms are more or less likely to fight entry. I believe the main reason for this gap in the literature is that explaining why some firms respond to entry more aggressively than others requires detailed firm-level data, including firmlevel pricing data as well as other firm and market characteristics. Yamawacki (2002) argues that incumbent pricing responses to entry are firm-specific, depending on the incumbent’s ability to respond to entry, which in turn depends on its relative resource position. But he does not specify what factors influence an incumbent’s ability to respond to entry. Another reason why incumbents may not cut prices after entry is that they may respond to entry with nonprice tools. In industries with differentiated products, firms have a variety of tools available to them for responding to new entry. Most prominently, these include advertising and promotional campaigns, and new product introductions (Thomas, 1999), as well as product improvements. Firms may use these other types of responses instead of or in addition to price responses. If firms substitute these non-price responses for price cuts, then the negative relationship between entry and price response will not hold. There is some evidence that firms substitute non-price responses for price responses in highly differentiated industries. Studies in the soda and ready-to-eat breakfast cereal industries find that while firms accommodate on price, they use advertising to deter or limit entry (Thomas, 1999). But this argument fails to explain why incumbents in the airline and the auto industry cut prices after entry. I suggest a more general explanation for why some firms respond to entry more aggressively than others: incumbents vary in their incentive to respond to entry. Incumbents with greater incentives to respond to entry are more likely to respond aggressively. In the following section I develop this explanation more fully, arguing that the incumbent’s age (time in the market), the incumbent’s corporate scope, and the market structure all can influence incumbents’ incentives to respond. First, however, I summarize the theoretical arguments for why incumbents are expected to cut price following entry. Copyright  2005 John Wiley & Sons, Ltd.

HYPOTHESES There are two distinct reasons why incumbents may cut prices in the face of entry: deterrence and current profit maximization. Entry deterrence strategies can be divided into two groups: those that predict a pre-entry price response, and those that predict a post-entry response. In the first category are limit-pricing models in which incumbents set a pre-entry price below the current profitmaximizing level as a way to reduce the potential entrants’ assessments of the benefits of entry. In the second category are game-theoretic models in which incumbents set post-entry prices below the current profit-maximizing level as a way to drive out current entrants and deter future entrants. While neither set of models has received much empirical support, post-entry price cutting has received more theoretical support. Absent any efforts to deter entry, incumbents may be expected to reduce prices in order to maximize current profits. New entrants force incumbents to reduce prices by increasing market supply, stealing market share, and reducing incumbents’ ability to tacitly collude. Moreover, because new entrants often introduce new technologies, they may force incumbents to become more efficient as well (Geroski, 1995). As a result, incumbents often have incentives to reduce prices that do not involve entry deterrence. While incumbents have incentives to cut prices in response to entry, this incentive is weaker in differentiated product markets because demand is more inelastic with respect to price and more elastic with respect to other marketing tools. In such cases, firms may accommodate on price and respond aggressively with another competitive weapon (Gruca, Kumar, and Sudarshan, 1992). However, firms in differentiated markets may also respond aggressively with more than one competitive weapon (Gatignon and Hanssens, 1987). For example, an airline may both reduce its price and increase the number of flights it offers as a way to drive out a new entrant from a particular route. Moreover, several papers in the marketing area argue that incumbents in differentiated markets are almost always better off reducing price post entry (Hauser and Shugan, 1983; Kumar and Sudarshan, 1988; Gruca et al., 1992). As a result, I offer the following hypothesis (given the substantial prior testing of this relationship, this hypothesis Strat. Mgmt. J., 26: 1229–1248 (2005)

Incumbent Pricing Responses to Entry is included for theoretical completeness rather than for its original contribution): Hypothesis 1: Incumbent prices are negatively related to entry. While theory predicts an incumbent response to entry, evidence shows that incumbents respond selectively (Geroski, 1995). The heterogeneous responses to entry may reflect varying incentives to respond. Incumbents that stand to lose more from entry, or gain more from deterring entry, are more likely to respond aggressively to entry. Incumbent characteristics, such as the incumbent’s time in the market, may influence its incentive to respond. Reflecting the liability of newness (Stinchcombe, 1965), newer firms are more vulnerable to entry than older firms. There are two possible explanations for this vulnerability. Both arise from the need for non-tradable assets that are characterized by time compression diseconomies (Dierickx and Cool, 1989). Older firms are more likely to have well-established reputations. Reputation cannot be acquired; a firm must invest in its reputation—for quality, for integrity, for innovation, etc.—over time (Dierickx and Cool, 1989; Fombrun and Shanley, 1990). The longer a firm has been in existence, the greater its opportunity to develop a favorable reputation with customers, suppliers and investors. A firm’s reputation can provide a competitive advantage by differentiating the firm’s products, allowing the firm to charge higher prices (Fombrun and Shanley, 1990). Newer firms that invest more to create reputations more quickly are likely to be unsuccessful because of time compression diseconomies in the production of a reputation. As a result, one might expect newer firms to be more threatened by entry due to their weaker reputations. Newer firms may also have higher costs or lower quality because they lack firm-specific resources and know-how that older incumbents possess (Geroski, 1995). Firm-specific skills and knowhow are developed through on-the-job learning and training (Williamson, 1979). This tacit knowledge is difficult to acquire because in most cases it is embedded in the company and cannot be codified (Nonaka, 1994). Firms with more experience have a greater opportunity to develop this tacit knowledge and move down the learning curve, reducing costs and/or improving quality. Learning by doing Copyright  2005 John Wiley & Sons, Ltd.

1233

may give older incumbents a competitive advantage over newer incumbents and new entrants.1 The case of Steinway & Sons (Kotha and Dunbar, 1996) exemplifies a learning-by-doing advantage. Manufacturing a Steinway piano is a complex process, requiring about 300 skilled artisans that develop their skills through years of apprenticeship (Kotha and Dunbar, 1996). Quality is embedded throughout the process, as ‘each step is contingent upon the success of the previous steps’ (Kotha and Dunbar, 1996: 10). Through years of trial and error, Steinway has developed a unique ‘craft approach’ that has ‘hardly changed for the last century’ (Kotha and Dunbar, 1996: 8). Using an apprenticeship system to transfer skills, Steinway has maintained its quality advantage in the face of entry: ‘It has proved impossible to produce a piano equal to the Steinway piano, even though Steinways were copied to the minutest detail’ (Dolge, 1911). Over time, Steinway has developed knowhow that protects it from the threat of entry. Empirical evidence supports the view that new firms are more vulnerable to entry as well. From 1963 to 1982, nearly 80 percent of all new firms in the United States failed within 10 years (Geroski, 1995), with new entry being a primary cause: ‘One of the clearest features of the data on entry . . . is that most entry results in exit, first by recent entrants’ (Geroski, 1995: 434). Because of their increased vulnerability to new entrants, newer firms are more likely to cut prices in response to entry, as a way to protect themselves. Hypothesis 2: Newer incumbents cut prices more in response to entry. Corporate scope can also affect a firm’s incentive to respond to entry. By fighting entry in one market, a multi-market incumbent can build a reputation for fighting entry that may deter entry in its other markets (Kreps and Wilson, 1982; Milgrom and Roberts, 1982b). The value of such a reputation increases with the number of markets in which a firm can use this reputation: The more markets that a firm serves, the greater its incentive for building and maintaining its reputation (Milgrom and Roberts, 1982b). Single-market firms 1 While older firms are likely to have moved further down the learning curve, not all knowledge must be developed through experience. In many cases new entrants have a cost or quality advantage based on new technology that they developed or acquired (Geroski, 1995).

Strat. Mgmt. J., 26: 1229–1248 (2005)

1234

D. Simon

have less incentive to build a reputation for fighting entry because the benefits of deterring entry are restricted to the market in which they compete. An incumbent competing in multiple markets can benefit in its other markets by deterring entry in any one market. For example, if a locally owned department store cuts prices when a new department store opens up in town, this may reduce the likelihood that other potential entrants will open a department store in town. However, if Wal-Mart aggressively cuts prices when a new store enters one of their markets, this may reduce the likelihood of entry into the focal market and into all other cities in which Wal-Mart competes. Potential entrants into other markets may infer that because Wal-Mart responded aggressively in one market, it will do so in others as well. Because it may deter entry in multiple markets, Wal-Mart has a greater incentive to respond aggressively to entry in any market. There are several counter-arguments to this reasoning. Some might argue that incumbents who compete in only one market have more to lose to new entrants, because they have no other alternative. But this argument applies only if exit is more costly for single-market incumbents than for multimarket incumbents.2 Another counter-argument is posed by Smith, Grimm, and Gannon (1992), who argue that firms with more complex organizational structures are less likely to respond to competitive attacks. They posit that in structurally complex firms decision-makers receive information more slowly and are more likely to receive misinformation. This weakens the ability of these firms to respond to competitive attacks. But, Smith et al. (1992) find little support for this hypothesis. Finally, research on multi-market competition indicates that firms may compete less intensely against each other when they meet in multiple markets (Gimeno and Woo, 1999). This suggests that when incumbents compete against new entrants in other markets the incumbents are less likely to respond aggressively. But, de novo entry (entry by 2 For this to be true, two conditions must hold. The incumbent must have assets that are transferable across markets, but firmspecific, such that it cannot sell them when it exits the market. Additionally, entering another market must require the entrant to incur some sunk costs (invest in assets that are both firmand market-specific) as well. However, to the extent that most of the important assets in the industry are not market-specific, then the single-market firm incurs little additional cost (relative to the multi-market firm) in transferring its resources to a new market.

Copyright  2005 John Wiley & Sons, Ltd.

new firms) is much more common than entry by diversifying incumbents (Geroski, 1995). Therefore, it is likely that in most cases incumbents do not meet entrants in other markets. As a result, the net effect of multi-market incumbency is likely to negatively influence incumbents’ pricing response to entry. Hypothesis 3: Multi-market incumbents cut prices more in response to entry. Market structure may also influence the incumbents’ incentive to aggressively respond to entry. Generally, the threat posed by entry should be greater in more concentrated markets (Hannan, 1979). In highly competitive markets entry should have little effect on incumbents as competition has already forced high prices down towards marginal cost, while in more concentrated markets entry threatens to erode rents by making it more difficult to maintain tacit collusion. Therefore, incumbents in highly concentrated markets have a greater incentive to cut prices, both to drive out entrants and to deter further entry (Hannan, 1979; Kessides, 1990). Alternatively, higher concentration may reflect entry barriers or incumbent capabilities which make it difficult for new firms to enter a market. In this case, the entry barriers would grant the incumbents a competitive advantage over new entrants, reducing the need for incumbents to aggressively respond to entry. But, while entry barriers appear to be high in many industries, they have little effect on rates of entry (Geroski, 1995). This suggests that they also have little effect on incumbent responses. Hannan (1979) finds that incumbents in concentrated markets cut prices more in response to entry. Similarly, Lieberman (1987b) finds that incumbents in concentrated markets increase capacity by a greater amount following entry. Therefore: Hypothesis 4: Incumbents in concentrated markets cut prices more in response to entry.

METHODS AND DATA To test these hypotheses, I use data on U.S. consumer magazines. The consumer magazine industry provides a good setting for studying incumbent responses to entry because it has Strat. Mgmt. J., 26: 1229–1248 (2005)

Incumbent Pricing Responses to Entry well-defined markets, allowing the researcher to identify entrants and incumbents. Furthermore, the existence of multiple markets allows me to examine how entry affects pricing behavior across markets while holding constant other industry characteristics which influence entry barriers and/or pricing behavior. Moreover, there is uniform price data available, allowing one to examine how pricing varies with entry. Magazines are differentiated products. Even within the same market there are differences in editorial quality, subject matter, and style. Therefore, publishers may cut prices or they may improve some dimension of quality to deter or limit entry. For example, an incumbent magazine might publish more articles about celebrities as a way to improve its quality and make it more difficult for an entrant to attract subscribers. But it is difficult to adjust quality in the short run, especially in the market for subscribers, which is the focus of this study. Data This study uses data on consumer magazines sold in the United States whose circulation is audited by the Audit Bureau of Circulation (ABC). ABC collects data on circulation, advertising rates, and cover prices, providing verified circulation figures to advertisers. The data span an 11-year period from 1990 through 2000. The sample includes 449 incumbent magazines,3 owned by 246 publishers, competing in 42 different markets.4 In total, there are 3715 magazine–year observations. Markets are defined by magazine topics. For example, there are markets for cooking magazines, car magazines, sports magazines, etc. Table 2 lists the magazine markets and a sample magazine from each. These market definitions reflect 3 I define incumbents as magazines selling at least four issues per year, offering subscriptions, which were not founded during the current or previous year. This is to avoid coding the same magazine as both an entrant and an incumbent in the same year. I also exclude magazines acquired during the current year. 4 I exclude magazines in three markets. I exclude men’s and general editorial magazines because the definitions of these markets used by the Audit Bureau of Circulation did not correspond clearly with the same markets in the entry data. I exclude magazines in the metropolitan/regional/state market because, owing to its geographic nature, magazines in this market generally do not compete with each other, e.g., Chicago magazine does not compete against Atlanta magazine. Therefore, incumbents should not be expected to respond to new entrants (except in the rare case when they compete in the same geographic market).

Copyright  2005 John Wiley & Sons, Ltd.

1235

Table 2. Magazine markets and sample magazines Market Art and antiques Automotive Aviation Babies Black/African-American Boating and yachting Bridal Business and finance Camping and outdoor recreation Computers Crafts, games, hobbies, and models Dogs and pets Dressmaking and needlework Education and teacher Entertainment and performing arts Epicurean Fashion, beauty, and grooming Fishing and hunting Fitness Gardening (home) Gay publications Health Home service and home Horses, riding, and breeding Literary, book reviews, and writing techniques Mature market Military and naval Motorcycle Music Nature and ecology Parenthood Photography Political and social topics Popular culture Religious and denominational Science/technology Sports Teen Travel TV and radio/communications and electronics Women’s Youth

Sample magazine Art & Antiques Road & Track Flying Twins Magazine Today’s Black Woman Salt Water Sportsman Modern Bride Entrepreneur Magazine Backpacker Byte Magazine Crafts Aquarium Fish Mccall’s Needlework & Crafts Creative Classroom Us Weekly Food & Wine Harper’s Bazaar American Angler Walking, The Magazine For Smart Health & Fitness Flower and Garden The Advocate Prevention Decorating Remodeling Equus The New York Review Of Books New Choices: Living Even Better After 50 Army Times Cycle World Rolling Stone Wildbird Child American Photo National Review Interview Catholic Digest Discover Tennis Tips & Tricks Conde Nast Traveler Video Cosmopolitan Disney Adventures Strat. Mgmt. J., 26: 1229–1248 (2005)

1236

D. Simon

the interaction among magazines competing for readers and advertisers. For example, ArtNews magazine and Art in America magazine are both in the art and antiques market, competing for the same customers and advertisers. Similarly, Bowhunting magazine competes for customers and advertisers with Deer & Deer Hunting magazine in the hunting and fishing market. Data on entry were collected from annual editions of Samir Husni’s Guide to New Consumer Magazines (Husni, 1991–2001). Each volume provides data on new consumer magazines sold in the United States, categorized by type of magazine. Variables Dependent variable The dependent variable is the log of the perissue price for a 1-year subscription.5 The per-issue price is the listed price for a 1-year subscription divided by the number of issues. I use the per-issue price rather than the total subscription price to facilitate analysis across magazines with different numbers of issues. I focus on subscription prices rather than single-copy prices because single-copy prices have been shown to be very ‘sticky’ (Cecchetti, 1986; Round and Bentick, 1997), and because subscriptions generally account for roughly three times more revenue than singlecopy sales (Papazian, 1997). In addition, subscribers tend to be more price sensitive than single-copy buyers (Round and Bentick, 1997), while single-copy buyers are much more influenced by non-price factors such as the magazine cover (Daly, Henry, and Ryder, 1997). As a result, publishers are more likely to emphasize non-price dimensions such as the magazine cover in competing for single-copy readers, while price is a more important competitive weapon in the subscription market. Finally, this study analyzes subscription prices rather than advertising rates because incumbents have a greater incentive to reduce subscription prices due to the positive effect that this has on advertising revenues (by increasing circulation the magazine is able to charge higher ad rates).

Independent and control variables I measure entry with a dummy variable that takes a value of one for a particular incumbent magazine if at least one new rival magazine enters the market during the current year, and zero otherwise. In order to measure entry it is necessary to determine what constitutes a new entrant. For the purposes of this study, I include all new magazines listed in Samir Husni’s annual Guides to New Consumer Magazines that were published at least four times per year in that year.6 In a very small number of cases, I find magazines listed as new entrants in the incumbent dataset for at least one prior year. These magazines are not considered to be new entrants. I also consider the identity of the publisher of the new entrants. If the publisher of an incumbent magazine introduces a new entrant into the same market, then the new magazine is not treated as a new entrant because it is not a rival. Therefore, I define entry specific to each incumbent magazine. Only new magazines introduced by rival publishers are considered new entrants. The incumbent’s age is the logarithm of the number of years since it was founded (since it entered the market). I use log of age because the effect of age is likely to diminish at higher levels. Similarly, I use the log of the number of markets in which it sells magazines to measure the publisher’s corporate scope. A Herfindahl index measures market structure. In addition to these variables, I include several control variables. The aggregate circulation of rival magazines controls for market demand. To further control for market structure, I include the number of rival incumbent magazines.7 Similarly, I include a dummy variable to indicate whether or not any rival magazines exited the market during the current year.8 This allows me to control for the effect of exit, which tends to be correlated with entry (Evans and Sigfried, 1994). Finally, to control for the publisher’s experience with entry, I include the percentage of entrants that a publisher has faced since 1990 that have survived to the current year. In addition, I control for unobservable magazine, publisher, and year fixed effects, as described below. 6

I exclude new magazines with an unknown number of issues. This variable excludes rival new entrants. 8 A magazine is defined as exiting the market if it no longer appears in the dataset and if it does not appear in that year’s edition of the Standard Periodical Directory. 7

5 I use log price because of skewness in price, and to be consistent with prior firm pricing studies.

Copyright  2005 John Wiley & Sons, Ltd.

Strat. Mgmt. J., 26: 1229–1248 (2005)

Incumbent Pricing Responses to Entry

1237

Model

RESULTS

To test the hypotheses, I estimate the following fixed-effects model:

Table 3 provides descriptive statistics and a correlation matrix. Looking at the sample means, there is an entry in 92 percent of the observations. While this is a high percentage, reflecting the low barriers to entry in the magazine industry, there are 284 observations spanning 21 different markets which experience at least 1 year without entry. The correlation matrix in Table 3 is a ‘within’ correlation matrix. That is, the correlations reflect only the variation within each magazine over time. This corresponds to the fixed-effects model which only exploits the variation within each panel (magazine). Despite the fact that the variables are centered around the magazine mean, in several cases the correlation between an interaction term and its components is very high. For example, the correlation between the incumbent age–entry interaction and incumbent age is 0.93, and the correlation between the markets–entry interaction and markets is 0.83. As I discuss below, these high correlations do not appear to cause multicollinearity. Table 4 reports results for the models used to test Hypotheses 1–4. Model 1 does not include any interaction terms. In this model, incumbent age has a negative and statistically significant effect on incumbent price, while the effects of corporate scope and market structure are statistically insignificant. Entry has a negative but statistically insignificant effect. This result fails to provide support for Hypothesis 1, suggesting that, on average, incumbent magazines do not cut prices in the face of entry. Model 2 includes the interaction of entry and incumbent age. In this model, the effect of incumbent age remains negative and statistically significant. The effect of entry is also negative and statistically significant, while the interaction of age and entry has a positive and statistically significant effect. These results provide support for Hypothesis 2, indicating that newer incumbents cut prices more in response to entry. Model 3 includes the interaction of entry and incumbent corporate scope. In this model, the effect of incumbent corporate scope becomes positive, but remains statistically insignificant, as does the effect of entry. The effect of the interaction term is negative and statistically significant, providing support for Hypothesis 3, and suggesting that multi-market incumbents cut prices more in response to entry.

ln(PRICEij kt ) = b1 Xij kt + ui + vj + wt + eij kt where i indexes magazines, j indexes publishers, k indexes markets, and t indexes years. In this model, PRICEij kt is the per-issue subscription price of incumbent magazine i, owned by incumbent publisher j , competing in market k, in year t. Xij kt is the set of independent and control variables described above. The ui is a set of magazine-specific dummy variables (magazine fixed effects), vj are publisher-specific dummy variables (publisher fixed effects), and wt are yearspecific dummy variables (year fixed effects). In the fixed-effects specification, only withinmagazine variation is used. In this model, the coefficient on entry compares a magazine’s price in those years in which it faces entry with its price in those years in which it does not face entry, averaging this difference across all magazines in the sample. A statistically significant coefficient on entry would indicate that incumbent magazines charge different prices in years when they face entry compared to years in which they do not face entry.9 I use a fixed-effects specification to control for unobserved differences across magazines, publishers, and time. Because each magazine only competes in one market, the magazine fixed effects also control for time-invariant differences in supply and demand conditions across markets. Moreover, publisher fixed effects control for the fact that some publishers may have already established reputations for fighting entry.

9 I also considered several different model specifications: (1) first-difference models; (2) models with a lagged dependent variable (DV); (3) models with entry as the DV and price (or lagged price) as an independent variable, in order to further explore the possibility of reverse causality; and (4) a model using 3-year moving averages, composed of the current year and the two previous years. In each of these four approaches, I examined a model without any interaction terms (comparable to Model 1 in the paper), and a model with all three interaction terms (comparable to Model 5 in the paper). Generally, the results are quite consistent across model specifications, and are quite consistent with those reported in the paper. There is little evidence of reverse causality, and the results are quite robust to different specifications.

Copyright  2005 John Wiley & Sons, Ltd.

Strat. Mgmt. J., 26: 1229–1248 (2005)

1238 Table 3. Variable

D. Simon Descriptive statistics and within-magazine correlation matrix Mean

S.D.

1.

2.

3.

4.

1. Pricea 0.66 0.50 2. Entry 0.92 0.27 0.01 3.14 0.95 0.06 0.08 3. Incumbent agea 0.92 0.91 0.06 0.03 0.28 4. Marketsa 5. Herfindahl 0.23 0.16 −0.05 0.01 −0.07 0.00 6. Market 12.11 16.69 0.15 0.01 0.24 0.04 circulation (millions) 7. Rival 19.71 14.00 0.19 0.00 0.26 0.08 magazines 8. Rival exit 0.12 0.32 0.01 −0.01 0.17 0.08 9. Experience −0.02 0.03 −0.03 −0.02 −0.22 −0.17 with entry 2.89 1.23 0.04 0.93 0.31 0.10 10. Incumbent agea ∗ Entry 0.87 0.91 0.03 0.39 0.27 0.83 11. Marketsa ∗ Entry 12. Herfindahl 0.20 0.16 −0.03 0.77 0.07 0.03 ∗ Entry 13. Experience 0.02 0.03 −0.03 0.13 −0.20 −0.16 with entry ∗ Entry a

5.

6.

7.

8.

9.

10.

11.

12.

−0.21 −0.27

0.65

−0.02 −0.08 −0.03 0.05 −0.17 −0.05 −0.07 −0.01

0.08

0.07

0.03 −0.07

0.01

0.04

0.07

0.07 −0.16 0.41

0.39 −0.11 −0.13

0.01

0.04 −0.17 −0.05 −0.07

0.00 0.70

0.30

0.94 0.07 −0.05 0.11

Logarithm. For correlations with absolute values greater than 0.04, p < 0.05.

Model 4 includes the interaction of entry and market structure. Here, both entry and market structure have positive and statistically significant effects. The interaction effect is negative and statistically significant, providing support for Hypothesis 4. This result suggests that incumbents in concentrated markets cut prices more following entry. Model 5 includes all three interaction terms. Of the main effects, incumbent age continues to exert a negative effect on incumbent price, while the effect of market concentration remains positive and statistically significant. The positive effect of corporate scope is statistically insignificant, as is the negative effect of entry. All three interaction terms continue to have statistically significant effects, and in the hypothesized directions. Taken together, these results provide strong support for Hypotheses 2–4. While the interaction effects in Models 2–5 are statistically significant, the R 2 in the interaction models is not much larger than in the basic model. However, a test for the joint significance of the three interaction terms in Model 5 yields an F statistic of 6.66, (significant at 0.0002). Although adding the interaction terms does not explain a lot of additional variance in incumbents’ pricing Copyright  2005 John Wiley & Sons, Ltd.

responses, the interaction terms do illuminate the relationships between the individual variables. To see this, compare the coefficient on entry in Models 1 and 2. In Model 1, the coefficient is −0.003, and is statistically insignificant, while in Model 2 the coefficient on entry is −0.125 and statistically significant. In other words, on average, an incumbent with 1 year of experience (i.e., where log incumbent age equals zero) reduces price by 12.5 percent in response to entry. Controlling for the differing incentives of old and new incumbents reveals the effect of entry on new incumbent prices to be substantial. Looking more closely at the magnitudes of these effects, the results of Model 2 indicate that while an incumbent magazine with 1 year of experience cuts price by 12.5 percent following entry, an incumbent with 11 years experience cuts its price by less than 4 percent when facing new entry,10 and a magazine with 21 years experience cuts its price by about 1 percent. Similarly, in Model 3, single-market incumbents do not reduce price post 10 I arrive at this number by taking ln(11) = 2.398 and multiplying this by the coefficient on the interaction term (0.037), and adding the product to the coefficient on entry, i.e., (2.398 ∗ 0.037) − 0.125 = −0.036.

Strat. Mgmt. J., 26: 1229–1248 (2005)

Incumbent Pricing Responses to Entry Table 4.

1239

Regression models for hypothesis tests

Entry Incumbent agea Marketsa Herfindahl Market circulation Rival magazines Incumbent agea ∗ Entry

Model 1b

Model 2b

Model 3b

Model 4b

Model 5b

Model 6b

−0.003 (0.011) −0.110∗∗ (0.017) −0.013 (0.013) 0.076 (0.055) 0.011∗∗ (0.003) −0.003∗ (0.001)

−0.125∗∗ (0.030) −0.141∗∗ (0.020) −0.015 (0.013) 0.075 (0.055) 0.010∗∗ (0.003) −0.003∗ (0.001) 0.037∗∗ (0.011)

0.023 (0.015) −0.110∗∗ (0.017) 0.021 (0.018) 0.075 (0.055) 0.011∗∗ (0.003) −0.003∗ (0.001)

0.036† (0.020) −0.108∗∗ (0.017) −0.014 (0.013) 0.164∗ (0.066) 0.010∗∗ (0.003) −0.003∗ (0.001)

−0.119∗ (0.049)

−0.038 (0.049) −0.132∗∗ (0.020) 0.015 (0.018) 0.152∗ (0.066) 0.010∗∗ (0.003) −0.003∗ (0.001) 0.028∗ (0.012) −0.032∗ (0.013) −0.103∗ (0.050)

3715 0.230

3715 0.234

−0.039 (0.048) −0.055∗ (0.021) −0.007 (0.018) 0.159∗ (0.066) 0.009∗∗ (0.003) −0.001 (0.001) 0.025∗ (0.011) −0.029∗ (0.013) −0.080† (0.046) −0.009 (0.009) 0.042 (0.374) −0.057 (0.375) 3335 0.238

Marketsa ∗ Entry Herfindahl ∗ Entry

−0.036∗∗ (0.013)

Rival exit Experience with entry Experience with entry ∗ Entry N Within R 2

3715 0.229

3715 0.232

3715 0.231

a

Logarithm Includes magazine, publisher, and year fixed effects Standard errors are reported in parentheses. †Significant at 0.10; ∗ significant at 0.05; b

entry, but an incumbent publisher competing in five markets cuts its price by 3.4 percent. In Model 4, a 10-percentage point increase in the Herfindahl index (on a 0–1 scale) results in incumbents cutting prices by an additional 1.2 percent post entry. In Model 6, I add additional control variables not included in the previous models: a dummy variable for rival exit, a measure of the incumbent’s experience with past entry, and its interaction with the entry dummy. The interaction term reflects the influence of the incumbent’s experience with entry on its response to entry. I include these variables separately because they each have missing values for each magazine’s first observation.11 The results of Model 6 are very similar to those of Model 5. Entry has a statistically insignificant

11

The exit variable takes missing values for the first year because it counts rival magazines that were in the dataset in the previous year that no longer appear. Incumbent’s experience with entry is missing for the first year because this variable measures experience up to, but not including the current year. Copyright  2005 John Wiley & Sons, Ltd.

∗∗

significant at 0.01

effect, while all three interaction effects are statistically significant. The effect of rival exit is negative, but statistically insignificant.12 Similarly, incumbent experience with entry has a statistically insignificant effect, as does its interaction with entry. Because of the high correlations between the interaction terms and the variables that make up the interactions, along with the modest increases in R 2 when the interaction terms are added, one might be concerned that multicollinearity is making the results in Table 4 unstable. To consider this, I split the sample at the median value of each of the three variables that I interact with entry and examine the effect of entry in each half of the sample. I report these results in Table 5. 12 I also consider a continuous measure of exit: the number of rivals that exit in the current year. The effect of number of rival exits is negative and statistically significant, suggesting that rival exits is picking up the effect of some unobserved change in market demand. The other results do not change in a meaningful way.

Strat. Mgmt. J., 26: 1229–1248 (2005)

1240 Table 5.

D. Simon Split sample regressions Model 7ab Low incumbent age (<24)

Entry Incumbent Agea Marketsa Herfindahl Market circulation Rival magazines N Within R2

−0.036∗ (0.017) −0.141∗∗ (0.025) 0.015 (0.018) 0.245∗∗ (0.081) 0.012∗∗ (0.004) −0.002 (0.002) 1857 0.293

Model 7bb High incumbent age (≥24) 0.023 (0.015) 0.280† (0.144) −0.060∗∗ (0.020) −0.123 (0.077) 0.008† (0.005) −0.005∗ (0.002) 1858 0.192

Model 8ab Low markets (≤2) 0.017 (0.016) −0.128∗∗ (0.029) 0.026 (0.035) 0.129† (0.076) 0.005 (0.006) −0.004† (0.002) 2033 0.135

Model 8bb High markets (>2) −0.032∗ (0.014) −0.092∗∗ (0.019) −0.014 (0.016) −0.069 (0.085) 0.013∗∗ (0.004) −0.002 (0.002) 1682 0.361

Model 9ab Low Herfindahl (≤0.187) 0.016 (0.027) −0.056∗ (0.026) −0.043∗ (0.021) −0.014 (0.377) 0.001 (0.005) 0.001 (0.002) 1852 0.200

Model 9bb High Herfindahl (>0.187) −0.003 (0.012) −0.201∗∗ (0.025) 0.002 (0.016) 0.113∗ (0.057) 0.009 (0.007) −0.001 (0.002) 1863 0.302

a

Logarithm Includes magazine, publisher, and year fixed effects Standard errors are reported in parentheses. †Significant at 0.10; ∗ significant at 0.05; b

For example, I first re-estimate Model 1 for those observations where incumbent age is less than the median value of 24. For these newer firms, the effect of entry is negative and statistically significant. By contrast, for those observations where incumbent age is greater than or equal to 24, the effect of entry is positive and statistically insignificant. These results parallel the results reported in Model 2, where the interaction of entry and incumbent age has a negative effect on incumbent price. Similarly, I split the sample at the median value of corporate scope. For focused publishers, those competing in one or two markets, the effect of entry is positive and statistically insignificant, while for broad publishers who compete in more than two markets the effect of entry is negative and statistically significant. These results parallel the results of Model 3; publishers competing in more markets cut prices more following entry. Finally, I split the sample at the median Herfindahl value (0.187). For those observations in concentrated markets, where the Herfindahl index is above the median value, the effect of entry is negative and statistically insignificant. But in competitive markets, where the Herfindahl index is less than or equal to the median value, the effect of entry is positive and statistically insignificant. Despite the statistical insignificance, the pattern of results is consistent with the results of Model 4; Copyright  2005 John Wiley & Sons, Ltd.

∗∗

significant at 0.01

incumbent pricing responses are less negative in competitive markets. Taken together, the results in Table 5 are highly consistent with those reported in Table 4, revealing the interaction effects to indeed capture important differences in how incumbents respond to entry. To better explain these results, Figure 1(a–c) provides graphical illustrations of the split sample results. I take the average price for each type of incumbent (old and new incumbents, incumbents owned by focused and broad publishers, and incumbents in competitive and concentrated markets), for those observations without entry. I then use the corresponding entry coefficient to compute the expected price when each type of incumbent faces entry. These figures show that new incumbents, incumbents owned by broad publishers, and incumbents competing in more concentrated markets all reduce price more than their respective comparison group, in response to entry.

Additional analyses Along with the incumbent and market characteristics discussed here, entrants’ characteristics may also influence incumbent pricing responses. Lieberman (1987b) and Thomas (1999) find some evidence that incumbents respond more aggressively to de novo entry than to incumbents that Strat. Mgmt. J., 26: 1229–1248 (2005)

Incumbent Pricing Responses to Entry

1241

2.70

Price

2.60 2.50

Old New

2.40 2.30 2.20 No Entry

(a)

Entry

2.60

Price

2.50 Focused Broad 2.40

2.30 No Entry

(b)

Entry

2.50

Price

2.40 Competitive Concentrated 2.30

2.20 (c)

Figure 1.

No Entry

Entry

Entry response by incumbent type (graphs of split sample results). (a) Old and new incumbents. (b) Focused and broad publishers. (c) Competitive and concentrated markets

expand existing operations or introduce new products. To consider whether incumbents respond more aggressively to de novo entrants than to incumbent entrants, I separate entry into two components: de novo entry and incumbent entry. De novo entry indicates entry by a publisher with no incumbent magazines, while incumbent entry refers to a new magazine offered by a publisher that offers at least one incumbent magazine. Table 6 reports these results. In Model 10, the effect of incumbent entry is negative, but statistically insignificant, as is the effect of de novo entry. In Model 11, I include a dummy variable for whether the incumbent competes in more than one market. In addition, I interact this variable with the two types of entry. The results indicate that multi-market incumbents Copyright  2005 John Wiley & Sons, Ltd.

cut prices in response to de novo entry by significantly more than single-market incumbents. On the other hand, multi-market incumbents did not respond differently to incumbent entry. Taken together, these results suggest that the type of entry moderates the negative effect of incumbent corporate scope on incumbent pricing responses; multimarket incumbents seek to build a reputation for fighting de novo entry but not incumbent entry. Along with the type of entry, multi-market contact between entrants and incumbents may also moderate the negative effect of corporate scope on incumbent pricing responses. To consider this, I construct a measure of multi-market contact between incumbents and entrants which counts the number of times in which the entrants into a market meet the market’s incumbents, across all markets other than the focal market. If either Strat. Mgmt. J., 26: 1229–1248 (2005)

1242 Table 6.

D. Simon Additional analyses Model 10b

Model 11b

Model 12b

Model 13b+ Ad rate

Model 14b Ad rate

−0.110∗∗ (0.017) −0.013 (0.013) 0.076 (0.055) 0.011∗∗ (0.003) −0.003∗ (0.001)

−0.109∗∗ (0.017) −0.017 (0.016) 0.075 (0.055) 0.011∗∗ (0.003) −0.003∗ (0.001)

0.027† (0.016) −0.108∗∗ (0.017) −0.017 (0.016) 0.073 (0.055) 0.010∗∗ (0.003) −0.003∗ (0.006)

0.001 (0.012) 0.111∗∗ (0.019) 0.001 (0.013) −0.064 (0.059) 0.001 (0.003) −0.0004 (0.0015)

−0.005 (0.011) −0.0004 (0.0068)

−0.082 (0.052) 0.094∗∗ (0.022) −0.005 (0.019) −0.100 (0.071) 0.001 (0.003) −0.0004 (0.0015) 0.019 (0.012) 0.005 (0.014) 0.047 (0.053)

0.025 (0.016) −0.010 (0.010) 0.053† (0.031) −0.058∗∗ (0.022) 0.015 (0.013)

0.592∗∗ (0.018) 3638 0.686

0.591∗∗ (0.018) 3638 0.686

Entry Incumbent Agea Marketsa Herfindahl Market circulation Rival magazines Incumbent agea ∗ Entry Marketsa ∗ Entry Herfindahl ∗ Entry De novo entry Incumbent entry Multi-market incumbent Multi-market incumbent ∗ De novo entry Multi-market incumbent ∗ Incumbent entry Multi-market incumbent ∗ Entry Multi-market contact ∗ 10 Multi-market incumbent ∗ Multi-market contact ∗ 10

0.058† (0.031)

−0.058∗∗ (0.022) −0.004 (0.002) 0.004 (0.003)

Circulationa N Within R 2

3715 0.229

3715 0.231

3715 0.231

a

Logarithm Includes magazine, publisher, and year fixed effects Standard errors are reported in parentheses. †Significant at 0.10; ∗ significant at 0.05; b

the entrants into a market or the incumbents in that market do not compete in any other markets, then the multi-market contact variable takes a value of zero. But, if one magazine enters into the women’s market, and its publisher also competes in the sports market, where one of the incumbent women’s magazine publishers also competes, then multi-market contact in the women’s market will be one (because an entrant and an incumbent meet once outside the focal market). Copyright  2005 John Wiley & Sons, Ltd.

∗∗

significant at 0.01

To examine the moderating effect of multimarket contact, I interact multi-market contact with the multi-market incumbent dummy. The results of Model 12 show that multi-market incumbents cut prices by significantly more than singlemarket incumbents in response to entry. Moreover, there is weak evidence that multi-market contact moderates the aggressive pricing response of multi-market incumbents; the interaction of multi-market contact and multi-market incumbent Strat. Mgmt. J., 26: 1229–1248 (2005)

Incumbent Pricing Responses to Entry Table 7.

Frequency of price cuts and advertising rate cuts in response to entry

Observations

Ad rate: cut

Ad rate: no change or raise

Magazine price: cut Magazine price: no change or raise Totals

848 190 1038 (31%)

1610 711 2321 (69%)

has a positive effect that approaches conventional significance levels (p = 0.125). While this study focuses on incumbents’ subscription prices, one might suspect that incumbents also respond to entry by reducing advertising rates. Although I argue above that incumbents have a greater incentive to cut subscription prices due to the effect that this has on ad revenues, advertising space is a more homogeneous product than magazine subscriptions, and therefore demand is likely more elastic with respect to price. As a result, publishers may also reduce ad rates in response to entry. To consider this possibility, I estimate a basic model and an interaction model of the magazine’s advertising rate. In both models I control for the magazine’s circulation. Models 13 and 14 in Table 6 report the results. In the basic model, the effect of entry is statistically insignificant, as are the effects of market concentration and corporate scope. The effect of incumbent age is positive and statistically significant. This may reflect older magazines’ ability to attract readers that are more valued by advertisers. It may also indicate that advertisers will pay more to advertise in older magazines because they have more information about the characteristics of the readers of older magazines. When I include the interaction terms, the negative effect of entry (p = 0.11) and the positive effect of the entry–incumbent age interaction (p = 0.12) both approach conventional levels of statistical significance, suggesting that newer incumbents respond to entry more aggressively than older incumbents. The other two interaction effects are statistically insignificant. To better understand these results, I examine the frequency of price cutting and ad rate cutting when facing entry.13 The results are reported in Table 7. The results indicate that, out of 3359 cases, magazines cut prices 2458 times (73%) 13

1243

I adjust nominal prices and ad rates for inflation.

Copyright  2005 John Wiley & Sons, Ltd.

Total 2458 (73%) 901 (27%) 3359 (100%)

when they faced entry, while they cut ad rates in only 1038 of these cases (31%). Combined with the results of Models 13 and 14, it appears that magazines are more likely to cut subscription prices in response to entry. Of the 2458 cases in which magazines cut prices in response to entry, they also cut advertising rates 848 times (34%). On the other hand, for the 901 cases where they did not cut prices in response to entry, they cut advertising rates only 190 times (21%). These results suggest that magazines do not use ad rate reductions as substitutes for price cuts in responding to entry. If anything, it appears that magazines use these two types of prices as complementary entry response mechanisms, as suggested by Gatignon and Hanssens (1987).

DISCUSSION Economic theory suggests that incumbent firms may cut prices after entry, either to deter future entrants or to maximize current profits. But, despite solid theoretical support, empirical research has produced inconsistent results: some studies find a negative relationship between entry and pricing, while this study and others do not. Regarding the direct effect of entry, the results of this study indicate that entry does not have a statistically significant effect on incumbents’ pricing. One industry-specific explanation for the lack of incumbent response is the high rate of entry in the magazine industry; the mean number of new entrants into a market in the sample is nine. Because entry is almost continuous, it may be that the threat of entry keeps prices low, even in the absence of entry (Hannan, 1979; Cool, Roller, and Leleux, 1999). In other words, the magazine industry may be a contestable market in which incumbents set low prices in anticipation of entry, regardless of whether or not it actually occurs. Frank and Salkever (1992) offer another possible answer to this question, but it is also somewhat Strat. Mgmt. J., 26: 1229–1248 (2005)

1244

D. Simon

industry specific. In explaining the positive correlation between brand name prices and entry by generic manufacturers, they argue that generic entry attracts price-sensitive consumers, making the residual demand curve for brand name incumbents more price inelastic. This allows brand name manufacturers to raise prices on the less pricesensitive market. But, as the authors note, such an explanation is only applicable in cases where there is clear market segmentation. Yamawacki (2002) argues that the heterogeneity in incumbent response is a function of incumbents’ varying ability to respond to entry. Many in the marketing literature argue that product differentiation moderates incumbents’ pricing responses. This study offers a more general explanation for the selective response to entry by incumbent firms: incumbents vary in their incentive to respond to entry. Incumbents with greater incentives to respond to entry are more likely to respond aggressively. To consider this explanation, this study assesses the effect on incumbent pricing responses of three factors that may influence the incumbents’ incentive to respond to new entry: incumbent age, incumbent corporate scope, and market concentration. While the average effect of entry was not statistically significant, these three factors were found to significantly influence incumbents’ pricing response to entry. The results indicate that while incumbents reduce prices over time, an incumbent’s time in the market weakens its pricing response to entry; newer incumbents cut prices more in response to entry than do older ones. While the negative effect of incumbent age on prices may suggest a liability of age, when considered jointly with the positive age–entry interaction effect, it appears that older incumbents have greater know-how that allows them to cut costs and in turn reduce prices below those of newer rivals. These lower costs and lower prices protect older incumbents from the threat posed by new entrants, and may explain both the negative impact of incumbent age as well as the positive age–entry interaction effect. However, there are two alternative explanations for the negative coefficient on incumbent age. First, reflecting the fact that entry occurs almost continuously in the consumer magazine industry, incumbent age may capture the effect of past entry. Therefore, incumbents may reduce prices over time because they face entry in most years. Copyright  2005 John Wiley & Sons, Ltd.

Other analysis suggests that age may also be picking up some of the effects of macroeconomic and industry-specific time trends measured by the year dummy variables. While it is difficult to identify the direct effect of age, the interaction effects along with the results of the split-sample analysis appear to provide robust evidence that older incumbents respond less aggressively to entry than new incumbents. Combined with theoretical arguments suggesting that older incumbents may have accumulated greater know-how, the results would seem to suggest that older incumbents enjoy lower costs which they exploit by reducing prices, thereby limiting their need to reduce prices when facing new entry. This is also consistent with evidence from prior studies that newer firms are more vulnerable to the threat of entry (Geroski, 1995), as well as evidence that the survival rate of new magazines is very low; roughly 70 percent of all new magazines survive less than 5 years (Husni, 1997). It appears that entry threatens the survival of younger incumbents, forcing them to respond aggressively by reducing prices. The results of this study also provide evidence that multi-market incumbents fight entry in one market as a way to deter entry in their other markets. The number of markets in which a publisher competes increases the amount by which it cuts prices following entry. This is consistent with game-theoretic models which argue that firms may cut prices aggressively in response to entry, as a way to develop a reputation for fighting entry. Firms competing in multiple markets gain more from developing such a reputation, as they can deter potential entrants in all of the markets in which they compete. It should be noted that the ability to develop a reputation in other markets assumes that the markets are linked in some way (Milgrom and Roberts, 1982b), such that potential entrants are likely to be aware of the incumbent’s responses in other markets, and such that entrants view behavior in one market as predictive of behavior in other markets. Therefore, while this relationship might be expected to hold for geographically diversified firms like airlines or banking, it would be less likely to apply to firms that are diversified into different product markets. Entrants may not be aware of the incumbent’s behavior in other product markets, or they may view the markets as sufficiently Strat. Mgmt. J., 26: 1229–1248 (2005)

Incumbent Pricing Responses to Entry different that incumbent behavior in one market does not predict behavior in another market. This relationship is counter-intuitive, challenging the notion that firms competing in only one market are more likely to fight because they have fewer alternatives. As noted above, this argument only applies if the cost of exit is greater for singlemarket incumbents. For this to be true, there must be costs associated with exiting the market, and/or sunk costs associated with entering a new market, which only the single-market incumbent would need to incur (the multi-market incumbent having already done so). In the consumer magazine industry it appears that the cost of exiting the market tends to be low because most of a publisher’s critical resources reside at the magazine level, in the form of the title and magazine layout, along with the publisher and editor. For this reason, publishing firms frequently buy and sell individual magazines. This suggests that exiting incumbents can sell their magazines to existing rivals. The cost of entering a market also seems to be fairly low for an existing publisher. First, the average publisher in the sample competes in four markets. Second, in the entry data many publishers enter multiple markets. Third, the fixed costs of printing, distribution, marketing, and administration can all be spread over titles and markets (Round and Bentick, 1997). Taken together, this suggests that the magazine industry has low mobility barriers, consistent with the high entry rates described above. Under these conditions, singlemarket incumbents incur little additional cost to exit a market. While the results indicate that multi-market incumbents respond more aggressively to entry, additional analysis suggests that this effect is moderated by two factors: the type of entry and multi-market contact. Multi-market incumbents cut prices more than single-market incumbents in response to de novo entry, but less than singlemarket incumbents in response to entry by incumbents. These results are consistent with Thomas (1999), who finds that large incumbents cut price in response to de novo entry while accommodating entry by incumbents. These results suggest that it is more difficult to deter incumbent entrants. Similarly, it appears that multi-market incumbents cut prices less in response to entry when they meet the entrants in other markets. This finding is consistent with the multi-market competition literature, which Copyright  2005 John Wiley & Sons, Ltd.

1245

stresses that firms competing against each other in multiple markets are more likely to recognize their mutual interdependence and avoid price wars (Gimeno and Woo, 1999). Finally, the results indicate that incumbents in concentrated markets respond more aggressively to new entry, cutting prices by more than incumbents in competitive markets. While it may seem counter-intuitive that firms respond more aggressively in less competitive markets, these firms have a greater incentive and a greater need to cut prices in response to entry. Firms earning monopoly rents have more to lose by not fighting entry, and more to gain by deterring future entry. This result is consistent with Hannan (1979), who finds that banks in concentrated markets are more likely to cut prices pre-entry to deter entrants, and with Lieberman (1987b), who finds that incumbents in concentrated markets increase capacity by a greater amount following entry As noted above, incumbents face incentives to cut price in response to entry both to maximize current profits and to deter future entry. New entrants threaten incumbents’ market share and reduce their ability to coordinate pricing decisions. But incumbents also may consider future profitability and determine that cutting prices, even beyond what maximizes current profits, may be optimal if future entrants are deterred. Previous studies have not distinguished between these two incentives for reducing prices following entry: Do firms cut prices in response to entry to maximize current profits or to deter future entry? This study’s results shed some light on this question. Finding that newer incumbents cut prices by more than older incumbents suggests that newer incumbents reduce prices to maximize current profits, because there is no reason to believe that newer incumbents would be more concerned than older incumbents with entry deterrence. Therefore, the marginal effect of entry on new incumbents’ prices should be attributable to the current profit maximization explanation. On the other hand, the interaction effect of corporate scope and entry suggests that multi-market firms cut prices to deter future entrants. Thus, the results of this study suggest that each incentive affects incumbent behavior under certain circumstances. Geroski (1995) comments that the absence of a pricing response post entry could either indicate accommodation by the incumbent or a limit-pricing strategy in which prices are reduced pre-entry. Strat. Mgmt. J., 26: 1229–1248 (2005)

1246

D. Simon

Again, empirical studies have not distinguished between these two explanations for the absence of a pricing response found in many studies. These results suggest that selective responses do not reflect limit-pricing, but rather that incumbents with weak incentives choose not to respond to entry. Those incumbents with less incentive to cut prices following entry also have weaker incentives to limit price. One concern with this study is that magazines may respond to entry with non-price weapons. For example, magazines may change the editorial format, or they may upgrade the quality of the paper used, moving to a glossier finish. Similarly, they may increase the number of pages per issue. These non-price responses may tend to be cost increasing, leading to price increases. As a result, these non-price responses may cause a spurious, positive relationship between entry and price. This may contribute to the absence of a direct relationship between entry and incumbent prices. Another concern in doing this study is the potential endogeneity of the entry variable. There are two potential sources of endogeneity in the entry variable. First, causality may work in the opposite direction, as a potential entrant’s decision to enter a market is likely to be influenced by the anticipated response of incumbents. As discussed above, firms may develop a reputation for aggressiveness that deters entry (Kreps and Wilson, 1982; Milgrom and Roberts, 1982b). Second, demand conditions may be correlated with both entry and incumbent prices. For example, if there is excess demand, then incumbents may set high prices, which will attract entry. I partially control for both of these sources of endogeneity. Fixed effects control for timeinvariant sources of endogeneity. Specifically, magazine and publisher fixed effects control for differences across magazines and publishers in their pricing behavior. Market circulation imperfectly controls for time-varying demand conditions that may affect entry and pricing. However, there may still be unobserved demand factors that are correlated with both pricing and entry decisions, creating a potential feedback from prices to entry. In this case, these sources of endogeneity should bias the estimated coefficient on entry, upwards, toward zero, perhaps helping to explain the absence of a statistically negative entry coefficient in most models. Copyright  2005 John Wiley & Sons, Ltd.

While the effect of entry may be biased, the interaction effects will be unbiased as long as the unobserved factors that are correlated with both pricing and entry have an equal impact across different types of publishers. That is, even if the ‘true effect’ of entry is more negative than that reflected in the estimated coefficients on entry, the interaction coefficients should provide good estimates of the differential effect of entry for different types of incumbents, as long as the unobserved demand shifters affect the different types of incumbents equally. One might also be concerned that changes in supply conditions, for example a reduction in costs due to a technological innovation, could induce entry and could also lead to lower prices, yielding a downward bias in the coefficient on entry. But, this is unlikely because factors that reduce production costs would be likely to affect all magazine publishers, regardless of the type of magazine that they produce. This effect would be captured by the year fixed effects. As a result, I do not believe that unobserved supply conditions pose a substantial threat to the qualitative nature of the findings discussed above.14 As the above arguments suggest, there are benefits to conducting this study in a single-industry setting. Using an industry such as the consumer magazine industry, which comprises many different markets, allows me to assess the effects of variation in market entry conditions, while controlling for a variety of other factors that might be correlated with both entry and pricing behavior, such as supply conditions. But, as with any single-industry study, there are concerns about generalizability. This may be a particular concern in this study due to the high rates of entry in the consumer magazine industry. Low entry barriers and the resulting high entry rates may partially explain the absence of a pricing response across all incumbents. On the other hand, as noted above, publishers have an extra incentive to cut magazine prices because of the positive effect this has on advertising revenues.

14 Most studies of incumbent pricing response to entry estimate models similar to the ones that I estimate here, using fixed effects, but ignoring any remaining endogeneity (Thomas, 1999; Marion, 1998; Hannan, 1979). Moreover, two studies that use instrumental variables estimation to control for the remaining endogeneity find that the results are not greatly affected (Frank and Salkever, 1997; Yamawacki, 2002).

Strat. Mgmt. J., 26: 1229–1248 (2005)

Incumbent Pricing Responses to Entry It would be interesting to see whether similar results are found in other industries including airlines, banks, and mutual funds, which, like magazines comprise many different markets, but have much lower entry rates. Anecdotal evidence from the airline industry indicates that incumbents do respond aggressively to entry (Carey, 2002).

CONCLUSION A great deal of research, theoretical and empirical, has examined the use of prices to deter entry. While theory suggests that firms often have incentives to cut prices to deter entry, empirical results have generally failed to find evidence of pricecutting after entry. This study provides evidence that several factors influence the incumbents’ response to entry. Results show that new incumbents respond more aggressively to entry, as a way to survive. Similarly, multi-market firms cut prices in response to entry, as a way to deter entry in other markets. Finally, firms in concentrated markets have a greater incentive to use price to deter entry, as a way to protect oligopoly profits. While this study examines the factors that influence incumbent pricing responses to entry, an interesting opportunity for future research lies in examining the effect of incumbents’ entry deterrence efforts. To what extent does price-cutting actually deter potential entrants from entering? To what extent does price-cutting prevent entrants from stealing incumbents’ market share? And, how successful are entrants in gaining market share in the face of incumbent price-cutting? Understanding the answers to these questions is important for managers and policy-makers alike.

REFERENCES Bain J. 1956. Barriers to New Competition. Harvard University Press: Cambridge, MA. Bresnahan T, Reiss B. 1991. Entry and competition in concentrated markets. Journal of Political Economy 99: 977–1009. Carey S. 2002. Can a big carrier book profit while battling a nimble rival? Wall Street Journal 240: B1. Cecchetti S. 1986. The frequency of price adjustment: a study of newsstand prices of magazines. Journal of Econometrics 31: 255–274. Cool K, Roller L, Leleux B. 1999. The relative impact of actual and potential rivalry on firm profitability Copyright  2005 John Wiley & Sons, Ltd.

1247

in the pharmaceutical industry. Strategic Management Journal 20(1): 1–14. Daly C, Henry R, Ryder E. 1997. The Magazine Publishing Industry. Allyn & Bacon: Boston, MA. Dierickx I, Cool K. 1989. Asset stock accumulation and sustainability of competitive advantage. Management Science 35: 1504–1511. Dixit A. 1980. The role of investment in entry deterrence. Economic Journal 90: 95–106. Dolge A. 1911. Pianos and their Makers. Covina Publishing: Covina, CA. Evans J, Sigfried L. 1994. Empirical studies of entry and exit: a survey of the evidence. Review of Industrial Organization 9: 121–155. Fombrun C, Shanley M. 1990. What’s in a name? Reputation-building and corporate scope. Academy of Management Journal 33: 233–258. Frank R, Salkever D. 1992. Pricing patent loss and the market for pharmaceuticals. Southern Economic Journal 59: 165–179. Frank R, Salkever D. 1997. Generic entry and the pricing of pharmaceuticals. Journal of Economics and Management Strategy 6: 75–90. Gatignon H, Hanssens D. 1987. Modeling marketing interactions with application to salesforce effectiveness. Journal of Marketing Research 29: 247–257. Geroski P. 1995. What do we know about entry? International Journal of Industrial Organization 13: 421–440. Gimeno J, Woo C. 1999. Multimarket contact, economies of scope, and firm performance. Academy of Management Journal 42: 239–259. Gruca T, Kumar R, Sudarshan D. 1992. An equilibrium analysis of defensive response to entry using a coupled response function model. Marketing Science 11: 348–358. Hannan T. 1979. Limit pricing and the banking industry. Journal of Money, Credit and Banking 11: 438–446. Hauser J, Shugan M. 1983. Defensive marketing strategies. Marketing Science 2: 319–360. Husni S. 1991. Samir Husni’s Guide to New Consumer Magazines. University of Mississippi: Oxford, MS. Husni S. 1992–1994. Samir Husni’s Guide to New Consumer Magazines. Folio: Stamford, CT. Husni S. 1995–1996. Samir Husni’s Guide to New Consumer Magazines. Hearst Magazine Enterprises: New York. Husni S. 1997–2000. Samir Husni’s Guide to New Consumer Magazines. Oxbridge Communications: New York. Husni S. 2001. Samir Husni’s Guide to New Consumer Magazines. RR Bowker: New Providence, NJ. Joskow A, Werden G, Johnson R. 1994. Entry, exit, and performance in airline markets. International Journal of Industrial Organization 12: 457–471. Kessides I. 1990. Towards a testable model of entry: a study of the U.S. manufacturing industries. Economica 57: 219–238. Kotha S, Dunbar R. 1996. Steinway & Sons. In Strategic Management: Competitiveness and Globalization (3rd edn), Hitt M, Ireland D, Hoskisson R (eds.) SouthWestern College Publishing: New York; C461–C481. Strat. Mgmt. J., 26: 1229–1248 (2005)

1248

D. Simon

Kreps D, Wilson R. 1982. Reputation and imperfect information. Journal of Economic Theory 27: 253–279. Kumar R, Sudarshan D. 1988. Defensive marketing strategies: an equilibrium analysis based on decoupled response function models. Management Science 23: 405–415. Lieberman M. 1987a. Excess capacity as a barrier to entry: an empirical appraisal. Journal of Industrial Economics 35: 365–378. Lieberman M. 1987b. Post entry investment and market structure in the chemical processing industries. Rand Journal of Economics 18: 533–549. Marion B. 1998. Competition in grocery retailing: the impact of a new strategic group on BLS price increases. Review of Industrial Organization 13: 381–399. Masson R, Shaanan J. 1982. Stochastic-dynamic limiting pricing: an empirical test. Review of Economics and Statistics 64: 413–422. Masson R, Shaanan J. 1986. Excess capacity and limit pricing: an empirical test. Economica 53: 365–378. Milgrom P, Roberts J. 1982a. Limit pricing and entry under incomplete information: an equilibrium analysis. Econometrica 50: 443–459. Milgrom P, Roberts J. 1982b. Predation, reputation and entry deterrence. Journal of Economic Theory 27: 280–312. Modigliani F. 1958. New developments on the oligopoly front. Journal of Political Economy 66: 215–232. Nonaka I. 1994. Dynamic theory of organizational knowledge creation. Organization Science 5: 14–37. Papazian E. 1997. Magazine Dimensions ‘98 . Media Dynamics: New York.

Copyright  2005 John Wiley & Sons, Ltd.

Round D, Bentick T. 1997. Magazine subscription discounts in Australia. Review of Industrial Organization 12: 555–577. Singh S, Utton M, Waterson M. 1997. Strategic behavior of incumbent firms in the UK. International Journal of Industrial Organization 16: 227–251. Smiley R. 1988. Empirical evidence on strategic entry deterrence. International Journal of Industrial Organization 6: 167–180. Smith K, Grimm C, Gannon M. 1992. Dynamics of Competitive Strategy. Sage: Newbury Park, CA. Spence M. 1977. Entry, capacity, investment and oligopolistic pricing. Bell Journal of Economics 8: 534–544. Stinchcombe A. 1965. Social structure and organizations. In Handbook of Organizations, March JG (ed). RandMcNally: Chicago, IL; 142–193. Sylos-Labini S. 1962. Oligopoly and Technical Progress. Harvard University Press: Cambridge, MA. Thomas L. 1999. Incumbent firms’ response to entry: price, advertising, and new product introductions. International Journal of Industrial Organization 17: 527–555. Williamson O. 1979. Transaction-cost economics: the governance of contractual relations. Journal of Law and Economics 22: 233–261. Windle R, Dresner M. 1995. Competitive responses to low cost carrier entry. Transportation Research: Part E: Logistics and Transportation Review 35: 59–75. Yamawacki H. 2002. Price reactions to new competition: a study of U.S. luxury car market, 1986–1997. International Journal of Industrial Organization 20: 19–39.

Strat. Mgmt. J., 26: 1229–1248 (2005)