Economics of Online Music and Consumer Behavior Prof. Dr. Marc Fetscherin Rollins College 1000 Holt Avenue - 2723 Winter Park, FL 32789 +1 407 921 1792

[email protected] marginal investment such as buying an empty cassette and the quality of the copy was inferior to the original.

ABSTRACT In this paper we first provide a literature review about the economics of online music. Second, we then present an economical model for online music by incorporating various parameters such as the price of a legal download, as well as parameters taking into account the quality and security risk associated with purchasing or copying music form the Internet. Furthermore, we incorporate legal and technological parameters in our model, such as law enforcement and its underlying penalties, as well as rights and usage restrictions enforced through so called Digital Rights Management Systems. We then present the optimal choice for a consumer for purchasing online music with the presence of online piracy. Finally, we discuss three possible scenarios. By using empirical numerical values we establish a link between our model and practical application. Each scenario is discussed in detail and we conclude that content providers should either demand a higher price but not restrict consumers or demand a lower price but restrict consumers more. However, doing both does not seem to motive consumers to purchase online music but rather copying it.

Nowadays, with technology advances in hardware and software, copying music has become much easier, cheaper and the quality of the copy is comparable to the original one. Thanks to Digital Rights Management Systems (DRMS), stronger law enforcement against illegal file-sharing makes copying more difficult. Moreover, emerging innovative business models [19], such as is the case of i-Tunes, provide consumers legal alternatives of purchasing music. The fact that i-Tunes sold more than 1 billion of such downloads in the first few years of operation1 shows that there are consumers willing to pay for legal downloads despite the fact that copies are freely and easily available [13]. Therefore, there are a number of questions which arise, such as: why and which factors make consumers buy digital music legally or to copy? Why do some consumers prefer to buy digital music whereas others prefer copying? Are there any microeconomic theories which might explain this behavior or do consumers follow random behavior and instinct? Despite these issues in practice and emerging interest in the research field of consumer behavior, psychology or economics, surprisingly current literature has not discussed these issues very broadly. Most papers dealing with digital content and piracy have focused on software rather than on music ([24]; [5]; [21]; [22]; [5]; [1]; [18]; [7]). Those focusing on music either look at the implication of copying on welfare or on the music industry ([4];[16]; [17]), or at the strategies content providers could follow to fight piracy [3] ; [2]). The few papers outlining the factors motivating consumer to pirate provide a good starting point and served as a basis for identifying some of the key parameters for our model. However, those models lack the modeling part of the consumer tradeoff between purchasing and copying ([15]; [25]; [6]; [14]). This paper wants to close that gap by focusing on consumer behavior when faced with the option of either downloading it legally or copying it. We propose a linear model where various influencing variables, which have an effect on consumer behavior and might be explained by consumer theory, will be taken into account. Our model should provide further insights into consumer theory for digital content by taking the music industry as an example.

Keywords Online Music, Consumer Behavior, Digital Rights Management

1.

INTRODUCTION

The sales volume of the music industry has decreased significantly the last few years. One of the main reasons, according to the industry, is the massive illegal sharing and downloading of music on peer-to-peer networks. These networks enable consumers to share, distribute and copy digital content, such as music, very quickly, easily and cheaply. As a result, consumers have new possibilities to access and acquire music which were previously not possible. Before the Internet, a consumer basically had two possibilities to access music: taping the music from the radio station or buying a disk or CD. Copying was very time consuming, required a Permission to make digital or hard copies of all or part of this work for personal or classroom use is granted without fee provided that copies are not made or distributed for profit or commercial advantage and that copies bear this notice and the full citation on the first page. To copy otherwise, to republish, to post on servers or to redistribute to lists, requires prior specific permission and/or a fee. ICEC’06, August 14–16, 2006, Fredericton, Canada. Copyright 2006 ACM 1-59593-392-1.

In order to more closely investigate this issue, one has to first define the term digital music. Digital content, i.e. digital music, compared to a conventional CD, is different in four main 1

599

http://de.wikipedia.org/wiki/ITunes_Music_Store

characteristics: the first characteristic is indestructibility, in the sense that a digital product is not likely to age like the underlying physical medium such as a CD [8]. There is also no quality degradation associated with the digital product. The second is modifiability, meaning a digital product can easily be modified, customized or personalized. This also makes a digital product vulnerable to fraud and piracy. The third is reproduction: a copy can easily and quickly be reproduced and copied, in most cases without any quality degradation. Thus, a copy of a copy, of a copy is still as good as the original copy or the original itself. This has an implication on the cost structure of a digital product, which is mainly characterized by high (initial) fixed costs and low or zero marginal costs of reproduction and distribution. The fourth characteristic is divisibility. It is possible to divide a digital product into smaller pieces which could be reused somewhere else. Hence, a digital product which can be copied easily and the use of the product is not restricted to one person has the characteristics of a public good. A public good is one which is non-rivaled in consumption and non-excludable. With the introduction of Digital Rights Management Systems (DRMS), the music industry has tried to create at new rivalry and excludability [3]. However, the economic implications of the introduction of DRMS will be further discussed later in this paper.

2.1 Assumptions There are a certain number of assumptions for the underlying model. First, all consumers behave rationally and have similar preferences and make their decision based upon their preferences. They always choose the product with the highest utility [20]. Second, it is assumed that there is only one price for a legal download and there is no price discrimination among products and consumers. Finally, we omit risk factors which do not have a direct effect on a download, such as a world crisis or demographical changes [20].

2.2 Basic Model We define x and y as the number of legal downloads, respectively the number of copies. Each consumer i, whereas i ! I and I represents the total number of consumers, has a budget constraint mi that he is willing to spend on either xi, or yi or a combination of these two products. When downloading the music legally, the consumer pays a price px, or a price py when copying2. The price px is fixed by the music providers and is assumed to be fixed for all consumers and downloads. The price for the copy py tends to be 0, lim py ! 0, as the copy can be acquired for free, but this will be shown later in this paper. In addition to the price to be paid, there are other costs associated with the acquisition of digital music, such as access, storage and search costs. Access and storage costs such as Internet access costs or hard disk space are assumed to be equal for both acquisitions, since in both cases, one has to have access to the Internet and needs a combination of hardware and software such as a computer and a media player. Therefore, they can be omitted for further analyses [9]. Furthermore the search costs, which is the monetary expression of the time needed to search for a song, is shorter for legal downloads than for copies, according to empirical analysis [11]. However, since the legal download requires additional payment information such as credit card information, this increases the duration and we will therefore assume, for simplicity, that the search costs are equal in both cases and are therefore also omitted in our model. We put nxi as the utility, or the value consumer i associates with the product x he is willing to pay for, the same applies for nyi. The effective utility or net benefit results from the difference between n on one side and p on the other side, thus nxi – px, resp. nyi – py. Table 1 summarizes the parameters outlined so far.

This paper is structured as follows. The following section presents the basic model and its underlying assumptions. The next section extends the basic model and presents additional parameters which have to be considered when modeling consumer behavior for digital music in the presence of copies. Section four outlines three scenarios which might reflect real world cases and show under which circumstances consumers might change their behavior. The final section provides a conclusion and a discussion of future research in this area.

2.

THE MODEL

On one hand the model should be as complex as possible in order to reflect the reality, on the other hand, it should be as simple as possible in order to enable its readability and to facilitate the understanding of which parameter has what kind of effect on consumer behavior. We distinguish between two acquisition routes: One is the legal download from music providers or web portals, the other the copying of music from websites or filesharing systems. We further assume that both products, the legal and the copy, are perfect substitutes in respect to their quality. This enables us to keep our model linear. Two products are perfect substitutes if the consumer is willing to substitute one product for the other at a constant rate [20], where it is assumed that the consumers substitute one unit of the legal with one unit of the illegal and vice-versa. Furthermore, this paper presents a model from a single consumer’s point of view; it does not provide an aggregate model showing the demand for digital music, nor the supply side. Thus, this paper does not discuss market equilibrium.

Table 1. Overview Parameter

Nevertheless, this paper deals with important questions: How do consumers react to restrictions on digital music? What effect will law enforcement have on consumer behavior? What effect does quality has on consumer behavior? Our model answer those and other questions and shows how they affect a consumer’s decisions.

2

600

Parameter

Description

I

Total number of consumers

i

Consumer i! I

{xi, yi}

Consumption bundle for the consumer i consisting of a quantity of product xi and yi , whereas " xi = x und" yi = y

p x, p y

Unit price for xi respectively yi, where py ! 0

nxi, nyi

Willingness to pay or utility of consumer i for the product x or y

Our basic model does not take into account an increase of income or any changes affecting the price of the legal download such as taxes or subsidies.

sales and could have closed its business. Thus, there must be other factors which make consumers buy digital music instead of copying it. What follows is an extension of the basic model with further parameters taking into account the latest technological, legal and business developments.

The budget constraint, which is the amount of money spent by consumer i on the two products can be no more than the total amount the consumer has to spend. It can be expressed as mi = pxxi + pyyi slope – px/ py In this case, pxxi is the amount of money the consumer is spending on the legal download and pyyi the amount of money the consumer is spending on the copies. The slope of the budget line measures the rate at which the consumer is willing to “substitute” the legal download for a copy and vice-versa [20]. The indifference curves for the two products can be expressed therefore as

2.3 Extending the Model There are a couple of risks associated when copying from filesharing systems. There are specialized companies3 flooding those networks with fake or corrupt files or files with interfering signals which make the file not playable most of the time [10]. Moreover, when copying form these networks one also faces the risk of getting viruses, either making the downloaded file unplayable or even damaging the computer4. In both cases, there is a direct risk associated with the download which makes the file unusable.

U(x, y)= nxi px + nyi yi slope – nxi / nyi

There is also another risk that users are exposed to, namely the risk of being prosecuted for copyright violation. Thanks to Digital Rights Management Systems (DRMS), it is possible to track music files, and thus users of such systems, which enables the prosecution of individuals for copyright violations. Since 2003, thousands of users have been prosecuted, not only in the United States but also in Europe5 and Asia. In this respect, we assume in our model that law enforcement is only possible with the usage of DRMS. However, DRMS also restricts the consumer in the usage of the download, as the consumer might not be able to copy the file, burn it onto a CD or other devices [13]. Thus, DRMS on the one hand, improves law enforcement for copying but on the other hand, restricts the access and usage of a music downloads.

On the indifference curves, the consumer is indifferent whether to buy product x or y. In the case of perfect substitutes and a consumer’s indifference to having a legal version of a song or a copy, the slope of the indifference curve is therefore – 1 [20]. Since we assumed in our basic model that the access, storage and search costs for a legal download and a copy are equal and the quality difference between the original and the copy is also equal, we could write nxi = nyi. In that case, consumer i is willing to substitute one unit of the legal product for one copy. Thus, the marginal rate of substitution (MRT) is 1. Figure 1 shows the transition of a consumer from having the possibility of copying music by taping on a cassette to a situation where he can download the song illegally over the Internet. In case the consumer is taping music, he has certain costs to bear such as buying a cassette. To simplify this case, we assume that px = py. However, the focus of this paper is on digital music, where copying music, compared to taping, is almost free and therefore py ! 0. Assuming that nxi = nyi, the budget line tends to have a slope of -! and implies an optimum on the border of the yaxis.

2.3.1 Quality and Security Whether dedicated companies are flooding file-sharing systems with corrupt or fake files or whether the file contains a virus, in both cases the quality of the copy is inferior to the original, mostly it is not usable at all. Therefore, this risk is expressed in our model with the parameter ", where " ! [0, 1]. It reflects the probability of the copy not being playable, where " = 1 indicates the copy is not usable at all and " = 0 indicates there is no quality difference between the copy and the original. The utility of the copy can be expressed as the utility of the copy multiplied by the factor 1 – ", which takes into account the potential quality differences. In the case of perfect substitutes we can express the following utility function as: U (x,y) = nxi xi + nyi(1 – ") yi slope –1/(1 – ") There is no effect on the budget line, as the quality affects only the utility of the copy.

2.3.2 Rights Restrictions and Enforcement There is a certain probability u, where u is expressed as ! [0, 1], that consumers are being prosecuted for copyright violations when copying music from file-sharing systems. In that case, the Figure 1. Optimal choice for consumer The arrow on Figure 1 shows the shift of the budget line when a consumer, who used to taping music on a physical medium, switches to copying music digitally. In that case, the consumer i chooses the copy over the legal download as indicated by point A. If this would be the case, i-Tunes would not have generated any

3 4

5

601

Examples such as Overpeer or Audible Magic Risk of virus damages on the computer are excluded in this model http://www.cdfreaks.com/news2.php?ID=9873

We can re-formulate this equation by the function of the price of the legal download, which results in the following equation:

consumer has to reckon with a fine f, representing the monetary amount for either paying the fine or the bail for getting out of jail6. The following equation shows the new budget line mi, under the condition that the consumer has a certain probability u of being prosecuted and paying a fine f when copying from filesharing systems:

px > (1 – #)uf In order to apply our model to reality, we have used figures from the real world as accurately as possible. In our calculations, the price of the legal download is USD 1, as this is the price which iTunes, the most successful legal music provider, is currently charging. We assume that the probability of being prosecuted for copyright violations is 0.01% when copying music from filesharing systems (u = 0.0001). It is very difficult to assess such a percentage, but taking into account that there are tens of millions of individuals using such networks – alone on KalaA, there are about 4 million users constantly online – and there have been approximately 8,000 people accused for copyright violations, we are convinced that 0.01% is a good estimate. In regards to the fine paid, individuals sued have so far paid fines of about USD 3,000, which we have taken as a starting point for our model (f = 3,000). Finally, we assume that DRM systems restrict consumers in their usage. In our model, we assume that users are only minimally restricted in their usage, as is also the case of i-Tunes where they have unlimited playing, burning on CD8, and barely limited in moving to other devices, hence we put # = 0.1. Therefore, we arrive at the following results:

mi = pxxi + (py +uf) yi slope – px / uf As mentioned already, law enforcement requires the existence of a Digital Rights Management System, which also restricts the consumer in the usage of the legal download. This restriction reduces the utility of the legal download nxi. An additional parameter, # ! [0, 1], will be defined which takes into account the degree of restriction consumers in the usage of their legal download where nxi (1 – #). We can formulate the following utility function as: U(x,y) = nxi(1 – #) xi + nyi yi slope –(1 – #)/1 Combining both quality and security issues on one hand, and legal and right restrictions on the other hand, we can formulate the following revised budget line and utility function. For the budget line we have:

1 > (1 - 0.1)0.0001*3000 => 1 > 0.27

mi = pxxi + (py +uf) yi

In that respect, the consumer would prefer the original to the copy as shown in Figure 2 with the optimal point B.

slope -px/uf 7 and the utility function can be expressed as: U(x,y) = nxi(1 – #) xi + nyi (1 – ") yi slope –(1 – # )/(1 – ") Using three scenarios, we want to show how consumer behavior might change from copying to buying and vice-versa.

3.

SCENARIO

Two additional remarks have to be made when presenting the three scenarios. First, we have assumed that law enforcement is only possible with the existence of a DRMS as mentioned above. This implies that when # > 0, the product of u f > 0 [26]. Second, when consumers might be indifferent, thus -px/uf = – (1 – # )/(1 – "), we assume that they then prefer to buy the product legally for ethical and moral reasons. Figure 2. Optimal choice scenario 1

3.1

Whether the consumer chooses the original or the copy depends on the chosen price of the original px, on the degree of law enforcement and the expected fine uf, as well as to what extend DRMS restricts consumers in their usage of the legal download #. In any case, the most influencing factor exercised by content providers is the price of the original px followed immediately by the rights restrictions faced by the consumer #.

Scenario 1

In the first scenario, we assume that there is no quality difference between the original and the copy, hence " = 0. However, there is the probability of being prosecuted when copying, therefore u, f as well as # have positive values. The consumer chooses to buy the original under the following condition: -px/uf < – (1 – # )/(1 – ") px/uf > 1 – # 6

iPure“ jail sentences are excluded in the model.

7

py, tend to py ! 0.

8

602

Only if combination of songs is different, if on same combination 10 times

3.2

0.19. With the following equation -px/uf = – (1 – # )/(1 – ") we receive:

Scenario 2

This scenario wants to assess consumer reaction when there is a quality difference between the original and the copy, with all other parameters taking the value 0. Again, the consumer acquires the original under the following condition:

-px/uf = - (1 – # ) / (1 – ") -px = -[ 0.9 / 0.9 ] 0.3 We obtain a value of approximately px = 0.3, at which the majority of consumers might switch their behavior from copying to buying or vice-versa. Surprisingly, this price is very close to an empirical study by [12], which found that the optimal price for a digital download would be around EUR 0.5 or about USD 0.4 [12] respectively. The second value calculated is how much restriction through DRMS consumers are willing to accept when downloading legally. Again, we can use the equation -px/uf = – (1 – # )/(1 – ") and the above mentioned values as a starting point, except that # will be calculated by:

-px/uf < – (1 – # )/(1 – ") therefore -px/$0 < – 1 /1 – " If we take the case of the copy not being unusable or almost unusable (" # 1), the consumer prefers the original shown in Figure 3. Point C is the slope of the utility function which is almost vertical and therefore parallel to the slope of the budget line. In other cases, the consumer prefers the copy to the original, as shown in Figure 3. Point D, the slope of the budget line, is steeper than the one of the utility function.

-px/uf = - (1 – # ) / (1 – ") [-1 / 0.3 ] 0.9 + 1 = # We obtain a value for the parameter # = -2, which is a negative value, hence not a possible solution. One explanation for this might be that, in order for consumers to prefer buying over copying it, the price currently demanded in the market should not be restricted at all. This is almost true in the case of i-Tunes, but not for the other legal music providers, which restrict the consumers in way or another most of the time. However, decreasing the price to USD 0.3, which was the calculated price at which consumers might switch from copying to buying, we arrive at a positive value of # = 0.1. This leads us to the conclusion that content providers should either demand a higher price but not restrict consumers or demand a lower price but restrict consumers.

4.

Figure 3. Optimal choice scenario 2

The aim of this paper was to present a model, by taking the view from a single consumer’s perspective, who has the option of either buying or copying digital music. In addition to the price paid for a legal download, there are other factors affecting a consumer’s decision to buy or copy online music. We have presented additional parameters such as quality and security risk associated with copying music. Furthermore, we have shown that legal and technological aspects such as law enforcement and its underlying penalties, as well as rights and usage restrictions enforced through DRMS, have all an effect onto the consumer behavior. After presenting the extended model, the paper discussed three possible scenarios. By using empirical data we have established a link between theory and practice. Our findings show that the most influencing parameters content providers can act on is the price of the original px, followed by the rights restrictions granted to the consumer #, both playing an important role in consumer behavior whether to purchase or copy digital music. We further show that they should either demand a higher price but not restrict consumers or demand a lower price but restrict consumers more. However, doing both does not seem to make the consumer buy online music.

Empirical analyses have shown that, to date, there is a high probability of getting high quality music on file-sharing systems, hence " is low [11]. Furthermore, the characteristics such as reproductability, destructibility and dividability will probably cause " to remain very low in the future. This scenario shows that the quality difference plays a minor role in the consumer’s purchasing behavior for digital music and that content providers have to focus on other parameters to make consumer change their behavior.

3.3

CONCLUSION AND DISCUSSION

Scenario 3

This scenario models a situation where all parameters have values and we analyze a consumer’s response to such a situation. As discussed earlier, content providers mainly have two parameters which they can influence directly, the price of the original px, as well as the degree of rights restriction granted to the consumer by DRMS, expressed in the model by the parameter #. In that respect, we calculate these values in order to know the critical point (or area) at which a consumer might change it’s behavior from copying to buying or vice-versa. We take as a starting point the same values as those used in the previous scenarios, such as px = 1, u = 0.0001, f = 3'000, # = 0.1, and "=

9

603

According to a study by [11], there is a almost 90% probability to get high quality music on peer-to-peer networks, thus a = 0.1.

getting a virus or spyware have been simplified. This paper also did not discussed the aggregated demand for legal and illegal digital music nor the supply side, presenting market equilibrium. Therefore, further research is required to better understand consumer behavior in respect to digital content such as music but also movies.

This paper further illustrates that a consumer’s decision to buy or copy is very complex and there might still be additional parameters which have been excluded in this paper. There are some limitations to our model, as it presents a static model instead of a dynamic one, and the model is based on some assumptions which might be perceived as too restrictive. Aspects such as network effects [23], taxation of digital products, or the risk of

[1] [2]

[3]

[4] [5]

[6]

[7] [8] [9]

[10] [11]

[12]

R. Andres, The European Software Piracy: An Empirical Application, 2002. S. Bhattacharjee, K. Lertwachara, R. Gopal and J. Marsden, No more shadow boxing with online music piracy: Strategic Business Models to Enhance Revenues, 36th Annual Hawaii International Conference on System Sciences (HICSS'03), IEEE, Big Island, Hawaii, 2002. W. Buhse, Systematisierung von Geschäftsmodellen für Online Musik unter Berücksichtigung von Marktunsicherheiten, Wirtschaftsinformatik, 43 (2001), pp. 383-391. A. E. Burke, How Effective are International Copyright Conventions in the Music Industry?, Journal of Cultural Economics, 20 (1996), pp. 51 - 66. Y. Chen and I. Png, Software Pricing and Copyright Enforcement: Private Profit vis-a-vis Social Welfare, in P. De and J. De Gross, eds., Proceedings of the Twentieth International Conference on Information Systems, ACM, Charlotte, North Carolina, USA, 1999, pp. 119-123. E. Chiang, Copyright Protection in U.S. Universities: An Overview of Copyright Piracy, Risk Attitudes towards Copyright Law, and Willingness-to-Pay, 2003. E. Chiang and D. Assane, Software copyright infringement among college students, Applied Economics, 34 (2002), pp. 157-166. S.-Y. Choi, D. Stahl and A. Whinston, The Economics of Electronic Commerce, Macmillan Technical Publishing, Indianapolis, 1997. M. Fetscherin, Evaluating consumer acceptance for protected digital content, in E. Becker, W. Buhse, D. Gtnnewig and N. Rump, eds., Digital Rights Management - Technological, Economic, Legal and Political Aspects, Springer, Berlin, 2003, pp. 342 362. M. Fetscherin and G. Knolmayer, Gegenschlag von Musik- und Filmindustrie, Neue Zürcher Zeitung 2004, pp. 59. M. Fetscherin and S. laugg, Music Piracy on Peer-toPeer Networks, IEEE International Conference on eTechnology, e-Commerce, and e-Service, IEEE Computer Society, Taipei, 2004.

[13]

[14] [15]

[16] [17] [18] [19] [20] [21]

[22]

[23] [24] [25]

604

N. Gehrke and M. Anding, Präferenzen von Konsumenten von Online-Musik - Eine empirische Erhebung, 2004. T. Hess, iTunes Music Store - Eine innovative Dienstleistung zur Durchsetzung von Property-Rights im Internet, Wirtschaftsinformatik, 45 (2003), pp. 541-546. H. Holm, Can economic theory explain piracy behavior?, Economic Analysis u Policy, 3 (2003), pp. 1082 - 1082. H. Holm, The Computer Generation's Willingness to Pay for Originals when Pirates are Present - A CV study, School of Economics and Management, Lund University, Lund, 2000, pp. 22. K.-L. Hui and I. Png, Piracy and the Legitimate Demand for Recorded Music, 2002. IFPI, IFPI Music Piracy Report, IFPI, 2002. A. Katz, A Network Effects Perspective on Software Piracy, 2003. W. J. Lee, T. U. Kim and J.-Y. Chung, User acceptance of the mobile Internet, M-Business, Athens, Greece, 2002. S. Liebowitz, Policing Pirates in the Networked Age, Cato Institute, 2002, pp. 28. M. Limayem, M. Khalifa and W. Chin, Factors motivating software piracy: a longitudinal study, International Conference on Information Systems, ACM, Charlotte, North Carolina, United States, 1999, pp. 124 - 131. T. C. Lin, M. H. Hsu, F. Y. Kuo and S. P.C., An Intention Model-based Study of Software Piracy, 32nd Annual Hawaii International Conference on System Sciences, IEEE, Maui, Hawaii, 1999. S. Poddar, Economics of Software Piracy and It's Global Impact, 2002. R. Sims, H. K. Cheng and H. Teegen, Toward a profile of student software pirates, Journal of Business Ethics, 15 (1996), pp. 839 – 851. J. Wijk, Dealing With Piracy. Intellectual Asset Management In Music And Software, Erasmus Research Institute of Management (ERIM), 2002.