You are on page 1of 19

MARKETING SCIENCE

Vol. 27, No. 4, July-August 2008, pp. 610-626 ISSN 0732-23991 EISSN 1526-548X1081270410610 DOI 10.1287/mksc.l070.0313 ©2008 INFORMS

Digital Piracy: A Competitive Analysis
Sanjay Jain
Mays Business School, Texas A&M University, College Station, Texas 77843, sjain@mays.tamu.edu

I

n recent years, the issue of copyright protection for intellectual properties such as computer software, music CDs, and videos has become increasingly important. It is often claimed that illegal copying of intellectual property costs companies billions of dollars in lost revenues and reduces firms' incentives to innovate. Some researchers have shown that copying can be beneficial to firms when there are strong network effects and copying expands the market. In this paper, we first examine the impact of illegal copying of software and other similar intellectual properties on firms' prices, profits, and quality choices, even when there are no network effects and the market is saturated. We show that contrary to the claims of manufacturers, there are conditions under which copying can increase firms' profits, lead to better quality products, and increase social welfare. This is because weaker copyright protection enables firms to reduce price competition by allowing price-sensitive consumers to copy. Thus, weaker copyright protection can serve as a coordination device to reduce price competition. We also examine how equilibrium copyright enforcement is affected by network externalities. In contrast to previous research, we show that strong network effects can sometimes lead to a firm choosing higher levels of copyright protection. Our results show that in the presence of strong network effects, stronger copyright enforcement by one firm can serve as a coordinating device to reduce price competition. Key words: piracy; pricing; innovation; network effects; game theory History: This paper was received October 5, 2005, and was with the author 14 months for 2 revisions; processed by Jinhong Xie. Published online in Articles in Advance May 7, 2008.

1. Introduction
Illegal reproduction of intellectual properties such as computer software, books, music CDs, videos, etc. is increasingly becoming a major concern for both industry practitioners and the government. The 2005 Clobal Software Piracy Report commissioned by the Business Software Alliance estimates that 35% of software is pirated, which leads to an estimated loss of more than $31 billion to firms. Software piracy rates are as high as 92% in Vietnam and 90% in China. In the United States, piracy rates are estimated to be 21%. The study concludes that software piracy is one of the industry's worst problems. However, copying of intellectual property is not limited to software. In fact, piracy of music and movies is a major concern to the entertainment industry, and some researchers have suggested that it will lead to radical changes in the industry (Moul 2006). The losses because of music piracy alone are estimated to be more than $10 billion (Murphy 2003). With the growth of the Internet, piracy is becoming even more prevalent because copying of intellectual properties is becoming easier and more difficult to prevent. For example, peer-topeer networks such as Kazaa, BitTorrent, and others can enable consumers to illegally download a variety of software, music, and movies for free. Many industry analysts see piracy as one of the key threats to profitability and irmovation. They claim
610

that piracy leads to higher prices for legitimate users, lower profits for the firms, reduced new product innovation, and is generally harmful to society (see, for example, BSA 2004). A 2003 survey of Business Software Alliance company's CEOs reveals that they believe that piracy is the number one hindrance to software innovation. To deter piracy, industry associations such as the Recording Industry Association of America are beginning to aggressively prosecute pirates. Some companies have also tried to combat piracy by making their products more difficult to copy using digital rights management software. For example. Intuit incorporated a feature in its 2003 TurboTax such that the software could only be installed on one computer and Windows XP has an activation code. Previous research has examined the role of copying and its impact on profits. This research has tried to identify conditions under which copying can lead to higher profits for the firm and the optimal level of copyright protection (e.g., Conner and Rumelt 1991, Civon et al. 1995, Hui and Png 2003). The general conclusion of prior research is that piracy could be beneficial if there are strong network effects, and some consumers who pirate would not have bought the product in any case. The intuition for the result is that network effects can significantly increase firm's market size and the consumers who copy the product therefore can benefit the firm by increasing the

The rest of the paper is organized as follows. Our results suggest that the role of copyright protection depends on the strength of network externalities. we describe our model and examine the implications of copying on firm's prices. Furthermore. The firms are on the opposite ends of the Hotelling line. that piracy is necessarily harmful in situations where network effects are weak and all pirates would buy absent the ability to copy. can lead to reduced price competition while the opposite may hold in the case of strong network externalities. weak copyright protection. The parameter a. Takeyama (1994) also shows that copying can lead to higher profits and social welfare in the presence of network externalities. Their models assume that copying by individuals confers a network externality that benefits users of the product. we discuss the relevant literature. represents the level of piracy for firm ¿'s product. To the extent that firms' policies can impact piracy levels. In contrast to the prior literature. In this paper. our results can provide insights into the pricing and R&D choices that firms should make. we show that as network externalities become stronger. Although direct network effects may be important for many software products. 2. we reexamine conventional wisdom. the potential for piracy to increase market size will be limited (see also Li 2005). The answers to these questions are important from both managerial and public policy perspective. ©2008 INFORMS 611 size of the potential market. Thus. Their results suggest that if network effects are large. In §3. In particular. we develop a game-theoretic model in which there are two firms in the market. as markets for digital products become more mature. we show that in mature markets when network externalities are strong. some consumers are willing to pay more for the product when the product is copied. From a public policy perspective. In particular. In §2. Our analysis of the impact of piracy on innovation strategy shows that piracy can sometimes lead to higher levels of innovation and improved social welfare. These results therefore show that strong network externalities are neither necessary nor sufficient for firms to choose weak copyright protection. However. we show that piracy can lead to reduced price competition. innovation levels.^ Intuition would suggest that piracy reduces the incentives for innovation and would necessarily lead to lower quality products. In §§4 and 5. we extend the model to consider various cases to see whether our model is robust to changes in assumptions. and therefore do not consider pirated versions to be substitutes for the originals. Absent network externalities.Jain: Digital Piracy: A Competitive Analysis Marketing Science 27(4). firms can choose weak copyright protection in equilibrium. The market consists of two types of consumers: the first type of consumers do not copy the product because they place much lower value on copies either because of the unavailability of support for pirated versions or the fact that they find copying morally unacceptable. 610-626. This reduction in ' In this paper. understanding the impact of piracy on social welfare is critical for the government to develop sound copyright enforcement policies. We show that even with absent network effects. price competition can sometimes lead to higher profits for the firm when piracy increases. copyright enforcement can help reduce price competition. We also reexamine the intuition that strong network effects lead to weaker incentives for firms to impose copyright protection. we assume that firm i can choose copyright protection level a¡ such that only a¡ proportion of these consumers can copy its product. Conner and Rumelt (1991) examine the impact of copying software on firms' profits. pp. but only when copying increases the market size. The second group of consumers consists of those who might copy the product. and profits. Related Literature Previous researchers have examined how piracy can affect firms' profits. In a monopoly setting. Our results show that piracy can change the composition of the market and alter strategic interaction between firms. copying can increase firms' profits by increasing the market size. one firm begins to more strongly enforce copyright protection. In other words. the value that consumers place on the product increases (see Katz and Shapiro 1985 for a discussion of network externalities). Section 6 concludes the paper with a discussion of the implications of the main results and directions for future research. Furthermore. we also study the impact of copying on the innovation process. This is because weak copyright enforcement enables firms to coordinate to charge higher prices from the segment of consumers that does not copy. Our research attempts to formally address how piracy impacts the level of innovation by firms and how firms should choose copyright protection levels taking these into account. In particular. . as the number of users (which includes people who copy the product) increases. These consumers consider pirated versions to be equal substitutes for the originals. it is important for the firms to understand how it should manage copyright protection. their ability to copy would depend on the firm's copyright protection policy. We then examine the impact of network effects on incentives to choose copyright protection. we use the terms piracy and copying interchangeably. To address these issues.

Conner (1995) shows that in the presence of network effects. Gu and Mahajan (2005) also examine how piracy affects firms' profits in ^ Another stream of research shows that if there are no network effects but firms are able to price discriminate. In their model. Thus. Thus. Givon et al. Finally. and usually choose among the available alternatives. By using data for spreadsheets and word processing software sales in the United Kingdom. Varian 2005). and can be beneficial to firms when markets have high wealth gaps. they find that piracy helped the diffusion of software and 80% of new software purchases were influenced by pirates. as they do in our framework. Using a diffusion-based model. pp. Xie and Sirbu (1995) show that the presence of network effects can encourage firms to share technology. Arrow 1962. a monopolist might benefit from entry by a clone. this implies that in their formulation. allowing piracy by both firms can intensify price competition. Furthermore. First. Novos and Waldman 1984. we also examine the competitive effect. unlike their paper. this research assumes that copies cannot be made from copies. In the context that we study. we endogenize copyright protection. so that products diffuse faster and a monopolist may therefore prefer entry by a compatible competitor. the presence of competition. Second. Modeling competition in the context that we study is important because most digital products compete with other products. then copying can iead to higher profits (Liebowitz 1985. for example. First. For example. Our work on the impact of copying on innovation is also related to the vast literature on intellectual property rights (see. we allow firms to set innovation levels. Word competes with WordPerfect. The producer can then charge the libraries an amount proportional to the amount of copying that occurs in the library. (1995) empirically demonstrate that piracy can help diffusion of legal copies. In our paper. This literature attempts to determine the optimal breadth and length of patents to ensure that firms have sufficient incentives to invest in innovation. firms allowing piracy increases the market size. Sun et al.612 Jain: Digital Piracy: A Competitive Analysis Marketing Science 27(4). firms do not compete for the low-valuation consumers because the marginal consumer is indifferent between buying the software or not adopting the software at all (see also Peitz 2004). Novos and Waldman (1984) show that lower profits because of copying can reduce the incentive to provide quality. but have sufficiently high valuations such that firms compete to sell to these consumers. Our work differs from the previous literature in significant ways. Besen and Kirby 1989). firms have little ability to appropriate revenues from the consumers who purchase the original product and are the source of the copies. McAfee's virus protection program competes with Symantec's Norton Antivirus. For example. There is also research that examines copying by firms. Hui and Png (2003) show that if we consider the positive network effects of piracy. firms sell to high-valuation consumers who do not copy and low-valuation consumers who potentially can copy. Thus the producer can indirectly appropriate revenues from users who copy. strict copyright enforcement by one firm can serve as a coordination device to reduce price competition. and music. Purohit (1994) shows that the presence of clones can sometimes lead to higher quality products. They consider a scenario in which the low-valuation consumers have sufficiently low reservation prices such that only a fraction of these consumers enter the market. Microsoft Money competes with Intuit's Quicken. while they do not. Their results show that firms can benefit by not protecting their software (thereby enabling piracy) if network effects are strong. which is clearly not true in the case of digital products. etc. unlike most of the prior literature on copying by consumers. then industry estimates of lost profits because of piracy more than doubled the actual losses. 610-626. in such situations. we also consider the presence of network effects and how equilibrium copyright protection is affected by network effects. but our analysis is different than theirs in many significant ways. Our research shows that the presence of competition can significantly alter . in their framework. movies. this research is only tangentially related to the context that we study. We also consider a duopoly framework and find that their results can be significantly altered if the lowvaluation consumers are more price sensitive. Therefore. Our research also extends previous research in that we allow firms to not only set prices but also to determine innovation levels in the presence of copying. Their result shows that piracy can lead to reduced price competition. We show that when the market is saturated and there is limited opportunity for market growth. This is because when firms compete for the lowvaluation consumers. we develop a duopoly model and study how the presence of competition can affect firms' preferences for copyright enforcement. For example.^ In an independent research. Johnson (1985) shows that copying can reduce firms' incentives to provide variety. ©2008 INFORMS Shy and Thisse (1999) extend the monopoly results of Conner and Rumelt (1991) and Takeyama (1994) to a duopoly framework. publishers of books can charge different subscription rates from individuals and libraries. We show that the presence of network effects can critically impact the role of copyright protection in reducing price competition. higher network effects can actually lead to higher levels of copyright protection. (2004) show that the presence of strong network effects can make it optimal for a firm to freely license its product. Even for entertainment software. In their Hotelling model. firms compete to the extent that consumers have limited time and money.

Swinyard et al. Constant by which potential copiers discount reservation price. and therefore might copy the product. results from prior research."* We assume that consumers are aware of the quality of the product. Profit function for firm 1 at the second stage where firms decide quality (given a. The expected network size for firm /. This segment could potentially consist of large organizations and professionals. strong network effects can lead to a more stringent copyright enforcement by some firms. pp. and therefore do not copy the product. In particular. Df(-) D. Base valuation without additional R&D. Potential Copiers R1{e)^v + q.1. 2006). The other segment of consumers view the pirated version to be the same as the original. The reservation price that a segment 1 consumer has for the monopolist's product is given by The consumer surplus that a noncopier consumer at 6 gets from buying the product 1 is given by where pi is the price for product 1. their ability to do so would depend on the company's efforts to protect their product. PI k q. v is the base quality. Index of market competitiveness. consumers may be unaware of the quality and use the pirated version of products such as movies and music to judge We assume that there is another segment of consumers who could copy. Alternatively. and both the firms and society may be better off with higher levels of piracy We also show that in some situations. ' The parameter ß could be affected by the exogenous factors such as government policies on copyright enforcement. However. ß may be smaller or larger than 1.Jain: Digital Piracy: A Competitive Analysis Marketing Science 27(4). It is easy to see that the sales of firm 1 in this segment is given by if Pi < Ü -H i/i .). we show that piracy can benefit firms even without strong network effects. In particular.-e. Therefore. Demand for firm / In the potential copier segment. The network externality parameter."() ni3() 0. The size of the potential copier segment. Note that q^ can be interpreted as the quality of an upgrade over the initial product that the firm had when it entered the market. Price charged by firm /.. Thus the parameter ß should not be interpreted in terms of the number of customers but the size of the sale. In particular. ©2008 INFORMS 613 Table 1 Notation e f{-) F() V S ß a. Noncopiers We assume that noncopiers have heterogenous preferences. 3. The level of piracy permitted by firm /.2 (•) n. depend on the location of firm 1. if they value the legal copies more than the pirated version. and therefore do not find copies to be acceptable substitutes for legal software (see. 3. Model We first consider a monopolist selling to differentiated consumers.). For these consumers. these consumers may end up purchasing the product. * ^ In many cases. The pdf for the distribution of 9. The R&D cost parameter. If the quality is acceptable. support may be less of an issue and they also do not find copying to be as morally objectionable. We assume that while all potential copiers could copy. We assume that the monopolist (firm 1) is at one end of the Hotelling line. for example. Profit function for firm 1 at the first stage where firms decide a. these consumers may consider pirating to be morally unacceptable. Bagchi et al. their ability to copy would depend on the copyright enforcement policies of the firm. and a. and may therefore be beneficial for firms whose expected quality is lower than the actual quality. we assume that the company can affect the level the quality.. . The cdf for the distribution of 6. In such cases. we assume that each consumer only purchases one product. Quality chosen by firm /. 3. Furthermore. Gopal and Sanders 1997. Profit function for firm 1 at the third stage where firms decide prices (given q. This analysis serves as a benchmark for the subsequent duopoly analysis. however.i?i otherwise. • Table 1 summarizes the model notation. We assume that a consumer buys the product as long as CS"{p-¡) > 0. piracy acts as a mechanism to allow consumers to costlessly sample products. in particular. We assume that the size of this segment is given by ß. for example. each consumer buys one unit. These consumers find the quality of the original and the copy to be the same. (1) where 6 is the location of the consumer.^ We assume that there are two segments of consumers. (•) y zf Ï) Model Notation Explanation Consumer location. Shy and Thisse 1999). Demand for firm / in the noncopier segment. and q^ is the quality improvement offered (over the base) by the firm. these consumers are distributed on the Hotelling line according to a continuously differentiable distribution / ( ). The first segment of consumers places a low value on the pirated version of the software. 1990. the availability of manuals and support (which is only available for nonpirated versions) is valuable (see. at 0.^ ^ The result for a monopoly does not.^ Note that in our formulation.2. 610-626.1 (3) I F{v -I. For these consumers.

After choosing the quality level. the firm decides on the copyright enforcement policy.). In Technical Appendix A (available at http://mktsci. The sequential structure is reasonable given that product quality is more difficult to change than the prices. •' In a monopoly. This assumption is also reasonable in the contexts that we are studying. The two-point cost distribution that we use is widely used in the literature in various contexts such as price promotions (see.pubs. Bagchi et al. Shaffer and Zhang 1995). advertising (see. we assume that R[{e) = 8R"{d) = 8{v-\-q^-6). In the second stage. hardware locks.. Jain and Srivastava 2000). Therefore. the assumption does matter. it is possible that Rj(O) > R"{1). the parameter a. Also. However. The sequence of decision is as follows. so that the profit function is concave. for example. Cheng et al. for example. 18 and footnote 18). then there is a probability a. A more general formulation would allow for a distribution of costs.. Therefore we assume that the cost of copyright protection is almost zero. ' The assumption of almost zero eosts is only used for equilibrium selection. then the firm produces a higher quality product. the total sales for product 1 in this segment is given by if (6) ß{l-a. represents the level of piracy in the market. Note that the assumption that these consumers have lower transportation costs (and therefore are more price sensitive) can be justified by the fact that. leading to the possibility that not all consumers in this segment copy. 2006). absent copyright protection costs. This would substantially complicate the analysis without affecting the basic arguments in the paper. where S^ and ^2 are different. in general. Furthermore. the firm decides on the optimal copyright protection level by choosing a^ to maximize n. Prior empirical work on copying suggests that consumers who copy have. in a duopoly. or not to use the product. Zettelmeyer 2000). ifv is sufficiently large. The reservation price for product 1 for a consumer at 6 is given by Ri{d). In particular. a firm would incur some costs to reduce piracy This could lead to firms choosing less than perfect protection. 1997. piracy is inversely related to income (Cheng et al. we are interested in examining whether a firm would prefer lower copy protection even absent such cost considerations. in this paper. 1997). we could have (5) Finally. We make the assumption that 5j = Sj for notational simplicity Note that the assumption R{{e) < R^id) does not imply that all potential copiers have lower reservation prices than the noncopiers. We assume that if firm i chooses a protection level a. for example. and others. for example. firm l's profit function in the third stage is given by'' The firm selects the optimal p* and then selects q* by maximizing li3('7i)-^iiV2l(8) where 0 < 5 < 1.. The monopolist chooses a^ = 0. and social welfare is higher under perfect copyright protection as opposed to no copyright protection. while these consumers are predisposed to copying. Our results would go through even with this alternate assumption. We assume that it costs firm a fixed amount kq^/2 to add quality improvement of q^. However. . Png and Hirshleifer 1987.Thus the assumption only implies that on average noncopiers have higher reservation prices than the potential copiers. ©2008 INFORMS As we discussed before. we assume that marginal costs for the product is zero. and probability (1 — a¡) that the costs will be so high that it will not be profitable to copy the product. where fc> 0 is sufficiently large. In the first stage. copy. the firm decides the quality level q^.^ The main insights of the paper continue to hold in the case where copyright protection costs are significant (see discussion on p. In other words. PROPOSITION 1. price matching refunds (see.informs.)F otherwise. in general. 610-626. and forced activation (Gopal and Sanders 1997). the firm sets the price in the third stage after which consumers decide whether to buy. a firm can affect their ability to do so. In general.(a. To incorporate this in our model. we develop a model in which consumers have a distribution of costs and show that the basic nature of our results remain true even in this more complicated scenario. In an alternative formulation. Using this formulation. For example.614 of piracy by using preventive measures such as the use of nonstandard disks. This assumption captures the empirical fact that some consumers copy one software purchase another software. softwarebased protection. a lower willingness to pay and are more price sensitive (see. pp. note that we are assuming that these consumers discount V and the transportation costs 6 by the same factor 8. it does not matter whether the firm sequentially decides on quality and price. a firm would prefer lower copyright protection if the profits are the same at different levels of «j.org). Without any loss in generality.that a consumer in this segment will have zero copying costs for copying a particular product. (4) Jain: Digital Piraey: A Competitive Analysis Marketing Science 27(4).

As in the monopoly case.a^){l . The third stage profit function for firm 1 (denoted by 1113) is given by i3 = Pi[f (oí) + ß{l -a. Thus j8(l . '" If /(•) is uniform with range (0. this assumption is particularly useful because it replicates conditions under which a monopolist firm would never prefer piracy and would produce lower quality products under piracy (see Proposition 1). 610-626. We start the analysis under the assumption that these consumers prefer to copy product / rather than purchase product . 4. social welfare is reduced as a. the firm would produce a quality level of 0..)il(14) " In any symmetric equilibrium with q. Analogously. each firm decides on the level of copyright protection a¡. Shaffer and 'For example. i. consumers make their purchase or copying decisions. the reservation price for firm 2's product from the potential copier segment is R¡{d)^SR'¡{e).6 and charge a price 0. Vx e (0. Thus.098 and u> 0. This benchmark proposition shows that under the setup that we have chosen.k = 2.1). Duopoly We now assume that there is a second firm in the market at the other end of the Hotelling line. the problem of underprovision of quality is further exacerbated. and will therefore copy a product. These can be shown to be -'I — (12) 25 (13) We assume that 5 is sufficiently high that the firm would prefer to cater to the potential copier segment.0:2) consumers in the copying segment cannot copy any product while ßa^ «2 can make copies of either product. Thus the reservation price for firm 2's product from the noncopier segment is given by Zhang 1995).1) and ß-l. for example.0. This proposition thus lends some credence to the advocates of strong copyright protection. in our context.(l . but not copy product /.78 with a = 1. Focusing on interior solutions.1) and ß = J." With this setup. Also. = q^. In cases where we use a specific distribution for /(. We also impose some additional restriction on the value distribution function / ( ). Therefore. if v is large enough so that all consumers are served. However.e. a2-ln the second stage. In the third stage.Jain: Digital Piracy: A Competitive Analysis Marketing Science 27(4).5+ 1/5. this would hold if Ö ~ Li(0. However. i = 1. Consequently.v = 2. piracy is invariably bad for the firm and society.96 without piracy and set quality level 0. pp. The other assumptions are very similar to those of the monopoly case. then a sufficient condition for these to hold is that S > (-5 + 4y/2)/7 ^ 0. then this proposition shows that under no piracy. in markets where network externalities are not strong. we can determine the demand for firms 1 and 2 for given prices. We also assume that v is sufficiently high and the competition is sufficiently intense that the market is fully covered in a duopoly.1) and 5 < 1. then the firm would serve all the consumers.'" This assumption is standard in Hotelling models because it considerably simplifies the analysis (see. In particular.) is uniform with range (0. the firm would prefer to impose perfect copyright protection and curb all piracy.8 = 0. j3a. we assume that /'(•) exists Vx e (0. In other words. we show later that the qualitative nature of our results is valid even if the total market size is not fixed and some consumers do not buy. when v is sufficiently large. We will look for the symmetric subgame-perfect equilibrium of the game.aj) consumers can copy product /. the firm produces a lower quality product under higher levels of piracy.' Consequently. a monopolist produces an inefficient level of quality (from a social welfare perspective). This enables us to highlight how the presence of competition might change the results. 6^ is analogously defined for the potential copier segment.1) and /(0.2. the firms decide on quality improvement levels q^ and q2 at a fixed cost kqf/2. As the level of piracy increases. note that the full market coverage assumption is required only for the second part of the proposition. we assume that / ( ) is a symmetric differentiable function with support (0. we explicitly check this condition—see the proofs of Propositions 5-8 in the appendix. then an increase in Oj leads to the firm investing less in R&D.Finally. and social welfare is higher. consider the case when /(.. in the first stage. Furthermore. (10) where 6" is the consumer in the noncopier segment who is indifferent between 1 and 2. at 1.). Furthermore. . if the firm can control piracy. the firms decide on prices given quality levels q-^ and ^2. as we show in the appendix.5). increases.6.3 and price 0. We now extend the analysis by considering a situation in which there are two firms in the market. the firm produces a higher quality product. where 0 < 5 < l . makes higher profits. the firms decide on the piracy levels a^.5-x) = f{0.5+x). and therefore first consider the third stage of the game. Also. the demand for firm 1 is given by where q2 is the quality improvement offered by firm 2. As before. ©2008 INFORMS 615 Proofs of all propositions are in the appendix. In this case.

In other words. as the distribution shifts more toward the center. ß {a = 1)flsopposed to the case whenfirmsenforce copyright is large. This is because the overall price increase leads to higher profits from noncopiers and higher profits from the consumers (who actually purchase) in the potential copier segment. there are many situations ^ 1 3 ^ m. the prices are given by (see the appendix for details) (16) Note that the optimal price is lower as fil/2) increases. The proposition shows that under the conditions specified in (17). Intuitively. in general.. Furthermore. (15) Using symmetry. we could encounter a protection even when there are no network effects. we also need to consider the strategic effect. ©2008 INFORMS 'î)]-0. however. In other words. the choose the level of copyright protection in our framenumber of price-sensitive consumers who are willing work. It is also important to note that while the monopoly effect is centered on lost sales from the potential copier segment only. both firms have chosen the same levels of copyright Note. The proposition below examines the impact of copying on prices and profits. the second term in (19). i. This term does not take into account '^ Cabrai and Villas-Boas (2005) refer to this as the Bertrand Supertrap. i. 610-626. pp. firms are better off with no copyright protection as opposed to complete protection.e. we will need to explore whether no copyright to purchase the product also decreases. i.. prisoner dilemma and firms might end up choosing To understand the intuition for the proposition. On the other ^ (18) hand.3 dp* (19) in which such an analysis is appropriate. We will explore this in sensitive segment of the market. we know that the monopoly effect must be negative. This is also intuitive since as 8 decreases.e. the strategic effect impacts profits from both the segments. dU^^/dp2 > 0. i. Note that the condition in (18) Firms' profits are higher under no copyright protection implies that if there is a large noncopier segment. the strategic effect dominates the PROPOSITION 2. the strategic effect is positive and increases profits. the potential copier segment becomes more price sensitive. To understand the complete picture. both firms make higher profits as also shows that if 8 is sufficiently small. Thus. copyProposition 4. we could have situations in which both The proposition examines the situation in which firms are better off with loose copyright protection. the price sensitivity of the copier segment and the level of copying. a company such as Symantec sells its software The first term represents the direct effect of copying on firm profits. which is captured in the second term in (19). then the monopoly effect would not be offset by the strategic effect and the firms are better protection if off with complete copyright protection. depend on 8. Nevertheless. a. ing reduces price competition between the two firms. ^^ The proposition increases. the copier a increases if segment is sufficiently price sensitive.616 where we assume that the marginal costs for both firms are zero. i. We consider the symmetric case when a^ = a2 — a and q-¡ = (]2 — ^. own direct effects. Note that from Proposition 1..e. because although sible for both firms to be better off with no copyright a = 1 is better for both firms. See their paper for other examples of this phenomenon. the effect of a on competition and only considers . first some copyright protection. first note that cations provide only an incomplete picture in that using the envelope theorem. Since. It is also important to note that these profit impliTo understand the profit implications. this term would determine the impact of copying on firm profits. i. dpl/da > 0. firms protection is indeed an equilibrium strategy under are less motivated to reduce prices to attract the pricethe condition specified in (18).^ ^ relevant first-order condition for firm 1 is given by 25 Jain: Digital Piracy: A Competitive Analysis Marketing Science 27(4). note that as 8 decreases. As indicated in the first part of the proposition... if the copier segments are sufficiently price sensitive. in general. The results show that it is poswill choose no copyright protection. For examda da dp^ da ple.e.e. price competition is increased and prices are reduced. We label this term as the monopoly effect. copying also reduces price competition.. and the firm's profits are higher ai — a and q^ = q2 = q. In the symmetric case. if there were no competitors. The total effect would. then firms' prices are higher as a under increased levels of copying. where a^ = monopoly effect. prices decrease. we have they do not consider the impact of piracy on innovation levels. Of course. Also. that this does not imply that firms protection and quality.e. because firms note that as copyright protection level decreases. This is because copying reduces sales.

it is not clear that this will have a substantial impact on the R&D levels. even when there are no network effects and the full market is served without copying.-J. However.e. perfect copyright protection is likely to be prohibitively costly. once we consider R&D. firm profits and social welfare are higher under no copyright protection than under complete copyright protection. however. then the negative effect of lost sales is greater than the positive effect of reduced price competition. in general. the second-stage profits of the firm are higher when we ignore the R&D costs. Thus. firms can use weak copyright enforcement as a coordination device to reduce price competition and make higher profits. charge higher prices. recall that there are three forces that determine the final impact of The proposition is the polar opposite of the result in the benchmark Proposition 1 with a monopolist firm. pp.). but are unable to commit to not serving (because of competitive reasons). even in this situation. for sufficiently large k. i. We have the following result: PROPOSITION 4. Furthermore. an increase in a decreases the total number of buyers. an increase in a also changes the level of market competifion. Indeed. Nevertheless. consider the case where the cost of setting a^ for firm 1 is given by —^ • ln(a. Recall. and make higher profits when (17) holds. Suppose e ~ U{0. = 0.27. However. in equilibrium. piracy reduces price competitior\.34. However. it is possible for the firms to be better off with copying. First. this is not the case.Jain: Digital Piracy: A Competitive Analysis Marketing Science 27(4). However. If firms choose a^ = «2 = a. it is possible that firms' profits increase with a even if (17) does not hold. First. which is defined as .S. i/(18) holds. The overall effect is such that firms produce a higher quality product as a increases iff (17) holds. We first have the following result: 3. note that a affects the incentives to innovate in two different ways. that under (17). while at 5 = 1. then as a increases. Note that this term is oo for aj = 0 and 0 for a^=l. firms set a. we analyze the first stage of the game in which both firms independently choose the level of copyright protection for their software. a* = a2 = 1. the innovation decisions are based on the global market and may not be infiuenced by piracy levels in small markets. To see this. Now. If k is sufficiently large. In other words. 610-626. Although it is difficult to analytically find the equilibrium a. and we look for the symmetric Nash equilibrium for the game. This. both firms are better off setting a — 1 for 8 < 0.9. piracy leads to lost sales. If (17) does not hold. if (17) does not hold. firms spend less on R&D and save developmental costs. In other words. This points to the fact that some of our intuition from the monopoly situation may not hold true when we consider competition.1. firms choose no copyright protection. while Symantec could alter its prices in response to the 92% piracy rates in Vietnam. in = this case. It is also useful to examine the case when (17) does not hold. For example. The proposition also shows that in situations where firms are better off when they do not serve a price-sensitive segment. In this situation. The proposition establishes that under conditions specified in (18). then the firms would set . which tends to reduce profits. if k is sufficiently large and (18) holds. In practice. then firms are better off with no copyright protection. we consider the second stage of the game in which the firms independently decide on quality levels.. This result also implies that there are conditions in which copyright protection may not be desirable even when there are no network effects and it is costless for firms to impose copyright protection. firms invest more in R&D and incur higher R&D costs. ^ = 0. L Firm 2 analogously determines q2{qi). The problem for firm 1 is therefore to choose qi{q2). ©2008 INFORMS 617 in the global market and can charge different prices in different local markets. they set a. For example.1). In this case. Of course. and ß =: 1. firms produce a higher quality product.• (20) piracy on profits.81. Second. piracy rates of 21% may indeed affect Symantec's R&D decisions. This effect tends to increase firms' incentives to invest in R&D. once we consider R&D. = 0. Thus the overall impact is ambiguous. then in the symmetric equilibrium firms produce a higher quality product and charge a higher price as a increases if and only if (17) holds. To understand this result. then the negative impact of a through increased R&D is outweighed by the positive effect of a on reduction of price competition. the U. However. This suggests that our results could be stronger if we allow for costly copyright protection. then both firms choose no copyright protection. when (17) holds. and finally piracy affects R&D expenditures. in this case. Even for Ô = 0. we can do so numerically. It is also important to note that the proposition is derived under the assumption that copyright enforcement is costless. Furthermore. will tend to decrease the firms' incentives to spend in R&D. which is substantially weaker than the condition S < 1/3 implied by (18). Now. the presence of piracy can solve the commitment problem. PROPOSITION This result shows that firms may spend more on innovation and make higher profits under piracy. if the firms were to ignore the strategic effect (and behave as if 5 = 1). the firm always produces a lower quality product as a increases.

The consumer in the noncopier segment who is indifferent between product 1 and product 2 is indexed by 6" and is given by Note that the 6s in the above equation are a function of z\ and z^. As before. it is useful to examine whether the same intuition holds in our framework. If both firms were to allow copying. where 0 < À < 1.e. '•* Note that we are modeling direct network effects as is common in this literature. we assume that noncopiers and potential copiers value network effects equally because there is no evidence to suggest that the valuations would differ. To accomplish this. This condition ensures that the network effects are not so large that small changes in network sizes can lead to large shifts in consumer demand. Thus although Proposition 4 shows that no copyright protection is beneficial to firms when (18) is satisfied. € {0. This assumption restricts the strategy space in the first stage of the game and makes the analysis tractable. 5. z{ is the expected network size for product 1 and y represents the intensity of network effects.618 more stringent copyright protection. we discuss the implications of relaxing some of the assumptions in our model. Note that the expected network size consists not only of sales of the product but also copies of the product. However.1|. Finally. Thus. We also impose the rational expectations condition z? = Zi z¡ = Z2. we make some further simplifying assumptions. closed-form solutions are difficult to obtain. Let the surplus that a consumer at location 6 in the noncopier segment gets from buying product 1 be given by CS"{p^. 6) = R"{6) + yz\—p^.2|) d The consumers who copy also care about the network benefits and would take the expected network size into account while deciding which product to copy. First. if consumers' cost of copying is not higher than the combined production and distribution costs. Also.g. we assume that / ( ) is urüform with range (0. Thus. To solve this. If we relax the assumption. However.I2) +P2-P1 + y(2i . so that the network effects are specific to each product. Thus the key insights in this paper do not require the full force of the assumption that (18) holds. we assume that firms coordinate to the Pareto-dominant equilibrium. we also assume that y < 5/(1 + 8). However.1) and set ß = 1. Network Externalities The base model assumes that there are no network effects. To ensure that a unique and stable rational expectations equilibrium exists. which is the number of consumers using a product. the strategic effect impacts the equilibrium copyright protection at all levels of 5. as we discuss later. pp. if we assume that the potential copiers value network benefits at AyjZÎ. '^ In our formulation. Church and Gandal 1992) and cross-market network effects (e." Jain: Digital Piracy: A Competitive Analysis Marketing Science 27(4). Other types of network effects include indirect network effects (e. our result holds. .g. we focus on equilibrium in which consumers expect that copiers would copy rather than buy (which we will show holds in equilibrium). which is given by (24) 5. than the social welfare results will be weaker. Chen and Xie 2007). If the cost of copying is greater than the distribution and production costs (which are also assumed to be zero in our framework). we assume that the market is fully covered. numerical simulations reveal that the basic nature of results is unchanged with this alternate and more complicated formulation. and yZj represents the network effects. the consumer in the potential copier segment who is indifferent between the two firms is indexed by 6{ and is given by'^ d' = ^ + ^(?i .. consumers who copy the product exert a positive externality on the utility of other consumers.. Zj is given by (25) where R"{d) is as defined before p^ is the price. as y increases. 610-626. then the network size for firm I's product.^* In particular. We assume that the products are incompatible. This helps us better assess the generality of our results and better understand the contexts in which our results apply. network effects become more important. we extend our basic formulation to allow for network externalities.. In particular. i.1. we assume that consumers also derive a utility from the size of the network. (21) 28 To proceed. in case of multiple equilibria. Model Extensions In this section. ©2008 INFORMS Analogously. The consumer who copies and is indifferent between copying 1 and 2 is indexed by 62. Because prior research has shown that piracy can only be beneficial when network effects are strong. then the results are similar. leading to multiple equilibria in which only one of the firm survives. we make a simplifying assumption that firms choose a. The consumer surplus for product 2 is analogously defined. (26) (27) '^ The results on social welfare do depend on the assumption that copying is costless for consumers.

first consider the case when 5 = 1 and there are no network effects.463. Note that for large values of 5. charges a higher price. Of course. which has shown that firms allow more copying as network effects become stronger. This makes piracy attractive. can make it attractive for firms to allow some copying.^^ Nevertheless.296. Although the result shows that there is an asymmetric equilibrium. Recall that absent network effects. enables firm 2 to charge higher prices. weaker copyright protection serves as a coordination mechanism to reduce price competition. the result shows that while the firms are a priori symmetric. Note that the firm that imposes copyright protection is at a disadvantage relative to firm 1. this leads to more intense price competition.25. we see that an increase in y can lead to higher levels of copyright enforcement. a firm may find that piracy also leads to increased network size. they end up making different profits because of the strategic copyright protection choices. = 0. The firm that allows copying (say. the firms switch to an asymmetric solution.5 and make profits of 0. and we have a prisoner's dilemma in which both firms end up allowing piracy. the utility In case. in equilibrium. In this case. 610-626. stronger copyright protection enables firms to coordinate to reduce price competition. The more general case is more complicated and difficult to analyze using closed-form expressions. note that an increase in y also makes piracy a less effective tool in reducing price competition. This allows both the firms to increase their prices.Iain: Digital Piracy: A Competitive Analysis Marketing Science 27(4). in turn. while firm 2's profits rise to 0. y > 72/flwrflarge k. firm 1) sets p* = 0. 16(1 + 5) (29) The result thus shows that as the level of network effect increases. in equilibrium. at y = 0. as is intuitive. although not allowing piracy may be better. £^2) and Piicji. This result is in contrast to the standard argument in the literature. This enables firm 1 to charge higher prices (because it has a higher installed base) and this. where 75 +15 72 = 32 (28) Furthermore. For (-5 + 4V2)/7 < 8 < 78/9 «s 0.94. For example. for larger values of y. However. In this case. the range of y for which the asymmetric effect holds is larger for smaller values of 5. pp. In this situation. at least one firm would benefit by switching to a. weaker copyright protection. in this case. To understand this result. let us consider the case when 5 is low. in this context. it is possible that some firms may choose to use stronger copyright protection. Figure 1 shows how firms' equilibrium copyright strategy changes as a function of 5 and y. The firms solve for the equilibrium qualities given p^{qi. and if one firm allows copying. The profits of firm 1 go up to 0. For lower values of y. one firm allows copying and the other firm does not. and therefore reduce price competition. enhancing effects are more important and firms prefer to have weaker copyright protection. if y & {y^. . and as we have established before. We have the following result: PROPOSITION 5. Also. In this case. since firms are symmetric and the market is of fixed size. The proposition thus shows an interesting role of copyright enforcement in the presence of network effects. <J2). we first solve the last stage of the game in which firms make pricing decisions given the qualities q-i and q2. where network effects invariably discourage firms from using strong copyright protection. as we discussed before. piracy reduces price competition. consider the case when 5 = 1/2 and k is large. where 3(2 + 55-Vl2 + 45 + . As y increases. piracy is only harmful in that it takes away sales from the potential copier segment. while the other firm sets pl — 0. however. because to reduce competition. as 5 decreases. then both firms allow copying. and the strategic effect of copying becomes more important. Therefore. Furthermore. However. it leaves the copier segment to firm 1. In this case. the second effect becomes more important. the firm that allows copying produces a higher quality product. and makes higher profits}^ However. large network effects lead to weaker copyright enforcement. However. it becomes attractive for the other firm to restrict copying.83.Our interest in the analysis is to examine how network effects impact the results of the previous section. This is consistent with prior research. However. then they would each charge a price of p* = 0. Now. enables firms to commit to charge higher prices. 72 > ^/(l + ^)f then the asymmetric solution will not exist. we conducted numerical simulations using a discrete version of the game in which each firm "An additional complexity is that we need to check for each parameter values whether consumers who can copy will copy or buy under the rational expectations framework. we need to consider the possibility that this result may be an artifact of our assumption that firms choose only two levels of a. y^^. In the presence of strong network effects. ©2008 INFORMS 619 As before. the role of copyright protection changes. and therefore increased ability to charge a higher price from the noncopier segment.25. If firms were to set a = 1. weaker copyright protection leads to even more intense price competition.67. This is consistent with prior research.

20 to lower levels. the market expansion effect because of increased network size. Further7 more. Again.8-.10 y^ ^^^ Copying y^ .7 •. as y increases.4-.620 Figure 1 Equilibrium Copyright Enforcement with Network Effects and Fixed tnarkel Size Jain: Digital Piracy: A Competitive Analysis Marketing Science 27(4). Also. then in equilibrium. 28 -^ (34) Note that if 17 = 1 and y = 0. If 1 = 1 and y > 0. the firms are effectively monopolists. 8* reduces to 1/3 and the firm never benefits by allowing piracy. is analogously defined as also considered the case when there is a cost of copyright protection given by —0.6-.201 combinations and numerically searched for the pure strategy Nash equilibrium of this large discrete game.05 0.2. as 1 ->• 1. 5. the results are stronger. both firms set a* = 1. as long as firms are sufficiently competitive. The demand from the potential copier segments. we consider the case when the market is not fully covered and there are network effects. so that the full market is covered and the total market volume does not depend on firms' prices. where y>8/(H-8) 0. when r] = 0. This is consistent with our result since 1 = 0 implies that the firm is a monopolist. To model this.3-^ 0. our main results hold. It is also important to note that our analysis assumed that the market demand is fixed.2. Our analysis does not incorporate a beneficial aspect of piracy in the presence of network effects. On the other hand.. we find that at y = 0. we assumed that the market is sufficiently competitive. 0. firm 2) sets Oj = 0 at y = 0.. so that at 7 = 0. while the other firm continues to set a\ — 1. This is intu7 itive because the full market coverage assumes strong competitive interaction and the results depend on the ability of piracy to reduce the level of competition. In other words. (31) 0.15 Asymmetric ^^^ Note that the analysis reduces to the base case when 17 = 1 (see Equations (12) and (13)).—1—. The demand from the copier segJ ment is analogously defined as 8+ 18 VP2 0.^* Thus the main insight of the analysis continues to hold even when we allow firms to choose multiple levels of a.01) (where the constant 0. We can incorporate this by postulating a linear demand funcfion of the form (30) Thus the proposition shows that the main result still holds even in situations where the market is not fully covered as long as T.» - where D. we analyze whether our results are still valid when we consider scenarios where the total market is not fixed and the presence of network effects can increase the size of the market.—1—A]—1—1—1—r-7—1—r—F—T— 8* = (32) was allowed to choose among 101 levels of a from 0 to 1 in increments of 0.-^^'^ 0. For large k. we have the base case. one of the firms (say. Note that the demand reduces to the constant demand formulation when 17 = 1. /3 = 1.01. However. However. the total demand is a decreasing function of p^ and p2.01 is used under the log term to avoid infinities in the numerical simulation).e. It is possible that the market demand could be influenced by the firms' prices. both firms allow copying. 610-626. we have the case analyzed 7 . we find that one firm confinues to set aj = 1. In our formulation with constant demand.5-i 0.30 0. For S = 1/2. 1/3). aj = 0. pp.32. network effects tend to intensify price competition. In this situation. Also. Market Is Not Fully Covered In our base model. i. note that for 77 = 0. In the next section.1 . We calculated the equilibrium profits for each of these 10. and k large. We have the following result: PROPOSITION 6. we assume that the demand from the noncopier segment is given by (33) where 0 < 17 < 1.The remainder of the analysis is as before. when they are unable to copy (which is a fraction (1 — ai)(l — 02)).19. as is well established in the literature. 0.01 • \n{a + 0. Now. 0. No copying / ^ 0. network effects tended to increase competition.25 0. if 8 € {8*. ©2008 INFORMS u.19. is not too small. We incorporate this in our formulation to examine whether our results would still obtain in this modified framework.2-. while the other firm switches from «2 = 1 st y = 0.() is the demand from the noncopier segment and 0 < T < 1.20 0. network effects can increase the size of the market.

6ments using different product lines. note the results show that the role of copyright protection in figure plots the equilibria for 5 = 1/2. Our results sugtion suggest that as markets become more mature and gest that in maturing markets for products with large there is limited potential for growth. Thus. this generalization allows the beneficial market expansion effect of network effects. since in the extreme case as those for entertainment products. lower values of 5. pp. a duopoly. Thus. For \ \ instance. ©2008 INFORMS 621 This paper examines how the presence of competition changes the effect of piracy on firms' prices. as 7 increases. Taken together. We have the following proposition: 6. network externalities always makes copyright mature in developing countries with large income disenforcement less attractive. one firm allows copying and higher levels of piracy.8the implications of relaxing these assumptions. and R&D levels. the effects identified in the paper one firm may choose to enforce copyright protection are stronger. profits. Future research \ can examine how our results change if the firms could 0. For lower valreducing price competition is substantially different ues of 5.27. piracy will lead to increased market size. the results change. we are more likely to 7 find that an increase in network effects can make protection in the presence of network effects. This. we first developed a model in which there were no network effects and the market is completely covered. 1 = 0. In partion holds increases. as the level of parities (such as China and India). Our results suggest that copying by more price-sensitive consumers can The proposition thus shows that even in situations enable the firms to credibly charge higher prices.U- in Froposifion 5.30 \ 0. 610-626. In developing the theoretical model. this effect is tempered by beneficial not to impose strict copyright protection. such as Microsoft. but we can easily deter. "We have also extended our model to examine how firm asymmetries can impact their decision to enforce copyright protection. of lost sales. Also. However. Let 8 = 1/2. we wished to examine how these effects impact firms' incentives to enforce copyright laws.15 y 0. we do not allow for the possibility that Copying 0. a firm will always be hurt by 7 if y > 0. Details are available in Technical Appendix B at http://mktsci.Jain: Digital Piracy: A Competitive Analysis Marketing Science 27(4). This is intuitive. we made sev\ s^ Asymmetric 0. However.25 = 1/2.7\ firms could price discriminate between the two seg\ 0. will find it beneficial to more strictly enforce copyright enforcement. firms may find it competifion (17) increases. our results are valid.informs.can sometimes outweigh the negative impact because mine the critical y for other parameter combinations.5^ 1 . . Furtherence of network effects can increase the total market more. in contrast to previous research. In a monopoly setPROPOSITION 7.4-. in equilibrium. such 1 increases. this positive effect of piracy on firms' profits size. The proposition is derived for specific values of 5 and 17. our results show that the presence of strong network effects is neither a necesFigure 2 Equilibrium Copyright Enforcement When iVIarket Is Not Fully Covered sary nor sufficient condition for firms to choose weak copyright protection. to impose strict copyright protection. This in which the market is not fully covered and the presleads to a reduction in price competition. Our copyright enforcement more attractive.05 Note. We also extend our model to consider copyright Consequently.^' the other effects that we have previously discussed.20 0. We establish conditions under which 7 in which an increase in y can lead to a firm choosfirms may choose weak copyright protection even in ing more stringent copyright protection. is lower as mature markets with no network externalities. then the effects network effects.10 0. We see that the range social welfare. We also show that piracy can someFigure 2 plots the equilibrium outcomes for different times lead to an increase in innovation and improve values of y and 1 at 5 = 1/2. Also.9\ eral assumptions and future research can examine 0.org. then for large k ting in such situations. Furthermore. Toward this goal. in turn. the region in which the asymmetric soluin the presence of strong network externalities. it may be beneficial for some firms identified in this paper will become stronger.8. Conclusion X 0. These results 7 when 1 = 0 and the firms are effectively monopo. n Q " No copying 0.suggest that as markets for entertainment products 7 lists. we find that when network effects are strong. the results of this secso as to reduce price competition. We find that large firms. can benefit both firms because the increased network size increases the price that consumers who purchase are willing to pay. once we consider the other firm does not. This is also intuitive because at ticular. S 0. because copying allows consumers to get a product at zero effective price.pubs.

If q. The sequence of decisions is as follows. First. note that if v is sufficiently large. then we would not expect such effects. when firms offer multiple products. the presence of the second product provides some ability to price discriminate. 18). Our model also does not allow for the possibility that over time piracy can affect the incentives for noncopiers. It is easily shown that da-¡ (39) da. piracy would lead to higher prices in the second period. However. For example. Appendix PROOF OF PROPOSITION 1. a firm that is maximizing long-term profits will need to take this effect into account. then the symmetric solution would imply that ÖJ = 0j" = 1/2 and f (1/2) = 1/2 for a symmetric distribution..kqf/2.a){l . their prices are still constrained because if they charge too low of a price from the potential copier segment. such consumers may also consider pirating. 610-626.Olln(ai). their argument is based on the diffusion model and the presence of network effects. both firms would set a¡ = 1. then the firm's profit function is given by Since the market is fully covered. in a two-period model in which piracy is allowed in only the second period. It can be argued that some consumers need to buy before the pirates can copy the product.33 and set a. and 0 < S < 1. while deciding on its copyright protection policy.. assume firms do not conduct R&D and ß = i and S = 1/2. and therefore firms do impose more copyright protection than before. then the firms are better off using no copyright protection and only have one product. otherwise. However. they would each make 0.e.a2)F(eí)] = 0. We will look for symmetric equilibrium T* and a' when there are two products. the profits in this case are 0. Therefore we have This implies that (43) Differentiating. Our results would suggest that absent network effects. Thus. ©2008 INFORMS how such dynamic aspects affect firms' incentives to enforce copyright protection.kq\ll.)(l . firms independently decide on a. which is the catered to the potential copier segment. (1995). if the cost of new product introduction K > 0. multiple product introduction coupled with strict copyright enforcement is costly. The relevant first-order condi- tion is [ 2' ' 25 ' J (41) .a. i. firms will set T. In the two-product case. Let us consider the case where each firm introduces another product of quality v — T. " On the other hand. Acknowledgments The author thanks Wilfred Amaldoss.86)f {9) dO .5.a)] . = q2. Ambar Rao. and then decide on prices for both their products. In contrast.. ^' This is consistent with the empirical work of Givon et al. they will not protect their software (see p. Furthermore. and the editor for helpful suggestions. we have dp. = 0. If these consumers do not pirate because they find piracy to be morally objectionable. absent a second product.93. they decide on T. whereas if firms were to set a. which come with a legal copy. (35) which is weakly negative. one could examine how piracy might affect the adoption decisions of consumers. da 2ß{l . Therefore. the area editor. to make things simple. we have . Using the Envelope theorem. Clearly. does give the firms the ability to price discriminate. pp. (37) Therefore <?. four anonymous reviewers.5007 — K.622 produce a cheaper version of its product for segment 2. if the noncopiers are not pirating because they value the support services.^^ Future research can further examine how piracy would affect diffusion of digital products even in cases where network effects are not significant. In other words. Social welfare is defined by + j ß{8{v + q\) .8)8 (44) . it is plausible that as piracy becomes more prevalent.^^ Future research can examine "^"In general. If we assume that there is a cost of copyright protection —O. Also. Thus the firm will choose a = 0.007. while introducing multiple products. Jain: Digital Piracy: A Competitive Analysis Marketing Science 27(4). This would lead to rational consumers buying early.* = • (38) Thus ql is decreasing in a. there are multiple issues that one needs to consider. we have dn'. and therefore piracy can accelerate product diffusion in the first period even when there are no network effects. in a symmetric equilibrium. = 1. p* = 8{v + q^) — 8. (40) where the inequality follows since q* is decreasing in a.. Next. D PROOF OF PROPOSITION 2. weaker enforcement of copyright protection to reduce price competition also simultaneously reduces costs. To see the second part. = 0.^° Our paper examined a single-period model in which both the pirates and the legal purchasers of the software enter the market at the same time. then there is the risk of noncopiers buying the cheaper version. The usual disclaimer applies. First.

If /'(1/2) > 0.. where we allow for the possibility that q^ ^ q2.ß). we need to consider the third-stage decision by allowing for the possibility that the second-stage qualities are different.Jain: Digital Piracy: A Competitive Analysis Marketing Science 27(4). = »2 = a. define ^ We have + t^-ï^-'^ii=O- (50) (51) A. it follows that Thus the relevant first-order condition for firm 1 in the second stage is given by 25 3Ä] = ^ - (66) Thus the first-order condition (50) becomes (67) (53) We therefore need to evaluate function theorem -^. 1 (47) Since we are focusing on symmetric solution. Using the implicit which reduces to ^' (54) 23 3k 3k[8-^-ß{l-a)^' (68) Differentiating. dpidp2 /(1/2) 2Ô (58) Similarly. a = 0 and a = 1. which completes the proof. assume it is not true. /(1/2 4. An analogous argument works for /'(1/2) < 0 case. the firms decide on qualities. Also. Thus under symmetry.a)^] .«isn • (69) . in a symmetric solution (64) Also. using (54) and (64). pp.af)[28 . ^13 dpi /(1/2) 28 /(1/2) Denote 28 (60) (61) (62) • + -~^-if^- (49) (63) The relevant first-order condition in the second stage is therefore Thus.3 Analogously. D PROOF OF PROPOSITION 3. the firms' maximum profits as a function of a can be determined by comparing the two extreme points. firms' profits are higher under a = 1 if ô < 1/(2 -I. /'(1/2) =0. Thus /'(1/2) = 0. we have dpi Also. We have 8(1 Similarly. Using the Envelope theorem. we have 3'n. then by continuity for sufficiently small e. we have /(1/2) (59) " " ^ " ~ ^ ' ~ 2/(1/2)^'"^ Thus. i..e.1 -f. (57) d'n. Given that a. after simplification. denote f{9{) = /j and /(ö") = /2. In this case. We need to determine the impact of q^ and q2 on the third-stage profit function. ©2008 INFORMS 623 where 23 Using the Envelope theorem for the third-stage profit function. /j = /2 = /(1/2).e) > /(1/2 — e) violating symmetry. Because the cost of enforcement is zero. = (65) Also. after simplification we get da (45) dpi We have dpi (55) which is greater than zero only if (46) Note that this implies that the firm profits are decreasing in a for small values and increasing after the critical value defined in (46) is reached./3(1 . To see this. For notational simplicity. in the second stage of the game. 2Ô (52) Thus. we have 3 7 _ ~2ß8(l-a){l <* da -h ^(1 . 610-626.

the result follows. In this case. then we know that 11. does not impact the final consumer choices. Now. which completes the proof. ß{l — al)a2 can copy only 2 and ßa*{l — a^) can copy only 1. . Then the profit functions are The network size for firm 1 is symmetric in the second stage. we get First.e. This.2 is maximized at Í = 0..18ky8 + 9k8 - • (87) . we get under the rational expectations equilibrium -2yg + 28q.25^^ . al). Assume that the equilibrium is (a*. is given by Analogous condition exists for firm 2. as k increases. Differentiating 2kq^^ =0. q^.oc*.2o)/ß. then the first term is higher for a = 1 than for any other a. Since by assumption copiers prefer to copy.2y By solving the game as usual. when (1 — a*){l — aj) = 0. when there are two products.y\^ . (81) The result then follows since the welfare function is concave and by the ordering q{i = 1) < qU = 0)<q^^. + 28.al. The socially optimal quality is given by (77) To calculate the equilibrium profits. (86) SW = 2\ f L-'o (v + q-e)f{9)de /•1 +ß . and only one firm allows copying. ß{l — a'¡){l — Oj) consumers can copy neither. (76) . the profits for both firms will be The rational expectations condition is given by m(73) I8k[ [8 -I (83) (74) If (46) holds. we investigate the impact on social welfare. we have + 2y8 .y^8^) • (18k(8 Both Firms Do Not Allow Copying 18yk8 . i. implies that there must be a symmetric equilibrium in prices. but then the other firm weakly prefers to set a = 1. Because prices are transfers from one group (consumers) to another (firms). a*y) represent the sibilities: both firms allow copying. Since. 610-626. D PROOF OF PROPOSITION 4. Then from (71). in (82) turn.+22 = 2. •'0 (75) Using the approach as before. D PROOF OF PROPOSITION 5. we can plug the equilibrium qualities back into the profit equation. (84) By solving. pp. + gp.9ky . a proportion ßa^a2 consumers can copy both products.q2. these have no impact on social welfare. q* increases as a increases iff a >1—. Therefore. First. which we can Both Firms Allow Copying denote by the parameter (. We have 8[l + ß{l-a)^] We know that (78) Therefore dq' (71) We know from previous proposition that if (46) holds.2 ^ .8p.624 Thus. to find the maximum value of q* over (. Social welfare is only affected by the impact of copying on innovation levels. i. (70) Jain: Digital Piracy: A Competitive Analysis Marketing Science 27(4). Since I varies from 0 to 1. The social welfare function. the second term becomes arbitrarily small. We have (80) Also. at least one firm must choose a* = 1. note that ing. we establish the socially optimal qualities. ©2008 INFORMS Note that the first-order condition implies that the social welfare function is concave in q and is increasing at q = 0. Also. the profits only depend on (1 — al)(l — a^). we only need to look at the extreme points. We will analyze the three posLet Yl-^2{qi.e. Note that piracy affects prices but in the symmetric case. the profit function in the third stage for the firm are (72) (79) Note that q* is decreasing at ¿' = 0 if S < 1/(2 + ß) and is increasing if ^ > (1 . C = 0 and 1 = 1. note that the full-coverage assumption implies that 2. and therefore there must be a symmetric equilibrium in quality even if a* 5^ «j.al) and Tl23{q2. both do not allow copyequilibrium profit functions in the last stage.

If it deviates and restricts copying.24y . however. It can be shown that if 7 > 7^. The difference in profits is given by 1-35 2(1+ 5)(2-77)2 < 0. we need to check that the condition that copiers copy rather than buy holds.y) . = 1. By substituting the equilibrium qualities in the profit functions.9ky . then firm 2 will not deviate and the equilibrium would hold. 16(1 + 5) (96) (103) Now.275) (95) Finally.94. then Aj > 0. If the difference is positive.605y -9-1. 1 2(2-- Note that 3(9/c . we need Aj > 0. where 75 +15 . we also check whether a firm can deviate and offer the product only to the noncopier segments. In this case. It can be shown that this is true as long as ô > (—5 + 4v'2)/7 « 0. We first consider firm 1.. We start the analysis by considering the case when both firms do not allow copying.v'4902 -1745 + 225 72 = Note. = 1. consider the asymmetric equilibrium. We need to show that neither firm would like to deviate. » Now. Assume without loss of generality that firm 1 allows copying. Also. (101) (94) which shows another part of the proposition. then copiers would prefer to copy. 72). then the profits it obtains are given by (86). D PROOF OF PROPOSITION 7. The analysis is similar to that before. after simplification A. then the asymmetric equilibrium holds. If the difference As we did in the proofs of Propositions 5 and 6. we need to establish that the asymmetric equilibrium holds under the conditions specified. y•^ < y^ for the relevant region. (90) (91) Also. pp. firm 1 would not deviate and would also allow copying. where 5* = —»7^ . If it does. Only One Firm Alloivs Copying (97) Without loss of generality.Jain: Digital Piracy: A Competitive Analysis Marketing Science 27(4). D PROOF OF PROPOSITION 6. To see which equilibrium dominates. Also. we assume that copiers prefer to copy rather than buy 2's product. and find that under the conditions specified in the proposition. In this case. we have 25(10-277)(50-27y) 27(9y-10)2(10-37) • For the asymmetric equilibrium to hold. . which where the second inequality follows since (q* — i/J) > 0. we get after simplification {9k-\-9ky)(6ky + 2(92) . the asymmetric equilibrium is valid. both firms allow copying. the asymmetric solution does not hold because firm 2 would like to deviate and allow copying. we solve the game as usual and obtain (99) It can be shown that both firms would choose to serve both the segments if 5 > 5*. we can show that y > 72. consider firm 2's incentives to deviate. then A2 > 0. ©2008 INFORMS 625 A2 > 0. this reduces to > ^ + 327^8 . We also find that in this situation. it is an equilibrium for both firms to set a = 1. This proves one part of the proposition. Again. we find that the equilibrium qualities are 32 • (98) 9fc -6ky-2 ' 3fc(9fc(l . Now. it immediately follows that the asymmetric solution is weakly dominated by the solution in which both firms allow copying. This also establishes that the symmetric solution will not hold because at least one firm benefits by deviating. To complete the proof. if 5 G (5*. However. we compare the profits in (99) and (101). 610-626. where 3(2 + 55 . assume that firm 1 allows copying and firm 2 does not. that we need to check in this case that both firms would choose to serve both the segments. Thus we have established that if 7 < 72. where A2 reduces to. we prove that under the conditions specified in the proposition. This completes the proof. .V12 + 45 + 52. (93) 12-772-477 (100) Now. We have „ . consider the case when a. After simplification and taking the limit asfc— 00. then the profits are given by (87).4 + 4^2772 + 4 . consider the case when y e (y.4 i . then the firm does not deviate.0938. = .2) 9fc-12fcy-2 3fc(9fc(l-7)-2)' Note that (88) (89) 9fc7+2-9fc Also. It can be shown that if S < ^8/9 = 0. no copier vvould purchase product 2. 1/3). in this case. the firm will not deviate and the equilibrium profits calculated are valid. With this. The profits for both firms in this case will be the same as the case when a. (102) Thus. In this case.2) > U. Therefore. After straightforward calculations.

1989. Rinne. 37 351-362. Xie. P. Shaffer. 74(2) (September 26).27. 2004. Marketing Sei.. 1997. Copying and indirect appropriability: Photocopying of journals. 2000. as Beijing stiffens penalties for the piracy of products.. Mahajan. J. and optimatch price. 2006. Sanders.10) 25(50 . Obtaining strategic advantage from being imitated: When can encouraging clone pay? Management Sei. /. K. 41(2) Swinyard. Management Strategy 8(2) 163-190. Piracy helps. 2004. Cross-market network effect with asymmetSei. continuous economic games. Murphy. the puppy-dog Williams. 2000. Lovelace. Econom.. The profits in this case are given by (103). 1985. 1986. Sundararajan. R.70)2(3y . Software piracy: Estimation the presence of positive demand effect. C. Forbes 172.. China pumps up limp protections for trademarks. Contributions to Econom. An experimental and theoretical analysis of price matching refund policies. The economics of copying. Managing digital piracy: Pricing and proand standardization. /.. Systems 13(4) Sun. ©2008 INFORMS Now. Shy. 2003. the asymmetric equilibrium holds. Maskin. Purohit. A. Preventive and deterrent controls for munications strategies when firms compete on multiple channels. /. Competitive coupon targeting. Management Inform. 53(1) Varian. if y > 0. /. Amer. M. Zhang.. A strategic approach to software protection. The day after tomorrow: Longer run issues in theatrical exhibition. Copying and copyright. Princeton. we examine whether firm 1 can benefit by deviating and not allow copying. Mahajan. duction of intellectual property in the presence of demand Dasgupta. 40 85-104. B. Sei. A. Kirs. I. M. Expanding to the Internet: Pricing and comGopal. 1991. /.9007 + 81 y^ 36(-10 -t. Piracy study. Asian Wall Street ].10) ' (105) (106) Now. 25(6) 665-666.. We also check this condition and find that it holds under the conditions in the proposition. Johnson. then firm 1 will not deviate. L. /. Indust. 361-366. Indust. 2003. Profits from piracy. (50 -9y)(3. The enemy strikes back. Marketing Sei. J. A. We find that if 7 > 0. 1985. tection. J. Rate and Direction of Inventive protection: An analytic approach. ric customer loyalty: Implications for competitive advantage.. Economic welfare and the allocation of resources Novos. Political Econom. 1984. Management Sei. the equilibrium profits for the firms in the asymmetric case are 1V Gu. Also. What should you do when your competitors send Cabrai. X. E. /. Perspectives 19(2) 1-26. Liebowitz. /. 1995.10)2(y . Econom. 121-138. R. Gandal. Prasad. Rumelt. Policy 2(1) 1160-1184. H. Marketing Sei. Waldman. 9-10) 10. The profits in this case are given by (104). We will assume that copiers copy product 1 rather than copy 2. 655-664. Png. Marketing Chen. D. B. K. http://www. Network effects. 1995. Srivastava. The Wall Street /.70)2(37 . Rev. Management Inform. C. 26(1) 52-66. Cheap talk and bogus network externalities. J. Econom. Austin. 2003.626 The analysis for the case when both firms allow copying leads to 144(37-10) (27y2-115y+50)2 (104) Jain: Digital Piracy: A Competitive Analysis Marketing Science 27(4). 14(4) 395-416. 236-246. J. How much anti-piracy effort is too much?—A study of the global software industry.bsa. 93(1) 158-174. Global software piracy: Can Peitz. V. Econom. http://www. Hutzler. Network externalities. Kau. I. Law Econom. Prasad. C. K.920y + 700)^ 36(-10 + 97)(97 . which completes the proof. V. W. The welfare implications of unauthorized reprotection strategies. Econom. /. Haruvy. Internat. 1995. Ethics 9(8) 209-225. Product strategy for innovators in markets with network effects. 75 424—440. Cao. H. ed. Sirbu. 60(3) 365-383. 2005. /. Anal. 1985.org. 2004. Bus.. pp. Rev. 2002. Arrow. J.. 51(4) 599-614. Management Sei. J. Cerveny. Princeton University Press. consider the asymmetric case in which firm 1 allows copying and firm 2 does not. 2003. S. Marketing 909-926. The morality of software piracy: A cross-cultural analysis. W. Conner.n. O. Png. S. Business Software Alliance. Fudenberg. V.. M. 2004. Shapiro. Rev. M. Teegen. Zettelmeyer. Management in the clones? Marketing Sei.9y)(117y2 . D.500... S. K. S. 37(2) 125-139. 1984. G. We find that if y > 0.128. Conner. 243-254. Y.1. Marketing Sei. /. V. 610-626. 32 255-280. Political Econom. Church.ll(Nov 24) References 218. 42(2) 155-166. 13(4) 392^11.. H. The existence of equilibrium in disexternalities.salon. Systems Res. Bus. 1995. Png. 70-75. Thisse. Price discrimination through offers to Besen. University of Texas. H. Moul. 1997. A.com ploy. Amer. M. I. By solving as usual. Working paper. P. Econom. Hui.. software provision. MuUer. 1999. 1987. R. we need to examine whether firm 2 can benefit by allowing copying. The fat-cat effect.. I: Theory. Private copying. competition and compatibility. Marketing 20(4) 337-352. Inform. 24(4) 531-543. Cheng. Econom. L. E. then firm 2 will not deviate.. Nelson. 37(3) 292-308. H. /. 59(1) 29-37. 1994. R. Marketing Res. 1962. /. R. 15(3) 287-308. Management Strategy 13(2) 371-374. F. Bus. /. 1996. Xie. Hirshleifer. /.. 41 of lost sales and the impact on software diffusion. G. Econom. 609-625. Software piracy: An analysis of proTakeyama. Marketing Res. E. Katz. 2005. Isn't tax time already frustrating enough? ¡nformationWeek (9/29) 96. I. and the lean and hungry look. .. D. Res. /. Political Econom.. Mahajan. J. 23(2) 49-60. 1992. /. To purchase or to pirate software: An empirical study. 2006. Bertrand supertraps. The effect of increasing copyright for inventions. Xie. mal copying royalties. 1994. (Dec 22) A. Stud. software piracy. Price competition and compatibility in Givon. D Mahajan. N. 2004. Jain.. H.27.. Li. Villas-Boas. 2005. Piracy and the legitimate demand for recorded music. NJ. How many pirates should a software firm tolerate? An analysis of piracy protection on the diffusion of software.. The effect of piracy on the market penetration of subscription software. enforcement still is suspect. 77(2) S81-S106. Kirby. 2004. M. Systems 13(4) 29-47. Bagchi. Sims. 1990. 93 945-957. appropriability. 2005. J. S. Thus. A strategic approach to software protection: Comeconomic factors alone explain the trend? Comm. K. Wu.10)^(7 ... V. (Feb. 2007. ACM 49(6) ment. R. 92(2) Activity.9y)(9y . C. Tiróle.

Although such a contracting practice creates a considerable administrative effort and channel conflict.l070. Dayalbagh Educational Institute. India.com Sanjeev Swami Department of Management. to capture strategic behaviors of channel members in a complex market environment. India. starting with the widely promoted The Greatest Game Ever Played and concluding with The Legend of Zorro. July-August 2008. British Columbia V6T 1Z2. Weinberg Sauder School of Business.g. processed by Koen Pauwels. We develop and analyze a game-theoretic model using the genetic algorithm approach and a decision support system. vied for the limited number of screens in a month that is not even . eleeO6@syr. limited shelf space. in the five weeks of October 2005. Jeuland and Shugan 1983.com Eunkyu Lee Whitman School of Management. pp. despite their valuable contributions. Channel contract structure has significant impact on channel member profitability and the exhibitor's movie-scheduling behavior. In such a market environment. In this paper. and consequently. and new products are frequently introduced to replace old ones. sumit. Moorthy 1988.ubc. . Agra 282110. No. then. the complexity of the market environment need not be reflected in the complexity of the channel contracts. with the demand being the highest in the first week after mass market release and then decreasing noticeably each week (27% per week. University of British Columbia. our results indicate that the flat rate contract structure represents an attractive alternative to the current practice for distributors. Ingene and Parry 1995. 627-641 ISSN 0732-23991 EISSN 1526-548X108 ¡270410627 DOI 10. the market is characterized by frequent new product introductions.raut@tcs. 2006. Key words: channel contracts. these studies have largely ignored three important characteristics of the marketing environment—dynamics. it is dynamic. 1. movie industry. in their sample). An example of such a complex market environment is found in the motion picture industry First. Mumbai 400096. we investigate the impact of charmel contracts on channel coordination and individual channel member performance in such a complex real-world setting. and product category management. 2008. Syracuse University. Chuan and Chen 2005). on average. New York 13244. 4. This study investigates this question by analyzing the impact of movie contract structure on movie scheduling and charmel member profitability.0315 ©2008 INFORMS How Complex Do Movie Channel Contracts Need to Be? Sumit Raut Tata Consultancy Services. In particular. Lehmann and Weinberg (2000) show that the weekly box office revenue for a movie typically follows an exponential decay over time. Purohit 1997. Chu and Desai 1995. 22 new movies. Introduction Designing optimal contracts between manufacturers and retailers is one of the most critical factors affecting channel coordination and the relative profit share of each channel member. and was with the authors 11 months for 1 revision. We find that simpler two-part tariff or 50/50 split contracts perform as well as the current contracts. sswamil853@gmail.edu Charles B.. Constrained by limited shelf space. Published online in Articles in Advance March 31. Consequently. charles. typically taking the approach of a static game-theoretic analysis within the context of one or a few products being marketed through established retail outlets.ca T he motion picture industry is characterized by a dynamic market environment. An extensive theoretical literature has investigated this issue (e. Dayalbagh.weinberg@sauder. Thus. Li 2005.MARKETING SCIENCE Vol. 627 must optimally decide when to carry which products to maximize the total profits generated from the product category over a period of time.1287/mksc. 27. Syracuse. Vancouver. Second. complex channel contracts specifying the split of box office revenue between distributors and exhibitors. Canada. retail managers. 2005. SilverScreener. it is not clear whether such complexity is necessary for superior channel performance. limited shelf space and product category management. existing products experience demand decay over time. game theory History: This paper was received April 10. Villas-Boas 1998. For example.