1
Market-Selectors and Illegal File-Sharing in the Movie Industry
Thomas Palm 10044817
MSc Business Studies, Management and Entrepreneurship in the Creative Industries track Master Thesis
Supervised by Frederik Situmeang 1 July 2014
2
1. Introduction
As James Madison, the 4th president of the United States and one of the leading founders of the United States constitution once said: “Some degree of abuse is inseparable from the proper use
of everything” Many hundreds of years later, the truth of the his words still hold true with
content providers in the music and movie industries suffering immeasurable amounts of dollars in lost revenues due to internet-wide copyright infringement. The scope of the problem is reflected in the statutory damages a copyright infringer can be held liable for in the West; from 750 dollars to 30,000 dollars per copied work (Lemley and Reese, 2004). From the perspective of the direct facilitator of modern-day illegal downloading such as Napster or Pirate Bay who may be in possession of hundreds (if not thousands) of files on their computers, this sum can easily rise up to the millions in statutory damages.
Although previous studies have disputed the actual effects of illegal downloading on for example a firm’s financial performance, understandably for many businesses these freely downloaded copies represent lost sales and thereby a major threat to their business. However, for a better understanding and consequently a better ability to act in the face of a rising volume of illegal downloads, one needs to understand the underlying factors that drive the supply and demand of illegal downloading. In particular one needs a better understanding of these factors in terms of the P2P networks where most of these downloads take place (Becker & Clement, 2006).
According to Becker & Clement (2006), participation in illegal downloading can be divided into distinct groups along a spectrum with those who share heavily on one end and free-riders on the opposite end. However, what is most interesting amongst their finding is that the principal motivation for those heavy sharers to share despite the risk of getting caught is the
3
notion of reciprocity, in other words the expectation that others will share files too if they do, thereby creating a large network of providers in which one could tap into for free copies.
However, what are the underlying factors that fuels file-sharers to create this network of file-sharing based on reciprocity? Thus, the purpose of this article is to further the understanding the drivers of illegal downloads. More specifically, this paper embarks on this attempt by focusing on illegal downloading solely in terms of the movie industry. Furthermore this paper looks at what kind of extent the different traditional factors and film attributes that drive movie goers to go attend a movie theater influences illegal downloading behavior.
For this purpose, the paper begins with a brief literature review to explore where present-day studies stand on this topic, and more importantly to establish the key factors that drive the demand for a movie and thereby a conceptual framework that will be used for this investigation. In the following section, the conceptual framework in question is presented along with the method of research and analysis. The paper then concludes with a discussion of the results and final words on managerial implications and the direction for future research.
2. Literature Review
As various papers have noted, peer-to-peer networks (P2P) with a file sharing nature have been argued to be a threat to content providers in the creative industries, particularly those content providers in the film business (Tschmuck, 2010; Henning-Thurau et al., 2007). However, in terms of the film industry these claims have been challenged and played down by numerous studies across the years as being minor or insignificant in terms of whether pirated movies act as substitutes to the authentic options (Fetscherin, 2005; Adermon & Liang, 2010). In some cases it has even been suggested that the practice of illegal downloading actually benefits by boosting
4
sales through the word-of-mouth effect (McKenzie, 2009). However, according to Bhattacharjee (2007) this effect on sales depends significantly on the pre-existing popularity and awareness of a certain work, for example the singer of a music album. Nevertheless as many scholars have pointed out that sharing movies or even other files in general are on the rise as technological advancement take place and as an increasing number of people in the developing countries are becoming connected. Furthermore, the occurrence of the behavior of illegal downloading has been identified to be dependent on not only one, but on numerous factors, and moreover on the place in the production process or the post-production product life-cycle of the work in question finds itself in (Byers et al, 2004; Adermon & Liang, 2010).
Although the concept and definition of illegal file-sharing is rather self-explanatory, it is worth noting that the definition of illegal downloading is very country dependent, and has been inadequately explored in a coherent way amongst previous literature, perhaps for this very reason. Nevertheless, the Copyright Directive of the European Union distinguishes three aspects within the copyright law, namely: the right of communication to the public, the right to reproduction, and the right to make available to the public (Hugenholtz, 2000). Therefore for the sake of simplicity, one could define illegal file-sharing here as the unlawful and unauthorized
reproduction, distribution or display of a copyrighted work. However, apart from the legal
aspects of downloading, previous studies have pointed out the significance of the moral aspects of illegal file sharing. According to Chen et al. (2008) one could look at illegal file sharing through the perspective of Kohlberg’s Model of Cognitive Moral Development (CMD), and suggests that illegal file-sharing foremost takes place due to the fact that internet users possess an underdeveloped moral reasoning ability. This finding goes hand-in-hand with findings as
5
presented by Jackson et al. (2012) that one of the pathways to legal compliance goes through a person’s moral alignment with law enforcers.
According to Landes and Lichtman (2003), modern day copyright infringement of music and movies began around the 1970s and 1980s in flea markets where unauthorized recordings of songs were being bought and sold. As late as 1991, a police raid on a California flea market ended with the seizure of over 38,000 illegal recordings (Landes, 1991). Since then, with the development of the internet and Napster in particular, as Lemley & Reese (2004) point out the scale of copyright infringement has grown to the extent that suing actual infringers have started to become rather passé. The primary tool used these days by music and film content providers is the notion of secondary liability, namely the suing of direct facilitators such as Napster and Pirate Bay. From an economic perspective, one could interpret the situation that copyright infringement has grown to the extent that dealing with the demand-side of the equation has become too cost-inefficient. Furthermore, as already mentioned above, Lander and Lichtman (2003) have identified that one primary result of this growth of availability and accessibility of copyright material to consumers have resulted in content providers increasing prices in order to cover for damages. From an economic perspective, this price increase may act as a welfare reducing tax so to say. In other words, according to Yen (2006) the number of computer-based copyright infringers is so large that these content providers cannot find and sue every single one of them.
However, whether or not this development in the movie industry is surprising, or even negative, is doubtful. According to Dana and Spier (1999), the movie industry, particularly Hollywood, has been transitioning from a business model based on vertical separation of upstream monopolism to a model based on revenue-sharing. In other words, Hollywood movie
6
studios have been transforming the way they conduct business with video retail outlets down the value chain. For example, traditionally video outlets such as Blockbuster bought videotapes of recently released movies through a distributor for approximately 65 USD per copy and kept all future revenues from subsequent rentals of the videotape to consumers. However, the downside of such a system was that consumers were often “stocked out”, in other words demand was significantly higher than supply, which left consumers unable to rent out their preferred title.
Under the new revenue-sharing model, movie rental outlets would purchase a copy of a title for approximately 8 dollars each, a significantly lower amount, while sharing future revenues from subsequent rentals with the movie studios and distributors. This new system has allowed these movie rental outlets to increase their stock, and thus equate supply more with demand (Dana & Spier, 1999). From this development in the movie industry, one could view the current rise in illegal downloading as a continuation of the move to the revenue-sharing model itself. More specifically, that the P2P networks where the illegal downloading takes place have replaced the movie rental outlets, and by paying less to have the movie file in the first place, these downstream actors have been able to significantly increase the supply and reach out to more consumers.
According to Becker and Clement (2006), those that participate in illegal downloading can be categorized into three distinct groups according to a spectrum. On one end of the spectrum there are those who heavily share files, sharing more or less whatever they can, despite the increased risk of getting caught. And on the opposite end there are those who heavily free-ride on their peers file-sharing, in other words those who connect to these P2P networks where illegal copies are made available to just download and providing nothing in return. However, a
7
major finding by Becker and Clement (2006) on this point is that the principal motivator for those heavy sharers is the notion of reciprocity.
Nevertheless, according to Fetscherin (2005) there are a range of factors that govern the propensity of downloading a movie and these can be classified according to the nature of factor in question. Firstly, on one hand there are those factors that are more technical in nature; for example the ability to access high quality copies of movies online or the availability of the movie for download at all. And secondly, on the other hand there are the other factors that are more directly related to the consumer itself. These would be for example those factors that would form the steps of a consumer’s behavioral thinking process as depicted in the theory of central or peripheral routes of attitude change as formulated by Petty & Cacioppo (1983). For example, how much information or interest does the consumer have on the movie in question. Other factors as outlined by Fetscherin (2005) are the individual’s perception towards the risk of getting caught downloading illegally, the perceived value of the original, et cetera.
However, despite the array of factors and explanations provided by various studies on illegal downloading, there is a convergence in the literature on that aesthetic products are in most cases meant to be appreciated by consumers for the perceived intrinsic values and not the utilitarian benefits that can be derived through them, or so at least is what appears to be the dominating belief (Holbrook & Anand, 1990; Klamer, 1996). This has been particularly true for film industries, where the dominant criteria consumers use to evaluate the movie is the enjoyment and pleasure one is able to get from watching a movie and not its utilitarian attributes (Neelamegham & Jain, 1999). Furthermore, previous studies have also pointed out how aesthetic value itself may be context dependent; for example a work of aesthetics presented in a gallery can be valued differently compared if the same work would be exposed in the everyday life of
8
the public (Kirk et al., 2009). In other words, the setting in which the work is presented can have an indirect effect on the way a consumer perceives the value of a particular work to be and the way in which the consumer enjoys the work. This point builds on the notion as put forth by Iser (1978) who studied literary works of art that words written down in written literature can have a different meaning or value to when the words are spoken instead of written down. These findings suggest that in determining the aesthetic value of a work, in general one cannot just look at the work itself, but one also needs to take into consideration the setting in which they are presented.
As Basuroy et al. (2003) point out highly uncertain demand is a key characteristic of the film industry. Although uncertainty is a key characteristic of the creative industries, it is especially true within the film business as established by Vany & Walls (1999). In their own words, “no one knows anything in the movie business”. Such claim has also been supported by the likes of Orbach and Einav (2007) who have characterized the motion pictures industry as an industry with a short product lifecycle and highly uncertain demand for new releases.
However, despite the ensuing uncertainty that surrounds much of what happens in the motion pictures world, there have been numerous attempts to predict demand with a certain degree of success. For example, production costs and gross box-office revenues were found to be strongly correlated (Einav, 2007). Furthermore, a lot of the uncertainty in terms of the success of a new release tends to be revealed only after its first weekend on the screens (Einav, 2007).
In addition to this, it has also been established that there is a strong correlation between show-time demand the time of the year, or week. According to Einav (2007) who studied the average weekly attendance of a movie theatre, there were significantly higher attendance rates
9
during holiday periods, and that on average a weekend day saw 3.5 times more movie-goers than on an average weekday.
Figure 1: Seasonality in movie attendance (1985-1999)
Source: Einav (2007)
But what identifies and defines the (perceived) aesthetic value of such experience goods as offered in the film industry and hence its demand? As Gemser et al. (2008) point out, previous literature agrees that the quality and thus the actual value of a cultural product such as movies are somewhat difficult to determine prior to consumption. Therefore, it has been widely recognized that awards presented to these movies are important quality signals that helps to shape the perceived value a consumer has of a movie, and hence the intention to consume the product (in this case, go to the theatre and watch the movie).
This notion is consistent with various studies carried out amongst movie goers. For example, according to Simmons (1994) a survey reported in the Wall Street Journal reported that one third of movie-goers reportedly chose a film because of a favorable review. While reviews in general may not equally influence everyone in a direct way as each consumer is different from one another, these reviews do have an indirect effect on movie-goers one way or another. The reason for this is because while a positive review may encourage one consumer to go watch a
10
movie, if the consumer is satisfied that consumer will then influence others to watch the movie through the word-of-mouth (Reinstein & Snyder, 2005).
According to Wijnberg (1994), these movie reviews and awards can be attributed to three distinguishable selection systems, namely: market-selection, peer-selection, and expert selection. In market- selection the selectors are market-based characterized by an impersonal environment, in other words this type of selection is based on the process of natural selection. In other words, the value that arises out of this system is based on what the consumers who make up the market as a whole think or feel about the work. On the contrary, peer-selection is where the selectors and the selected are primarily the same group of people. This is where the peers of whoever produced a work assess and evaluate the value of something. A notable example would be peer-reviewed articles, where academics would review what others have written. And finally expert selection is where a small group of individuals who are not part of the group of selected is delegated a special evaluative capacity (Wijnberg, 1994). This would be for example where a panel of appointed judges would dictate the assessment and value of something.
From the perspective of signaling theory, one may argue that the role these reviews play in influencing a movie’s box office performance is substantial especially the smaller the discord between the reviews or ratings given by the different selection systems. As Kirmani (1997) points out, the reason for this positive effect on a movie’s box office is because the very reason a consumer has doubts about watching a movie in the first place is the high uncertainty on the quality and value of the movie. For example, a positive accord between different selectors as described by Wijnberg (1994) would significantly reduce the uncertainty on a movie’s quality and encourage consumers to actually watch the movie.
11
An interesting model to take into consideration here is the Elaboration Likelihood Model (ELM) as initially proposed by Petty and Cacioppo (1986). In this model, the authors outline the formation of attitudes in terms of central and peripheral routes individuals can take in forming attitudes. To quickly summarize the model, one could conceptualize that there are two kinds of consumers, those who are interested in possession of information on something, and those who are relatively less interested with little information; central-route takers, and peripheral-route takers. The central model is the phenomenon where a person’s attitude changes due to richness in information and the ensuing consideration the person makes of this information. On the contrary, the peripheral route is where attitude change happens based simply on negative or positive cues, or even a simple decision rule, about something due to the lack of information or interest.
This model on dual process theory is both interesting and relevant to this study for the reason that it may explain the importance of movie reviews or rating systems as outlined by Wijnberg (1994). For example, those consumers who aren’t especially interested in dissecting and critically evaluating a movie may simple base an attitude towards a movie on the movie’s rating. Furthermore, when individuals are relatively interested and in possession of significant amount of information about a movie, they may follow a central route and base their decision on watching a movie after diligently consideration the information available. Therefore, from this information above, one may see the importance of selectors in a movie’s box office performance; the information that comes out of a movie’s rating is relevant for both the involved and not so involved movie-goers.
However, as outlined in previous findings by Reinstein and Snyder (2005), the degree of influence a rating or review has on a movie’s box office performance is not generalizable and
12
straight-forward. According to Reinstein and Snyder (2005) film reviews were significantly more influential when it came to independent movies contrary to mainstream ones, Gemser et al. (2008) found that the effectiveness of awards handed out by the different selection systems differ significantly. More specifically in terms of mainstream movies it was found that the credibility of a reviews no matter the selection system it originated from are comparable and similar in terms of influence. This is in contrast when compared to independent movies. This is because for independent movies reviews led by expert-selectors were significantly more influential compared to reviews by other type of selectors. An example of such finding was that for example for mainstream movies, a consumer-selected MTV Movie Award was found not to be a more credible cue than winning an Academy Award as Gemser et al. (2008) point out.
Furthermore in terms of awards, it was also pointed out by Simonoff and Sparrow (2012) that some awards such as Oscar nominations did seem to increase the revenues for mainstream movies, but only if the movie hadn’t been in release for many weeks already prior to the nomination. Therefore for the purpose of this study, the aesthetic value of a movie can be defined as the overall evaluation in terms of reviews and ratings as presented by the relevant selectors.
Furthermore, according to Hsu (2006) there is a significant correlation between the genre of a movie title and the attention of critics a movie receives. The reason for this is that the ability of which a movie critic is able to convey him or herself as credible depends on the appeal for and the defensibility of the standard used in evaluating a movie product. Thus, producers that are in areas where such standards of evaluation exist are more likely to receive a disproportionate amount of attention from critics contrary to other areas. This would suggest that the particular movie genres are more susceptible to reviews and rewards, whereas other genres are relatively less affected.
13
According to Basuroy et al. (2003), these awards have a significant correlation with a movie’s box office revenue, perhaps the most crucial aspect of a movie’s product lifecycle. It was found that in the first eight weeks of a movie’s lifecycle, a movie’s financial performance was significantly correlated with both positive and negative reviews, with the impact of negative reviews diminishing over time. Furthermore, building on findings by De Silva (1998), star power and budgetary strength were identified to be significant moderators when it came to cushioning the impact of negative reviews, though their effects were significantly lower once the movie had already received positive feedback (Basuroy et al., 2003).
The notion of star power being beneficial for a movie has also found support in findings by Elberse (2007) who found a positive correlation between trading behavior on a simulated stock market setting and casting announcements of actors with varying star power. Elberse (2007) also found that on average stars are worth around $3 million in revenues.
However, this notion is somewhat contested. Although Elberse (2007) was able to find support for the hypothesis that star power does indeed boost film-level revenues, no evidence was found of star power increasing the valuation of film producers which brings into question whether star power actually creates more value than it costs. Furthermore, recent literature has not been able to challenge the claim by Vany and Walls (1999) that movies are complex products. According to their study it is the audience themselves who make a movie a hit, and there is no amount of marketing or star power that could alter the outcome. However as Elberse (2006) points out, it is the complexity and its one-off nature of movies that make hypotheses like this difficult to test and explore.
14
3. Conceptual Framework
Building on previous literature as outlined above, this paper arrives to the following research question, namely: What is the correlation between a movie’s aesthetic value as expressed in
terms of online ratings and the number of times it gets downloaded illegally? The subsequent
section presents the research design and the methodology of the study.
3.1 The Framework
Firstly, in order to arrive at a conceptual model for this study, one needs to take into account the unique feature of “experience goods” to which movies belong to. As Gemser et al. (2008) point out, the fact that movies are experience goods, it is difficult to determine a movie’s aesthetic value prior to consumption. In other words, a consumer cannot foresee how much value he or she would get out of watching a particular movie. This is arguably the primary reason why the movie industry can be described the way it can be described; short-lifecycles with high uncertainty of demand (Orbach and Einav, 2007).
As already pointed out earlier, one way for consumers to overcome this problem of gauging a movie’s quality or value prior to consumption is by looking at reviews and awards. Wijnberg (1994) mentions three distinct systems or groups of people who issue movie reviews, namely market-selectors, peer-selectors and expert-selectors. From a signaling theory perspective, one could envision that equal quality signals from the different selection systems in this case would effectively influence a consumer’s choice. For example, if there is a clear positive agreement between different selection systems, one would become more likely to watch the movie. The reason for this is that this would lower the uncertainty, and thereby greatly
15
increase a consumer’s confidence prior to consumption that watching a movie would be a good decision (Kirmani, 1997).
Of course the above notion is based on findings by numerous studies that positive reviews and ratings are favorable for a movie’s box office performance and vice versa. As Basuroy et al. (2003) points out, this finding is particularly true during the first eight weeks of a movie’s release, with the negative reviews diminishing over time. But either way, as Gemser et al. (2008) point out, for a potential movie-goer a good place to start to evaluate whether or not to watch a movie are the ratings and any award nominations it has, and furthermore what kind of information about the movie is circulating via word-of-mouth (Reinstein & Snyder, 2005). Thus one cannot discredit the role movie ratings play in a film’s box office performance. Therefore, through this we are able to arrive to and formulate the first hypothesis of the study:
H1: Positive movie ratings are favorable for a movie’s performance in terms of its box office revenue.
However, as indicated above Orbach and Einav (2007) have pointed out that the movie sector is a highly uncertain place, where nobody knows anything. This is due to the fact that movies are aesthetic and experience goods, the type of goods that are very difficult, if not impossible, to evaluate the quality of the work prior to consumption.
From the perspective of signaling theory, such uncertainty can be further exacerbated through information asymmetry. In other words, conflicting or highly diverse reviews and ratings would contribute to higher uncertainty through ambiguous signaling. In terms of Wijnberg’s (1994) selection system theory, for newly released movies the most relevant sources of ratings would in this case be market-selectors and expert-selectors; those ratings generated by experts
16
and users. The reason why awards are not very relevant for new releases is because unless the movie has already been released once at a film festival somewhere, naturally the peer-selectors would not have had the time or the opportunity yet to consider the movie for any awards. Thus from here we are able to arrive at our second hypothesis:
H2: The higher the congruence between expert and user generated movie ratings, the stronger the correlation between a movie’s overall ratings and its box office performance.
However, in terms of movie reviews one also needs to take into consideration the role of the type of film production a movie has come from as pointed out by Gemser (2008). In other words one needs to take into account whether or not the film was produced within a major film studio, or if it is an independent film produced outside of a major film studio. The reason for this varies, but the primary reason for this may be attributed to the communication sciences.
One primary attribute of a mainstream movie produced in one of the major film studios is that it is mainstream; it has a relatively large budget and thereby marketing expenditure, reaching out to many more consumers unlike many independent films which usually can only afford a few releases. For mainstream movies this would naturally lead to a relatively larger audience, and as a consequence, a higher volume of feedback and critique from consumers and experts alike.
According to research by Flanagin and Metzger (2013), such a high volume and availability of user-generated information on movie ratings allows consumers to perceive it as adequately trustworthy, and thus giving a user-generated movie rating a heavier weight in their decision making process. On the contrary, because independent movies do not usually have as high of a number of user-generated content, a potential movie goer places a heavier weight on expert critique as it may be perceived as more trustworthy in this case.
17
Furthermore, what are both interesting and relevant to this point are the differences between the characteristics of mainstream and independent films. Contrary to mainstream films, those which are produced outside major studios are targeted towards specific audiences, and distinguish themselves by, generally-speaking, the style and artistic way directors portray the movies (Gemser et al., 2007). According to theories on normative social influence, one could argue that certain types of people become influenced and conform in order to be liked and accepted by a particular group. This leads to findings by Gemser et al. (2008) on how ratings by different selection systems differ in terms of the effect they have on a movie’s performance. According to their study, it was found that ratings that came out of the market-selection system had a larger positive effect for mainstream movies. On the contrary, the opinion of expert-selectors was more relevant when it came to independent or art-house movies. This may be because independent movies are more concerned with the artistic and stylistic aspects of things which for natural reasons can only be appreciated by someone with knowledge on the subject.
In addition to this, this may also explain findings by Hsu (2006) that the relevance of a review or rating as issued by expert-selectors depends on to what extent there is an established criteria for evaluating a movie. For independent movies a criteria would be more likely to be established as there are specific stylistic and artistic features a viewer can look for. On the contrary, critiquing a mainstream movie is more of an ambiguous matter owing to the fact that stylistic and artistic factors are relatively much less evident. This would suggest, as forward by Hsu (2006), that expert feedback would be more relevant for independent movies compared to mainstream ones.
This difference in effect of expert- and user-generated movie reviews and ratings can be understood from Petty & Cacioppo’s (1983) theory on central and peripheral routes to
18
advertising effectiveness. In this study they gauged the effect the quality of an argument had on consumers who were highly involved, and not as highly involved. Their major finding was that the attitude of those individuals who were highly involved or interested in a product were only affected by information or arguments for a product that were of high quality. On the contrary, those were not as involved or interested in the product generally settled with arguments of lesser quality. Because independent movies are aimed towards niche groups who are keen on conforming and getting accepted by other members of the same group, one could understand that experts who are naturally more able to offer arguments of higher quality are more influential than consumer-generated reviews and ratings. Therefore, building on the theory as mentioned above, we arrive to the third hypothesis of this study:
H3a: Positive user-generated ratings more favorable for box office performance of mainstream movies
H3b: Positive expert-generated ratings have a bigger favorable effect on the box office performance of art-house and independent movies.
However, apart from these rather basic but essential hypotheses this study intends to test, the ultimate objective of this study is to study the correlation that may exist between these movie review and ratings, and the demand for the movie in terms of illegal download. But what is also essential for this study is the effect illegal downloading may potentially have on the financial box office performance of a film.
To start off, what is relevant to note is the idea put forth by Dana and Spier (1999) that illegal downloading of movies is not surprising in itself. As already mentioned in the literature review, according to them the movie industry, more specifically, the upstream movie studios,
19
have been integrating downstream with the theatres and movie retailers. The emergence of illegal downloading of movies via P2P networks is according to Dana and Spier (1999) only a further development down the same road. The primary significance of this theory is comes with the allusion that demand for illegally downloading a film follows a similar, if not the same, demand curve as the demand for watching a movie in the traditional sense.
In addition to this, one could also argue that when there is a smaller consensus between expert- and market-based opinions, the more frequently the movie will be illegally downloaded. The reason for this is that when there is a lack of consensus between the different ratings, the ratings become less effective as a signaling device.
Therefore one could arguably apply the previous hypotheses 1 and 3 for illegal downloading as well, namely:
H4a: There is a positive correlation between a movie’s rating and the number of times it gets downloaded illegally.
H4b: Positive user-generated ratings leads to more illegal downloads for mainstream movies.
H4c: Positive expert-generated ratings leads to a more illegal downloads for independent movies.
H4d: when there is a lack of consensus between market and expert ratings, it will increase the number of downloads
Furthermore, what is of ultimate interest to this study is how illegal downloading affects the financial performance of newly released movies, namely its box office revenues. Although
20
one can accurately claim that illegal downloading represents lost revenues, unfortunately there is no real consensus between different scholars on whether or not illegal downloading ultimately has a negative effect on a movie’s box office.
But as the adage goes: there is no such thing as bad publicity. According to a study by Sorensen & Rasmussen (2004) on the correlation between New York Times book reviews and the subsequent sales of these books, it was found that any publicity is good publicity. In other words, although positive reviews had a far larger impact on sales, even negative reviews led to increase in sales of these books. The explanation Sorensen & Rasmussen (2004) put behind this is that without the negative review of a book, consumers would not have heard about the book. Therefore in this way, a negative review can act as a marketing tool just by mentioning the books existence out there.
The relevance of this finding Sorensen & Rasmussen (2004) to this study is relevant if one considers a key characteristic of independent films for example, that such films are targeted towards specific niches and thus relies heavily on word-of-mouth between like-minded groups of consumers (Gemser et al., 2007). Therefore one could speculate that the diffusion of illegal copies of an independent movie to a broader audience would be beneficial in the sense that more people would be able to know about it.
On the contrary, mainstream movies are, as described by Gemser et al. (2007), more directed towards the general audience and thus is not especially concerned about implementing artistic or stylistic components in the film like independent movies. This suggests that the average movie-goer for mainstream movies is shallower in their desire to watch the movie; purely for entertainment purposes. Compared to this, the average viewer of an independent
21
movie may have a desire with more depth in watching a movie. A viewer of an independent movie would think more of the experience watching the movie itself, looking out for these artistic or stylistic devices as a poetry-enthusiast would do when reading a poem.
As Iser (1978) points out, a typical factor one needs to take into account when studying aesthetic works is not just the quality of the work itself, but also its surroundings or the context in which it is presented. Due to the purely entertaining nature of mainstream movies, these movies are more likely to be able to be enjoyed on your screen at home. On the contrary, because directors of independent movies work with the stylistic aspects of showing the movie itself, it is more likely that an independent movie enthusiast would be willing to pay for a movie in the theatre, partly to conform to the way it is supposed to be viewed. Therefore, this study arrives to the fifth set of hypothesis:
H5a: The illegal downloading of mainstream movies has a negative correlation with the box office performance of the movie.
H5b: The illegal downloading of independent movies has a positive correlation with the box office performance of the movie.
22
Following these hypotheses as presented above, the following conceptual framework can be formulated for this study. The framework here is based on the idea by Petty and Cacioppo (1983) that there are indeed two distinct routes to persuasion:
Figure 2: Conceptual Framework
As depicted in the figure above, there are two routes a consumer can take to watching a movie. The central route to watching a movie involves going directly to a movie theatre to watch a newly released movie. On the contrary the peripheral route of watching a movie involves downloading a movie online and then ending up viewing a movie that way. Both routes lead to
23
the box movie which leads to positive or a neutral box office revenue, and an increase in review and ratings input.
The framework above also relates to Petty and Cacciopo (1983) ELM theory and hypothesis 2 in the way that when there is low uncertainty over the quality of a movie a consumer would follow the central route to watching a movie. Compared to this, when there is incomplete or asymmetric information, a consumer would pick up on negative or positive cues to make a decision, thereby more likely leading to the taking of the peripheral route to watching a movie.
Therefore, to summarize, this study relies on three sources of data, namely: 1) the number of illegal downloads, 2) the review and ratings of a movie, and 3) the weekly box office of a movie.
3.2 Data Collection
For the purpose of this study, this paper will take a social media research approach studying the interaction between online review mechanism and illegal downloading of movies through P2P networks, and more specifically the paper takes on a netnography-related approach to research. This also entails that strands of database research methodologies will be employed as well. Therefore, the research setting of this study is the array of different popular P2P networks (torrents) as compiled through torrentz.eu, where links not only to movies from the United States can be found but also links for more oriental movies such as those out of Bollywood can also be found in significant numbers.
Information on user-generated and expert-generated reviews and ratings will be obtained from the popular movie review website IMDB, and this data will be compared to the current
24
download statistics as indicated on Torrentz.eu. This channel is chosen because it lists a compilation of downloads based on the most popular torrent sites online, and it allows a good estimate on what is being download the most.
Furthermore, weekly box office data on each movie title is obtained online from sources such as boxofficemojo.com.
3.3 Sample
For the purpose of this study, a sample consisting of newly released movies during the weeks between the months of January and March are used. This list is obtained from looking at the weekly release schedule that can be found on boxofficemojo.com. Furthermore, out of the list of movies that are released during this time period, only those movies with a release of more than 1 are included in the sample for the week. However, in the case of a week where the movie release is not that widespread, this guideline is dropped. This resulted in a sample size of 55 movie titles. According to Saunders et al. (1997) this would be an adequate number as a sample size of at least 30 would generally lead to a sampling distribution with a mean close to a normal distribution.
Additionally, since this study is focused primarily on the Western movie industry, the scope of the study is the US movie market and thus the sample of this study consists of movies that were released in the US market. The reason for this is that movies produced in the US, which have the highest market share in the global movie market, are generally speaking released in the US market first before other places. Therefore collecting data from the US market may be a good indicator of the overall picture on illegal downloads and the other variables in question.
25
Two independent variables are used in this study. The first dependent variable for this study is the number of illegal downloads as indicated by torrentz.eu relative to the other movies during the same period. This is operationalized in the form of number of leechers and the number of seeders as indicated. The seeders are in this case those that supply the illegal file in question and the leechers the number of people downloading the illegal file of the movie. A reading of the number of illegal movie download is taken every 2 days by looking at the top 3 links as indicated on torrentz.eu, and a weekly ranking of the number of downloads is then compared. This one week timeframe is used in the study in order to compare the results to the box office of a movie title which is reported every week.
Secondly, another dependent variable used in this study is box office revenue of a movie. This is measured in weekly terms and in dollar figures from the US movie market. Furthermore, the weekly fluctuation in weekly box office revenues is also measured and recorded.
3.5 Independent variables
This study employs one independent variable for its purpose. Firstly, one of the independent variables in this study is the aesthetic value of the movie which is operationalized as the consumer review and ratings that can be found on the popular website for movies IMDB. This is chosen as the operationalization as in terms of market selectors, as it is a relatively well visited platform this would provide a good representation of where the market stands in terms of how well they appreciate a certain movie.
3.6 Control Variables
The study controls for several variables, namely: the type of movie in terms of the number of releases, the origin of the movie, the star power of the movie, and the movie genre.
26
Firstly, a variable that is controlled for in this study is the type of the movie in terms of the number of releases. In other words, the movie title is controlled for depending on whether it’s a ‘blockbuster’, or in other words a major budget movie, or an ‘independent movie’ with a lower budget. This may have an effect on the results as De Silva (1998) points out; the size of a movie’s budget may have the ability to cushion the impact of negative reviews on the financial performance of movies. Furthermore, as Gemser (2008) points out, the influence market-based reviews have are contingent on whether the movie is a mainstream or an independent movie. As already indicated above this will be operationalized by looking at how many releases a movie has. The reason for this is rather straight-forward, a movie with a higher budget is able to secure more screens in theatres. A blockbuster with a larger budget would have a number of releases above 1,000 nationwide in the US. On the contrary, smaller independent movies would have a much smaller release in the region of less than 100. During the months of February and March, there were no movies that had more than 100 but less than 1,000 releases. A dummy variable
Mainstream-Independent that represents this variable is created.
Secondly, the origin of the movie in question is also controlled for. This is because there are a considerable number of movie releases that originated from abroad, most notably the movie industry in India. This may affect the outcome of the results as Dinwoodie (2001) writing on copyright law points out; the classical system of international copyright law has allowed individual nations to implement those standards in their countries tailored to their own cultural and economic norms. It is common knowledge that intellectual property is relatively less protected in developing countries. And therefore, this study takes this into account by controlling for the movie of origin, as for example, a Bollywood movie generally speaking has an audience
27
predominantly of Indian background which may make it relatively more susceptible to illegal downloading. A dummy variable Domestic US-International is created for this purpose.
A third variable controlled for in this study is star power of a movie. This is done in the the study due to the reason that it may influence the outcome of the study as indicated by the likes of Basuroy (2003) and Vany and Walls (1999). This is operationalized in terms of the success the lead actor of a movie has had in terms of their previous movie. As the famous saying by stage actress Marie Dressler goes, “you’re only as good as your last picture”, this study operationalizes star power in terms of the success an actor has had in terms of his or her last movie. For this purpose the following methodology is employed:
𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆 𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑆𝑆 = 𝑂𝑂𝑂𝑂𝑃𝑃𝑂𝑂𝑂𝑂𝑂𝑂𝑂𝑂 𝑊𝑊𝑃𝑃𝑃𝑃𝑊𝑊𝑃𝑃𝑂𝑂𝑊𝑊 𝐵𝐵𝑃𝑃𝐵𝐵 𝑂𝑂𝑂𝑂𝑂𝑂𝑂𝑂𝑂𝑂𝑃𝑃 (𝑈𝑈𝑆𝑆𝑈𝑈)100,000 × 𝑅𝑅𝑃𝑃𝑆𝑆𝑆𝑆𝑃𝑃𝑂𝑂 𝑇𝑇𝑃𝑃𝑇𝑇𝑆𝑆𝑆𝑆𝑃𝑃𝑃𝑃𝑇𝑇 (%)𝑆𝑆𝑂𝑂𝑃𝑃𝑆𝑆𝑃𝑃 ×𝐹𝐹𝑆𝑆𝑂𝑂𝑃𝑃𝐹𝐹𝑃𝑃𝑃𝑃𝑊𝑊 𝑙𝑙𝑂𝑂𝑊𝑊𝑃𝑃𝑇𝑇10,000 The following formula was obtained from an individual who conducted a study on the star powers of The Avengers1. The following metric takes into account the opening weekend box office, the movie rating as found on Rotten Tomatoes, and the number of likes an actor’s primary Facebook page has to create a value for an actor’s star power. This formula is employed on the previous movie of an actor to evaluate his or her star power today.
Following the step as described above, the star power of the actors can then organized onto a scale ranging from 1 to 5, with the designations: Very Strong (5), Strong (4), Average (3), Low (2), and None (1). This idea is based on the methodology Forbes used to construct their Star Currency system, a ranking of star power of actors back in 2008 (Forbes, 2009).
1
Throwinitoutthere (March 21, 2013). Measuring Star Power: The Avengers. Retrieved from: http://wp.me/p33Ux6-3Z
28
In addition to this, another variable that is controlled for in the study is the genre of the movie. The reason for this is that as mentioned above, certain movie categories may be more prone to be affected by movie reviews. As Hsu (2006) points out, some movie categories are prone to receive a disproportionate amount of critic attention as they may have a more defensible and credible standard of evaluation, allowing more movie critics to evaluate titles in that category. However, as different movies may be assigned multiple categories at the same time. To counter this problem, the genre of the movie is evaluated by reading its description and studying available trailers and movie clips. The optimal movie genre out of the multiple genres is then chosen and assigned to the movie.
Finally, another variable that is controlled for in this study is whether or not there is an age restriction for the movie. The reason this should be controlled for is because the demand for age restricted, and non-restricted movies are arguably different; the two types of movies are intended for different audiences. For the purpose of this study, the relevant ratings of the sample are: Unrated, PG, PG-13, and R.
3.6 Method
As mentioned above, therefore the method of research employed here is netnography. This is because in order to study such a phenomenon as market-based movie ratings, and its correlation with illegal downloading, it is necessary to study the freely expressed opinions and such of the different users who contribute to the different movie forums and rating websites. Furthermore, since illegal downloading is illegal and thus a sensitive topic to ask some individuals, it makes more sense to take a more naturalistic approach in studying this behavior: in other words,
29
manually going onto these P2P networks and counting the number of downloads that are happening by hand.
The hypotheses in this study propose several factors that may seem to have an impact on illegal downloading behavior. For this reason independent sample T-tests are carried out to test these hypotheses. In addition to this, a regression analysis is also carried out to study the correlation between the dependent and independent variables. SPSS 21.0, software dedicated to statistical computations is employed for this purpose. Furthermore, it is worth noting that the T-test relies on several assumptions. These assumptions are namely that the samples are independent, the sample data is (approximately) normally distributed, and variances are equal between the different groups.
In terms of the normal distribution of the sample data, as already mentioned above a sample size of 56 is used. According to Saunders et al. (1997) a sample size of more than 30 is generally speaking adequate to attain a near normal distribution. Additionally, in terms of equal variances, Levene’s test is carried out to assess the degree of homogeneity between the different sample groups.
Additionally, missing values that arose during the data collecting process were dealt with using the hot deck imputation method as outlined by Myers (2011) as the level of missing values was fairly low at less than 5% of the total data.
4. Results
The results are presented in this section. Firstly, the descriptive statistics related to the sample are presented and described. Additionally, then the correlations are presented. Thirdly, the results of the regression analyses that were conducted are discussed.
30 4.1 Descriptive Statistics
Table 1: Descriptives
N=55
Variables Frequency In %
Type of Movie Mainstream 22 40
Independent 33 60
Country of Origin Domestic US 44 80
International 11 20 Genre Action 9 16.4 Adventure 2 3.6 Animation 2 3.6 Biography 1 1.8 Comedy 15 49.1 Crime 2 3.6 Drama 13 23.6 Fantasy 1 1.8 Horror 4 7.3 Romance 4 7.3 Sci-Fi 2 3.6 Thriller 3 5.5 War 1 1.8 Rating Unrated 4 7.27 PG 6 10.91 PG-13 26 47.27 R 19 34.54
Star Power Very Strong (5) 11 20
Strong (4) 15 27.27
Average (3) 13 23.64
Weak (2) 3 5.45
31
33 4.2 Correlations
Table 2 above shows the standard deviation, mean values, and correlations of the data as generated using SPSS. As shown in the table above, the mean of the total number of downloads that occurred, and the box office revenues incurred was contingent on the week since the movie was released. What is interesting to note, however, is how the mean value of box office revenues decreased as weeks passed since the first week of a movie’s release, confirming findings by previous studies that time and box office revenues of a movie are generally speaking negatively correlated. The mean market rating was 6.54 with a standard deviation of 1.89 (minimum 1.3; maximum 8.8). The mean expert rating was 5.66 with a standard deviation of 1.43 (minimum 8.30, maximum 2.90). The mean difference between market and expert rating of movies was -0.63 with a standard deviation of 1.4 (minimum -3.50; maximum 6).
According to the first table, in overall terms market ratings did indeed have a significant positive correlation with box office revenues. However, this correlation was only significant for weeks 5, 6 and 7 (resp. r = 0.295, p ˂ 0.05; r = 0.307, p ˂ 0.05; and r = 0.311, p ˂ 0.05). There does not seem to be a significant correlation in the early stages of a movie’s product life cycle. In turn, according to the table expert ratings did not have any significant correlations with the box office revenues of a movie nor its illegal downloads. However, a closer look as seen on the second and third tables show that although there is no significant correlation with independent movies, there is indeed a significant correlation between market and expert ratings, and the box office for mainstream movies. Moreover, any differences between market ratings and expert ratings also did not seem to have any significant correlations with box office revenues and illegal downloading, except for in the week 5 when it seemed to have a positive correlation with illegal
34
downloading (r = 0.384, p ˂ 0.01). However, in macro terms, there seems to be no significant correlations no matter independent or mainstream.
Furthermore, the control variables mainstream-independent and movie genre had significant negative and positive correlations respectively with the box office revenues of a movie. In addition to this, with the exception of weeks 5 and 7 mainstream-independent also seemed to be negatively correlated with illegal downloads, in other words main stream movies were more likely to be downloaded than independent ones.
For hypothesis 4, firstly it was hypothesized that positive movie ratings had a positive correlation with the number of times a movie gets downloaded. According to results above, no significant correlation between movies’ rating and its propensity to get downloaded was found. However, a closer look reveals that there did seem to be a significant negative correlation between market-based user ratings and downloads for mainstream movies. Furthermore, no significant correlations were found to exist between the difference between expert- and market-ratings and downloads for neither mainstream nor independent movies.
Furthermore, the control variable domestic-international seemed to have a negative correlation with the box office revenues of a movie through weeks 1-3 (respectively r = -0.299, p ˂ 0.05; r = -0.272, p ˂ 0.05; and r = -0.268, p ˂ 0.05). In other words, movies made abroad in foreign studios made less money in box office than those created by an U.S. studio. This is interesting if one considers how a movie from abroad is generally screened in theatres for less than 3 weeks. Furthermore, domestic-international seems to have a significant positive correlation with expert ratings, suggesting that movies from abroad were more likely to receive favorable expert reviews.
35
Lastly, in hypothesis 5 it was hypothesized that there would be a negative correlation between the number of downloads of mainstream movies and its box office, while there would be a positive correlation between the downloads of independent movies and its box office. However, no significant correlations were found between the variables for neither type of movies.
4.3 Analysis
In this section, the different hypotheses that were mentioned in the conceptual framework are presented again, and tested in relation to the results of regression analyses that were conducted using SPSS. Two separate models are used in the regression analyses for the reason that the dependent variable in the first one is box office revenues, and in the second the number of downloads of a movie. Furthermore, the two models are also sub-divided based on mainstream and independent.
Model 1
In this model, the dependent variable is box office revenues of a movie. For the purpose of this study, a bivariate regression analysis is used to study the empirical relationship between box office revenues and movie ratings.
The independent variables expert rating and market rating had p-values of 0.227 and 0.862 respectively for mainstream movies, and for independent movies 0.664 and 0.888 respectively, and therefore they are not significant predictors of box office revenues. Hypothesis 1 is thus rejected. Moreover, because the null hypothesis for 1 is not rejected, hypothesis 2 can be rejected as well as there is no correlation to make stronger. The p-value for this congruence is 0.989 for independent movies and 0.862 for mainstream movies, not a significant predictor either of box office revenues.
36
The p-value of market ratings for mainstream movies is 0.862, while the p-value of expert ratings for independent movies is 0.664, thus not significant predictors of box office revenues of the respective movie types. Therefore hypothesis 3a and 3b can be rejected.
A significant control variable on the 0.05 level is Movie Genre in mainstream movies. Although it is slightly above the 0.05 level, one can state that the variable has a rather medium presumption against the neutral hypothesis.
37
Model 2
In this model, the dependent variable is the illegal downloads of a movie. For the purpose of this study, a bivariate regression analysis is also used to study the empirical relationship between a movie’s downloads and its ratings.
In hypothesis 4, it was hypothesized that positive movie ratings are positively correlated with a movie’s downloads. However movie ratings were not found to be significant predictors of movie downloads in either for mainstream and independent movies. In addition to this, the difference in market and expert ratings, with a p-value of 0.856 and 0.914 also do not seem to be significant predictors of illegal downloads of a movie. Therefore hypothesis 4 and its sub-hypotheses can be rejected.
However, in both cases the inclusion of the variables market and expert ratings did improve the R-squared of both models. For independent movies, the R-square improved from the control-model value of 0.069 to a value of 0.235, while for mainstream movies the R-square improved from the control value of 0.335 to 0.429. This shows that the inclusion of these variables do indeed improve to explain variations in the variable downloads.
In terms of the control variables, it seems that movie genre is a significant predictor of
downloads for mainstream movies at the 0.05 level. In addition to this star power and age restriction seem to be weak predictors of downloads.
Furthermore, in hypothesis 5 it was hypothesized on the effect of downloading on the box office of movies, an important issue in the film industry. However, as shown in table 5 no significant correlation was found between the two variables; a p-value of 0.809 was found while
38
regressing box office revenues against movie downloads. Thus support for hypothesis 5 is not found.
39
Table 5: Regression Movie Downloads and Box Office
5. Discussion
In this section, the results from the analyses above are discussed and evaluated. Following this, conclusions are presented and future recommendations are provided. The paper ends with a brief description on potential limitations that this research may have faced.
5.1 Discussion and evaluation
In this paper, different variables that may or may not affect illegal downloading in the movie industry, and its ultimate effect on movie box offices was studied. In the hypotheses presented in the conceptual framework, it was argued that movie ratings as provided by market and expert sources are significant predictors of box office revenues, and also illegal downloading. Furthermore, it was argued that the number of illegal downloads is negatively correlated to the box office revenues of a movie.
40
Rather surprisingly, the results of the analyses did not find any support for the hypotheses that the conceptual framework came forward with. Firstly, hypothesis 1 argued that box office revenues are positively correlated with movie ratings. However, although positive correlations were found between box office revenues and movie ratings, both in terms of market and expert ratings, for mainstream movies, after performing regression analysis movie ratings were not found to be significant predictors of a movie’s box office revenues. The reason for this may because a movie could still have good ratings but have relatively smaller revenues by only having an appeal for a small group of people (and also be viewed favorably by this small group of people). Furthermore, a major mainstream movie with large revenues can still have the same ratings as a relatively smaller mainstream movie with smaller revenues, and thus it is to some extent understandable that movie ratings did not seem to be a significant predictor of box office revenues.
In hypothesis 2 it was argued that the higher the congruence between the expert and market ratings of a movie, the stronger the correlation between movie ratings and box office revenues. Therefore, there were no correlations in significant terms to strengthen or weaken. However, a significant correlation was not found between movie ratings and box office overall. Furthermore, the regression analysis has shown that this difference within ratings is not a significant predictor of box office revenues.
Hypothesis 3 argued that positive market ratings would be more favorable in terms of box office revenues for mainstream movies, while expert ratings would be more favorable for independent movies. While a significant correlation was found between mainstream movies and market-based user ratings, no significant correlation was found between independent movies and expert movies. However, according to the regression analyses that were performed, it was found
41
that both expert and market ratings did not seem to be significant predictors of box office revenues. Therefore both hypothesis 3a and 3b may be rejected. The reason for this may be the same as the reasons that were given for the rejection of hypothesis 1. The lack of correlation for independent movies for example may be explained by the fact that independent movies are generally enjoyed by a limited small group of people, and while this group of people who watch the movie may rate the movie highly, they do not necessarily lead to particularly high box office revenues. For mainstream movies, it is understandable that with more screens, a movie with higher ratings generally has higher box office revenues. However, these variables were nonetheless not found to be significant predictors of box office revenues.
Furthermore, in hypothesis 4 it was argued that positive movie ratings would increase the illegal downloads of movies, and furthermore more specifically in hypotheses 4b and 4d it was argued that a higher disagreement between expert and market ratings of a movie would lead to increased downloads. However, according to the regression analyses carried out, movie ratings and any differences within these ratings were not found to be significant predictors of movie downloads. The reason for this may be because when a consumer downloads a movie, they are not particularly concerned with the aesthetic value of the movie as there is no cost incurred as a result of the download. When there is no cost to an action, one becomes less concerned about the value.
Finally in hypothesis 5 it was argued that movie downloads are negatively correlated with box office revenues. This hypothesis was brought up to test the long standing claim by the movie industry that illegal downloads does indeed hurt the industry through lost revenues. However, no significant correlation was found between the two variables, and furthermore, a regression
42
analysis did not find the illegal downloads of a movie to be a significant predictor of box office revenues. Thus hypothesis 5 may be rejected.
5.2Limitations and Recommendations for Future Research
It is essential to list out some limitations that this research may have faced in order to properly interpret the results of this research. Furthermore, such identification of limitations may help to point future research towards the right direction.
One limitation that this study faced is the limitation of only being cross-sectional. In other words, although the objective of this paper was to study illegal downloading in the film industry, it only took a small sample over the course of 2 months. This is a limitation worth mentioning as already mentioned above in the literature review, movie attendance and box office revenues strongly vary depending on the time of the year, season and other variables. Thus future research may have to look in terms of one year in order to account for the seasonal differences.
Secondly, another significant weakness worth mentioning is that the number of downloads as recorded in the data of this study is not completely accurate. Although the most popular torrent site was used in order to study the behavior of illegal downloads, there is the possibility that downloading behavior may or may not have been different on other smaller file sharing websites. Furthermore, during the course of the research it was also found out that not all illegal downloading happens over file sharing sites. In some cases, this happens through illegal streaming of the movie. A strong recommendation for future research may be to look into illegal movie streaming as well as illegal movie downloading to better understand this phenomenon.
Lastly, a better way of coding for a movie genre should be come up with in any future attempt to study movies. The reason for this is because in most cases movies can be classified as
43
different genres by film review sources. In this study, these films were categorized in the genres that were deemed best fit by the researcher, which have led to a certain degree of bias. In order to accommodate the complexity of this variable, future research should look into taking into account multiple genres of movies.
However, in terms of future research there is another recommendation worth mentioning, which is that future studies should look into the role of marketing film studios put into movies and box office and illegal downloads. The reason for this is that a consumer who downloads movies illegally may download a movie based on the interest the person has for the movie, not necessarily the aesthetic value of the movie based on its rating. With more intense marketing, interest for a movie can be encouraged, and thus box office and downloads may also go up. This is a recommendation for future research worth following.
44
References
Adermon, A. & Liang, C. (2010). Piracy, Music, and Movies: A Natural Experiment. Uppsala University Department of Economics, Uppsala.
Basuroy, S., Chatterjee, S. & Ravid, S. (2003). How critical are critical reviews? The box office
effects of film critics, star power and budgets. Journal of Marketing , 67(1) pp. 103-117
Becker, J. & Clement, M. (2006). Dynamics of Illegal Participation in Peer-to-Peer
Networks-Why Do People Illegally Share Media Files? Journal of Media Economics, 19(1), pp. 7-32
Bhattacharjee, S., Gopal, Ram, D., Lertwachara, K., Marsden, James R. & Telang, R. (2007).
The effect of Digital Sharing Technologies on Music Markets: A Survival Analysis of Albums on Ranking Charts. Management Science.
Byers, S., Cranor, L. F., Cronin, E., Korman D. & McDaniel, P. (2004). An analysis of security
vulnerabilities in the movie production and distribution process. Telecommunications Policy,
28(7) pp. 619-644
Chen, Y., Shang, R. & Lin, A. (2008). The intention to download music files in a P2P
environment: Consumption value, fashion, and ethical decision perspectives. Electronic
Commerce Research and Applications, 7(4), pp. 411-422
Dana, J. D. & Spier, K. E. (1999). Revenue Sharing, Demand Uncertainty, and Vertical Control
of Competing Firms. Journal of Industrial Economics
De Silva, I. (1998). Consumer Selection of Motion Pictures in The Motion Picture
Mega-Industry. Barry R. Litman, ed. Needham Heights, MA: Allyn Bacon, pp. 144-71
Dinwoodie, G. B. (2001). The Development and Incorporation of International Norms in the
Formation of Copyright Law. Ohio State Law Journal. 62(1)
Einav, L. (2007). Seasonality in the U.S. Motion Picture Industry. Rand Journal of Economics, 38(1)
Fetscherin, M. (2005). Movie piracy on peer-to-peer networks – the case of KaZaa. Telematics and Informatics, 22(1) pp. 57-70
Flanagin, A. J. & Metzger, M. J. (2013). Trusting expert- versus user-generated ratings online:
The role of information volume, valence and consumer characteristics. Computers in Human
