Communications of the ACM

Communications of the ACM

Digital Music and Online Sharing: Software Piracy 2.0?

By Sudip Bhattacharjee, Ram D. Gopal, G. Lawrence Sanders
Communications of the ACM, July 2003, Vol. 46 No. 7, Pages 107-111
10.1145/792704.792707
Comments

Rapid advances in Internet connectivity and digital compression technologies have dramatically increased online sharing of digitized material, raising issues of intellectual property rights and lost sales. For instance, online music sharing has prompted legal challenges and industry alliances, while raising significant concerns regarding the industry's future. A study in 2000 reported 14% of Internet users had downloaded music for free [11]. This number has grown rapidly, and online music sharing has been estimated to result in annual sales losses of $3.1 billion by 2005 for the music industry [2].

The effect of such sharing—induced by technology—is inconclusive. Proponents—consumers and some artists—argue that many consumers download music to sample and subsequently purchase a CD¹ if they like the music, hence it provides an effective advertising channel [1]. It also benefits recording companies by helping new artists to become known at little cost to the companies themselves. Opponents—in particular the recording industry and some other artists—argue that it undermines sales. Their fundamental concern is that piracy threatens the business of artists, composers, and producers.

Despite its scope and publicity, little research exists on this problem from a consumer's economic perspective. Substantial research has examined the related problem of software piracy [1, 6–8]. While software and music are both information goods in that the marginal cost of production is virtually zero, certain key characteristics differentiate the two: audio quality of the original CD is better than its electronically transferable compressed version. Software, however, requires a loss-less compression for proper functioning; music files are much smaller than a typical software application, and take much less time to transfer; individual audio files are not perceived to be as expensive as average software packages, and the volume of available audio files is significantly larger than the existing volume of software products.

The phenomenon of sharing music files online has been accelerated by various software packages and increasing Internet connection speeds. With decreasing data storage cost and higher bandwidth, users are able to send large collections of music via email [2]. It is important to recognize this phenomenon is not limited to audio content. Other digitized goods (movies, photographs, published articles, for example) are increasingly shared as technology improves.

Here, we strive to understand individual motivations to "freeload" digital music, such that affected industries could develop effective measures to combat the problem. Online music sharing is analyzed from economic, technological, and demographic perspectives, and contrasted with software piracy. In particular we study whether: consumers' music purchasing behavior is affected by free music; the effects of online sharing are consistent across types of music; and how alternative pricing and distribution mechanisms would affect consumer behavior.

Our results suggest that, under certain conditions, sharing digitized music can serve a useful marketing function, and that demand for music depends critically on price, distribution mechanisms, technology, and the type of music. While our results suggest a number of similarities with software piracy, some unique characteristics have important implications for the music industry and technology providers in particular, and other digitized products industries in general.

Back to Top

Digital Music Piracy: Lessons from Software Piracy

Digital goods have high initial production costs, and very low—approaching zero—reproduction costs. They also have characteristics of a public good in that sharing with others does not reduce a consumer's utility for the product. These traits facilitate their widespread and often illegal distribution worldwide.

Software piracy research suggests that increasing software prices increases piracy [3, 6, 8]. Also, consumers with a high value for a product typically tend to purchase rather than pirate [3, 4, 7]. Additionally, in the face of increasing preventive controls (encryption and other technologies), individuals who do not legitimately own the digital good simply do without it [6]. This represents an opportunity loss on producer profits. Females and older individuals (as opposed to younger college students) pirate less, and individuals with an ethical disposition tend to pirate less [5–7].

We focus on the factors and relationships summarized in Figure 1. Specifically, we study whether:

• CD pricing, consumer income and Internet connection speed (bandwidth) affect piracy (bandwidth measures the ease of copying music);
• Value of music to a consumer and quality of online music affect piracy;
• Demographic factors influence piracy;
• Consumers' available bandwidth influences their music purchase decision; and
• Piracy provides a word-of-mouth advertising, leading to "sampling" that leads to a future purchase [5].

Behavioral models, which are important to initiate educational and legal campaigns to reduce piracy, are investigated in detail in [9].

Back to Top

Online Music Experience Survey

Young people form a significant portion of the music fan base, and online music copying and sharing is rampant among students. We surveyed over 200 respondents during 2000–2001 as part of an ongoing study of consumer attitudes toward online music. Respondents were primarily enrolled as full-time (15%) or part-time (54%) students in colleges; ages ranged from 19 to 54 years, with 61% males. A total of 52% of the respondents reported a very high level of interest in music, while another 37% listened to music regularly. The sample group is sufficiently diverse in demographic, economic, and social aspects, and represents a significant component of the music industry customer segment. Respondents were asked to reveal their online music experiences, provide some demographic information, and specify preferences for certain online music activities under different music CD pricing schemes (economic) and Internet connection speeds (technological).

Online music sharing is conceptualized as a four-step process, where economic, technological, and demographic factors guide a consumer's decision at each step. Given a music file, and an economic and technological scenario, a consumer is asked to take one of six actions, as shown in Figure 2. Once a user searches and finds a certain file online, the user may decide to download, listen, and purchase it based on: a "value" the user attaches to it, the user's connection speed, and the price of the related CD. For example, if a user downloads, listens to, and deletes the music file, then buys the CD of the music, the user has performed Action A shown in Figure 2. Similarly, a user performs Action D when finding some music, downloading, listening to, and keeping a copy of the music, but does not purchase it. Actions A and C are considered sampling², as these lead to purchases. Action B also constitutes sampling and is similar to listening to a radio broadcast. Action D constitutes freeloading³ or piracy, because files are downloaded and kept with no subsequent purchase. This may also lead to subsequent sharing of the file. Actions E and F constitute traditional purchasing decisions.

If a consumer has listened to particular music before, its value is assumed to be "known" to the consumer; otherwise it is "unknown." Consumers are presented with five music categories:

Two categories of known music:

• Category (i)—one of their top five songs (to which they attach a high value); and
• Category (ii)—one of their top 50 songs (to which they attach moderate value).

Three categories of unknown music:

• Category (iii)—unheard music from their favorite artist;
• Category (iv)—unheard music from the type of songs they like; and
• Category (v)—unheard music recommended by a friend or colleague.

Back to Top

Key Analyses and Implications

Comparison of Music Freeloading and Software Piracy. Overall, we find certain key factors that influence software piracy also extrapolate to music piracy. Females display a significantly lower tendency to freeload music than men. Older individuals pirate less; however the effect is small. Previous research has similarly categorized young males as archetypal software pirates. While the demographic profile is similar to that of software pirates, implications to the industries are dramatically different. Younger individuals constitute a minor consumer segment for typical business software, however, they comprise a significant portion of the music industry's consumer base. Therefore, the demographic effect is likely to translate to a larger drain on the bottom line for the music industry than for the software industry.

Increasing the price of a music CD has a strong effect of increasing its related online piracy, pointing to a fairly elastic demand for music. This is similar to demand for software products, despite the significant price differential between the two. An average music CD retails for approximately $10–$15 and is significantly less expensive than typical application software. Additionally, as prices increase, piracy of known songs (categories (i) and (ii)) increase almost twice as fast as unknown ones (categories (iii) to (v)). This indicates the need for differential pricing for different music categories. Coupling individual music downloads with emerging micropayment methods may also yield promising results.

Inclination to pirate music increases dramatically as Internet bandwidth improves, with similar trends for all music categories. This suggests consumers would pirate more music as the ease of piracy increases.

Income has a negative effect—only for unknown songs, suggesting that individuals with lower incomes are likely to pirate rather than purchase and sample new music, based on current prices. Absence of income effect for known songs suggests the decision to purchase music containing favorite songs is not significantly influenced by disposable income.

Perception and Impact of Online Audio Quality. Digital audio from CD loses some of its sound quality in a compressed format such as MP3, which compresses the original audio source at rates up to 12:1. The online download process itself leads to files becoming corrupted or to incomplete downloads, further deteriorating audio quality. In fact, 60% of respondents reported experiencing corrupted downloads frequently or at least occasionally.⁴

Interestingly, more than 90% of respondents rated compressed music quality as almost the same or very good compared to CD quality, as shown in the bar chart in Figure 3a. The quality perception did not vary across demographic and economic factors. This suggests that with widespread prevalence of compressed digital audio, users have become accustomed to and now accept this level of quality. Also, it is probable that users' demands for degrees of quality are influenced by free availability of digital audio offerings. Other results also indicated the quality of downloading music alone is not a significant determining factor leading to a purchase.

This has crucial ramifications for the recording industry. Pricing and other economic models that rely on audio quality are unlikely to be successful under the changing circumstances. Moreover, with improving compression technology and faster connections, even higher-quality digital files will become available, which would severely undermine marketing efforts that rely heavily on acoustics.

Consumer Willingness to Purchase. A natural corollary of the non-impact of quality is that as users' Internet connection speeds increase, their price sensitivity to purchasing music increases. On average, for category (i), a consumer with a slow connection is willing to pay 8% more (for CD) than one with fast connection. For category (ii), the difference increases slightly to 8.4%, while for unknown songs it is 7.4%. Hence users with high bandwidth are willing to pay less than those with slower connections. Importantly, this result holds true across categories clearly implying that users expect to find freely shared music online. In terms of pricing, as more music becomes available online, users with fast connections would rather download and listen than buy. This is similar to results obtained in software piracy research—as copying software becomes easier (hence convenient), the willingness to purchase decreases. Hence as more music becomes available online (network effect), and as more consumers upgrade to high-speed connections (technology effect), the offered price should decrease from current levels to positively impact a user's buying decision.

A user's willingness to pay also depends on perceived value of music, as shown in the bar chart shown in Figure 3b. For example, users are willing to pay over $5 more for a CD with a category (i) song than an unknown song. In all comparisons between categories, a user is willing to pay more for a CD with a known song than for one with an unheard song. This is natural, considering the speculative risk of buying unheard music. The price difference is also affected by users' available bandwidth, and is consistently lower for consumers with high bandwidth. This again seems to imply that consumers expect to find and download music online.

The results suggest current uniform CD pricing system is suboptimal under a changing technological scenario. The music industry should introduce newer economic models of pricing that incorporate both the value of music to consumers and technological factors that affect sharing. For instance, two complementary approaches can be employed to increase users' buying probabilities: users should be provided more exposure to new music in order to familiarize them, which would potentially result in a higher willingness to purchase, and the pricing structure should be more flexible for unknown music.

Technologically, copy protection mechanisms are evolving to prevent users from copying digital music files. Similar to the effects of preventive controls in software piracy [6], music listeners may simply do without the copy-protected goods. Further study is warranted to determine consumer acceptance of such technologies.

Issues of Piracy vs. Sampling: For categories (i), (ii) and (iii), downloading clearly leads to piracy. For other categories, neither piracy nor sampling is evident in a statistical sense. Hence consumers with high perceived value for music primarily pirate it, and might sample if the value is completely unknown. Overall, results point to the existence of piracy across categories, with indirect evidence for sampling for categories (iv) and (v). Further studies are ongoing to determine the effects of such sampling.

---

Recent variations of the subscription-based model are beginning to provide consumers greater flexibility in purchasing and listening to digital music.

---

Online Subscription. Users with high-bandwidth connections (via cable modem or DSL service) indicated they were willing to pay over $10 per month for an online music service, while those with slower connections (56Kbps modem or less) were willing to pay about $5. This demonstrated a significant positive price sensitivity on subscription rates based on Internet connection speeds. It is interesting to contrast here our earlier result that users with high-speed connections were willing to pay a lower price for a CD than those with slower connections. This strongly suggests users find subscription services more convenient, and hence fee-based subscription would become a better model due to its ease of use and greater customer convenience. However, doubts have been raised about the viability of subscription-based business models in their current form [10].

The results did not vary with the age of consumers. However, with a fast connection, men are willing to pay higher subscription rates than women; with a slower connection, women are willing to pay more. This suggests that gender-based, rather than age-based, marketing strategies are likely to have greater impact for signing up subscription members.

Back to Top

Conclusion

Our study of online digital music sharing highlights several key influential factors. Sharing audio files shows some similarities as well as uniqueness compared to software piracy. Price of music and available bandwidth are found to have significant effects on piracy. The price impact becomes more pronounced as technology improves. We find existence of piracy across all music categories, and weak evidence of sampling for "unknown" music. Interestingly, the perceived quality of compressed audio did not seem to play any significant role. We also find viability of subscription-based models, which exhibit sensitivity to gender differences and differentiated pricing based on bandwidth. These insights could influence enhanced pricing models in the future. Recent variations of the subscription-based model, for example, Apple Computer's á la carte online music service iTunes, are beginning to provide consumers greater flexibility in purchasing and listening to digital music.

Other industries that produce digital goods and face similar problems need to be studied. In particular, the TV and movie industry has already felt the pressure, and it is only a matter of better compression technology and increased bandwidth before full-length movies are shared the same way as music files [12]. The global scope of the Internet calls for the development of generalized models for information goods that are supranational and that transcend cultural, legal, and economic barriers.

Back to Top

Back to Top

Back to Top

Back to Top

Back to top

The Digital Library is published by the Association for Computing Machinery. Copyright © 2003 ACM, Inc.

---

No entries found