1
Consumer File Sharing of Motion Pictures:
Consequences and Determinants

Thorsten Hennig-Thurau*
Department of Marketing and Media Research
Bauhaus-University of Weimar
Bauhausstrasse 11, 99423 Weimar, Germany
Phone +49 (0) 3643 58 3822
Fax +49 (0) 3643 58 3791
Email: tht@medien.uni-weimar.de

and

Victor Henning
Department of Marketing and Media Research
Bauhaus-University of Weimar
Bauhausstrasse 11, 99423 Weimar, Germany
Phone +49 (0) 3643 58 3736
Fax +49 (0) 3643 58 3791
Email: victor.henning@medien.uni-weimar.de



*corresponding author

Acknowledgement: The authors thank Martin Spann for his helpful comments on the
modeling parts of this paper.


This paper has been awarded
BEST PAPER OF THE E-COMMERCE AND TECHNOLOGY TRACK
of the 2005 Summer AMA Educators’ Conference and was also selected as the
OVERALL BEST PAPER
of the conference.


Accepted for presentation at
2005 Summer AMA Educators’ Conference,
E-Commerce and Technology Track

2
Consumer File Sharing of Motion Pictures:
Consequences and Determinants

Abstract
Consumer file sharing of motion pictures is considered a major threat to the movie industry.
While industry advocates postulate a cannibalistic effect on traditional forms of movie
consumption such as movie theater visits and DVD purchases, some researchers have argued
against such an effect. In this paper, we test a structural model of consumer file sharing
consequences and determinants and are the first to provide sound empirical evidence of the
impact that consumer downloading of motion pictures via peer-to-peer networks has on industry
revenues. Using two samples of 450 and 547 online survey respondents, we apply partial least
squares structural equation modeling and find no support for the cannibalization hypotheses.
Moreover, our findings help to understand the phenomenon of downloading movies by
identifying five significant drivers of downloading behavior.

1
Consumer File Sharing of Motion Pictures:
Consequences and Determinants
THE PHENOMENON OF CONSUMER FILE SHARING
The current crisis in the music industry has been extensively covered by the media.
Worldwide record sales have been plunging with an annual growth rate of -6.4% every year since
1999, with the figures for 2003 signifying another 7.6% decrease. Overall, the value of the
worldwide music market dropped from $38.7 billion in 1999 to $29.3 billion in 2003 (IFPI 2003;
2004a). The music industry has declared online piracy, i.e. consumer file sharing, to be the main
cause of the problem, as the sales slide has largely coincided with the growing popularity of peer-
to-peer networks which allow consumers to download music and other content from other
consumers without having to pay for it. In fact, the loss of CD sales has been shown to correlate
with an increase in CD-R sales which are needed to store the downloaded material (FFA 2003;
IFPI 2004b).
Today, consumer file sharing also covers the exchange of video content between consumers
without paying fees to the content copyright owners. As for video files, the estimated number of
150,000 downloaded movies per day in 1999 quadrupled to 600,000 downloaded movies per day
in 2002 (Reuters 2000; Websense 2003). According to the Motion Picture Association of
America (MPAA), around 130,000 movies per day are traded through peer-to-peer networks in
the United States alone (MPAA 2004c). The peer-to-peer monitoring company Sandvine reported
that 65% of all Internet traffic in the US and 70-80% of all Internet traffic in Europe results from
the swapping of music and movies (Forbes 2004). Although Hollywood seems not to show any
signs of economic crisis, with worldwide box office revenues in 2003 higher than ever at
USD 20.34 billion (MPAA 2004a) and worldwide home video revenues growing thanks to the
rise in DVD consumption (International Video Federation 2003), the MPAA claims that “illegal

2
movie trafficking represents the greatest threat to the economic basis of moviemaking in its 110-
year history” (MPAA 2004b). However, this perspective has been criticized by researchers who
suggest that neither the music nor the movie industry suffer economically but in fact benefit from
peer-to-peer networks for a variety of reasons. As a result, the impact of file sharing on creative
industries has become the subject of controversial discussion over the last few years. However,
up until now, no peer-reviewed study has been published on the subject of movie file sharing.
In this paper, we address two related research questions. First, we investigate whether the
revenue streams of the motion picture industry generated through traditional channels such as
theatrical showings, DVD rentals, and DVD sales are affected by the phenomenon of consumer
file sharing. Second, we analyze the determinants of the consumers’ downloading activity. We
aim to spur on the discussion of motion-picture downloading by testing a structural model of
downloading consequences and determinants using partial least squares against data from two
studies with a total of 837 responses. Results and implications are discussed.
A MODEL OF MOTION PICTURE DOWNLOADING CONSEQUENCES AND
DETERMINANTS
Figure 1 presents a structural model that contains two groups of hypotheses. While the first
group addresses the impact of motion-picture downloading on traditional movie consumption
channels, namely, movie theater attendances, DVD rentals, and DVD purchases, the second
group focuses on the determinants of downloading movies. For the latter, we distinguish between
three fundamental drivers of downloading, i.e. consumer characteristics, movie characteristics,
and infrastructure characteristics. With regard to downloading behavior as the core concept of
this research, we differentiate between the number of movies a person has downloaded and the
number of downloaded movies a person has watched. Although both concepts are closely related,
with downloading serving as an influencer of watching downloaded movies (see Figure 1), they

3
are conceptually distinct as (a) movie downloaders do not necessarily watch every motion picture
they have downloaded and (b) non-downloaders are able to watch movies that have been
downloaded and handed to them by friends or colleagues.
------------------------------------------------------------------------
Insert Figure 1 approx. here
------------------------------------------------------------------------

Consequences of Motion-Picture Downloading
Several scholars have addressed the issue of consumer file sharing and its impact on the
entertainment industry. Although all existing work focuses on music file sharing, we draw on this
literature to understand the impact of motion-picture downloading on traditional modes of movie
consumption because of the similarities between music and motion pictures (i.e. hedonic
experience goods, expensive to produce and market, can be digitalized and copied at zero
marginal cost, star-driven, few successful products subsidize a larger number of failures). As
such, some authors explicitly claim that their music file sharing models are equally applicable for
motion pictures and other digital experience goods (Gopal, Bhattacharjee, and Sanders 2004).
The idea that the downloading of motion pictures has a negative impact on other kinds of
movie consumption has been opined mostly by industry representatives based on industry studies
which use a direct-question approach (i.e. consumers are asked directly about the impact of
downloading on traditional forms of movie consumption). Specifically, the German Federal Film
Board (FFA) released a study on movie piracy based upon a random sample of 10,000 German
consumers, with 3.2 percent of respondents admitting to having downloaded movies already
(implying a sample size of n = 320 downloaders) (FFA 2003). Respondents were asked directly
how movie downloading or copying movies with a CD/DVD-burner had influenced their
consumption of motion pictures through other channels (no distinction was made between the

4
effects of downloading and copying). 46 percent of the respondents indicated that they had
reduced their number of movie theater visits, while 7.1 percent said that they go to the movies
more often. Similarly, 51.4 percent of downloaders said they rent fewer DVDs, and 43.5 percent
replied that they buy DVDs less often.
Similarly, in a report by the French Centre National de la Cinématographie (CNC) based on
a sample of 3,086 French Internet surfers aged 15 and above, 19 percent (n = 586) had already
downloaded movies at a rate of 132 movies per year (CNC 2004, p. 6). As with the FFA study,
respondents were asked directly if their movie consumption behavior had changed. In this study,
21 percent answered they had reduced their movie theater visits, and 38 percent indicated they
had cut back on DVD rentals, 35 percent on DVD purchases.
Further support for the negative impact of movie downloading on other kinds of movie
consumption is provided by an eight-country study by the MPAA (2004d). According to the
MPAA, “400 respondents were initially recruited from each country”, and “the sample was
augmented in several countries in order to provide a minimum sample of 100 movie downloaders
per country” (MPAA 2004d, p. 3), resulting in a total sample size of 3,600 Internet users with a
special emphasis on broadband users. The authors report that “about one in four Internet users
(24%) have downloaded a movie” (MPAA 2004d: 1) and that 26 percent of downloaders report
that they purchase movies “much less” or “a little less” often than in the past. Global results are,
however, limited to the impact of movie downloading on DVD/video purchases and the results
(an unweighted mean score) are biased by the outlier Korea (excluding Korea lowers the
unweighted mean from 26 to only 14 percent).
Based upon the empirical results of the aforementioned studies and industry knowledge, we
would expect to find a negative impact of motion picture downloading on traditional kinds of
movie consumption, such as visits to movie theaters, DVD rentals, and DVD sales, as expressed

5
in hypotheses 1 to 3. These hypotheses represent the main argument of the motion picture
industry against consumer file sharing, i.e. that the consumption of illegally downloaded movies
substitutes legal movie viewing and thus threatens the industry’s foundation (MPAA 2004c). It
should be stressed that all of the studies cited here are more or less limited in their
methodological approaches, founding their results on a direct-question approach only and lacking
transparency. Although none of the studies cited explicitly differentiate between the impact of
downloading movies and watching downloaded movies, we interpret the findings in a broader
sense and expect the substitution effect to be true for both kinds of consumer behavior.
H1: The more movies a person downloads and the more downloaded movies this person
watches, the less he or she will visit movie theaters.
H2: The more movies a person downloads and the more downloaded movies this person
watches, the less he or she will rent DVDs.
H3: The more movies a person downloads and the more downloaded movies this person
watched, the less he or she will purchase DVDs.

The aforementioned studies conflict with a number of theoretical and empirical articles
which argue against the existence of a substitution effect of entertainment product downloading
on traditional consumption of entertainment products or even suggest a positive impact. Although
none of these studies directly address the consumer file sharing of motion pictures, their
arguments can be easily transferred into the context of motion pictures.
Gopal, Bhattacharjee, and Sanders (2004) present a model of online music sharing
economics and derive implications for consumer surplus and producer profits. Following the train
of thought that consumer file sharing is a form of “sampling” for experience goods, they
conclude that consumer-to-consumer networks over the Internet lower the total cost of the
evaluation and acquisition of experience goods such as music and movies, resulting in increased

6
purchases and industry profits. Although the authors do not distinguish between the download
and the consumption of experience goods, their argument clearly focuses on the latter concept, as
it is the consumer’s actual consumption of a song or movie, not the technical process of
downloading itself, which reduces his or her uncertainty.
Using a different argument, Levine and Boldrin (2001; 2002) model competition with sunk
costs under the assumption that it takes a small amount of time to reproduce copies of the product
and that the elasticity of demand for the product is greater than one. They argue that in such a
setting, decreasing costs of reproduction through file sharing can make it easier, not harder, for
the producer to recoup his cost – and as the rate of reproduction increases, the competitive rents
increase as well. This conclusion is based on the concept of indirect appropriability which
assumes that the copyability of a digital product increases its value through a potential increase of
the market price. Consequently, the authors argue for the legalization of file sharing, as this
would increase the amount of entertainment products produced and distributed. Like Gopal,
Bhattacharjee, and Sanders (2004), they also do not provide empirical findings to substantiate
their conclusions. However, unlike Gopal et al. their argument applies to the actual downloading
process rather than the consumer watching downloaded material.
Finally, empirical support for the non-existence of a negative impact of downloading on
traditional distribution channels is presented by Oberholzer and Strumpf (2004). Over the course
of four months, Oberholzer and Strumpf (2004) monitored 1.75 million file downloads on
Internet consumer-to-consumer networks and then matched the downloads to U.S. album sales
data. Empirical analysis yielded the result that music file sharing has no statistically significant
impact on album sales. Their results have been criticized by industry representatives as being
skewed as the authors had collected all empirical data in the fourth quarter of the year 2002,
which is traditionally the busiest sales period with over 40% of the year’s total sales (IFPI

7
2004c). As in the case of Levine and Boldrin, the focus of this study is on downloading behavior
rather than on the consumption of the downloaded material.
Based on the arguments and/or findings that propose either no negative or a positive
association between downloading of entertainment products over the Internet and traditional
distribution channels, we follow the recommendation of Armstrong, Brodie, and Parsons (2001;
see also Sawyer and Peter 1983) and present competing hypotheses, proposing a non-negative
impact of downloading movies and watching downloaded movies on traditional forms of movie
consumption:
H1Alt: The number of movies downloaded and downloaded movies a person watches has no
or even a positive impact on how often that person visits a movie theatre.
H2Alt: The number of movies downloaded and downloaded movies a person watches has no
or even a positive impact on how often that person rents a DVD.
H3Alt: The number of movies downloaded and downloaded movies a person watches has no
or even a positive impact on how often that person purchases a DVD.

Determinants of Motion-Picture Downloading
In addition to learning about the potential impact of movie downloading on outcome
constructs, a second objective of this paper is to shed more light on the factors that determine the
intensity with which consumers download movies from peer-to-peer networks on the Internet.
Research on this topic is rare; more specifically, we are not aware of a single academic study that
addresses this question. We distinguish between three groups of factors that can be expected to
drive consumer movie downloading, namely consumer characteristics, infrastructure
characteristics, and content characteristics.
Consumer characteristics. Knowledge levels of technical issues related to the downloading
of motion pictures through computer networks differ considerably among potential downloaders

8
and can be expected to influence the consumers’ downloading intensity. This notion is based on
the task-technology fit model (Goodhue and Thompson 1995) and the concept of user
competence (Marcolin et al. 2000). The former model proposes that an individual’s performance
using information systems is determined by the fit between task (e.g. downloading motion
pictures), technology (e.g. the personal computer, the Internet, the file sharing client), and
individual characteristics (e.g. consumer knowledge). Marcolin et al. extend this model by the
concept of user competence which is considered a mediator between individual characteristics
and information systems usage. User competence is viewed as the capability of an individual to
translate knowledge into advantage (e.g. free access to entertainment goods such as motion
pictures).
Specifically, we consider three categories of consumer knowledge as relevant: (a) consumer
knowledge that refers to the use of personal computers, as such knowledge is a precondition for
installing software that is needed to download motion pictures from the Internet; (b) consumer
knowledge about the Internet and how to connect to the Internet, as downloading requires the
consumer to be online, and (c) consumer knowledge of file sharing itself, as downloading motion
pictures from peer-to-peer networks requires the consumer to grasp the possibilities of file
sharing, to get access to file sharing clients, and to configure and use them properly to download
files. For all three categories of consumer knowledge, we expect a positive impact on movie
downloading. As knowledge refers to the technical process of downloading, no direct impact is
proposed for watching downloaded motion pictures. In other words:
H4: The better a consumer’s knowledge about (a) computers, (b) the Internet, and (c)
consumer file sharing, the more motion pictures he or she will download.
The availability of motion pictures on peer-to-peer networks differs between movie genres
and varies with other movie characteristics (e.g. country of origin, star participation).

9
Specifically, in his empirical investigation of files available through the Kazaa network, Henning
(2004) demonstrates that high-budget blockbusters are available more often on file sharing
networks than independent movies, and rankings of the most frequently downloaded movies
provided by the Internet monitoring company BayTSP almost exclusively list major studio
blockbusters (BayTSP 2004). As a consequence, consumers who prefer blockbuster-type motion
pictures will find that their preferred movies are more readily available on peer-to-peer networks
than fans of independent movies will, and therefore download movies more often. Again, as this
effect is related to the technical process of downloading, we do not expect a direct impact for
watching downloaded motion pictures. We offer our next hypothesis:
H5: The higher the perceived availability of movies in peer-to-peer networks, the higher the
likelihood that consumers will download motion pictures.

Infrastructure characteristics. In addition to consumer-related factors, downloading
intensity can also be expected to be impacted by the consumer’s technical infrastructure. A slow
Internet connection represents a bottleneck to downloading activity (Hess, Anding, and Schreiber
2002), with the technical infrastructure limiting the extent to which a consumer can engage in
motion picture downloading. A faster “broadband” Internet connection increases the consumer’s
discretionary scope for deciding how many motion pictures he or she wants to download. We
would expect consumers to make use of this discretionary scope, resulting in an increase in
downloading intensity. This leads to our next hypothesis:
H6: The faster the consumer’s Internet connection, the more movies he or she will
download.
The individual consumer’s ability to watch downloaded movies on his TV/DVD set (e.g.
via a video connection cable or a burned CD-R or DVD-R) will also influence the pleasure he or
she might derive from downloading movies and from watching them. Under these circumstances,

10
the consumer’s expected enjoyment of the downloaded file will be greater as a result of the better
“cinematic” experience he or she may expect in comparison to viewing on a PC screen.
Moreover, watching movies on TV can be expected to be a more socially attractive option than
watching it on the PC which is typically located at a work desk with seating for one person only.
Consequently, we propose:
H7: If a consumer has the option to watch movies on his or her TV, he or she will
download more movies and will watch more downloaded movies.

Content characteristics. In the age of home theater systems, picture and sound quality has
become a critical issue for consumers when buying or renting a DVD as well as when going to
the movies (Business Week 1997). That said, in the case of downloaded movies, quality often
varies considerably depending on the source the material has been taken from (e.g. digital video
camera, DVD, or movie reel) and the respective compression rate and/or file size. Given that
consumers value the quality of a movie presentation, the consumer’s perception of picture and
audio quality as experienced from previously downloaded files is expected to influence the
attractiveness of downloading movies and watching downloaded movies. In other words:
H8: The better the perceived image and sound quality of downloaded movies, the more
movies he or she will download and the more downloaded movies he or she will watch.

AN EMPIRICAL TEST OF THE CONCEPTUAL FILE SHARING MODEL
Methodology
Data collection. To test our hypotheses, we carried out two empirical studies in the summer
of 2004, collecting consumer data through online questionnaires. In the first study, the potential
outcomes of consumers’ downloading behavior were addressed, while the second study focused
on the determinants-side of our conceptual model. Downloading and watching of downloaded

11
movies as core concepts of the model were measured in both studies. Both questionnaires were
distributed using a snowball sampling technique. Specifically, the web address of the
questionnaires were e-mailed to all students of a small-sized German university. Recipients were
asked to fill out the questionnaire and to forward the web address to other persons. In addition,
the links were posted in a number of Internet forums discussing movies and file sharing.
Sample characteristics. In study 1, 415 respondents participated, 232 of which were
downloading movies through peer-to-peer networks. Respondents were on average 24 years of
age, with a standard deviation of 8.2 years, and 85.8% were male. Countrywise, 60.8% were
residing in Germany, 8.0% in Switzerland and 5.0% in the US, with the rest divided mostly
among a number of Western European countries. In study 2, 547 respondents filled out the
questionnaire, 378 of which were downloaders. The demographic pattern was similar to study 1,
as the average age of respondents was also 24 years of age, with a standard deviation of 6.0 years,
and 79.8% were male. 60.2% resided in Germany, 10.9% in Switzerland, 4.2% in the UK, and
the rest mostly in the US and Western European countries. In both surveys, the majority of
respondents were students, followed by employees, pupils/apprentices and self-employed or
unemployed persons.
Method of analysis. Partial least squares structural equation modeling (PLS) was applied to
test both the determinants and the outcomes part of the conceptual model (e.g., Anderson,
Fornell, and Mazvancheryl 2004; Smith and Barclay 1997; White, Varadarajan, and Dacin 2003).
PLS allows a simultaneous testing of hypotheses, taking indirect model effects into account.
When compared to covariance-based structural equation modeling approaches such as LISREL
and AMOS, PLS enables single-item measurement as well as multi-item measurement and the
modeling of constructs as either reflective or formative. As a distribution-free method, PLS has
fewer constraints and statistical specifications than covariance-based techniques. PLS Graph

13
had not downloaded movies themselves. For the variable ‘watching downloaded motion
pictures’, the mean was 28.1 movies in study 1 (33.7 in study 2) with a standard deviation of 64.5
in study 1 (61.7 in study 2). Table 1 lists descriptive statistics and correlations between
constructs.
------------------------------------------------------------------------
Insert Table 1 approx. here
------------------------------------------------------------------------

Study 1 Results: Consequences of Downloading Behavior
We used PLS to test the impact of downloading behavior on traditional movie consumption
patterns in two ways. First, we used the absolute numbers of movies downloaded and
downloaded movies watched by respondents as well as the absolute frequencies of DVD rental,
DVD sales, and visits to movie theaters in the calculations. Second, we calculated the logarithmic
transformation of all variables and ran PLS with these transformed measures. As a logarithmic
regression implies a degressive/progressive course of the function, this was done to account for
potential non-linear relations between downloading and DVD consumption/movie theater visits.
Although not formally stated, the two key constructs of movie downloading and watching
downloaded movies are significantly correlated, but, as theoretically predicted, the correlation is
far from perfect with a path coefficient of .781 (standard model) and .812 (logarithmic model),
respectively. Table 2 lists the standardized path coefficients as calculated by PLS and the
respective t-values generated through bootstrapping with 479 subsamples.
------------------------------------------------------------------------
Insert Table 2 approx. here
------------------------------------------------------------------------

The variance explanation is 3.6 percent for DVD rentals and less than 1 percent for both
DVD purchases and theater visits. For each of the three outcome variables, a significant negative

14
impact of either movie downloading or watching downloaded movies as predicted in H1 to H3
could not be demonstrated, whether using the linear or the non-linear model. The only significant
path of the consequences-side of the conceptual model is the one from watching downloaded
movies to DVD rentals, which is positive, indicating that DVD rentals are higher when
consumers watch a large number of downloaded movies. In other words, the empirical results
provide support for H1Alt, H2Alt, and H3Alt. More specifically, the positive paths from watching
downloaded movies to DVD rentals as well as to movie theater visits (although non-significant)
is consistent with the argument provided by Gopal, Bhattacharjee, and Sanders (2004) that
consumers use downloaded material as a “sample” to reduce uncertainty. No such positive impact
is found for DVD purchases, which suggests that the prior viewing of the movie makes it
unnecessary for the consumer to obtain more information through downloading. Downloading
per se (i.e. excluding the effect of watching downloaded movies) correlates slightly negatively
(but non-significantly) with movie theater visits and DVD rentals. This indicates that the
availability of downloaded movies tends to be accompanied by a slightly reduced amount of
traditional movie consumption. This effect is, however, fully compensated for when the
downloaded material is watched, as the total effect of downloading behavior on theater visits,
DVD rentals, and DVD purchases is close to zero for each of the variables, adding further
support to the non-existence of a negative effect of downloading movies on traditional
consumption channels (see Table 2).
To obtain a better picture of the impact of movie downloading, we extended the model by
considering the consumers’ involvement with motion pictures in general, adding paths from
involvement to movie theater visits, DVD rentals, and DVD purchases, and then tested the model
again with PLS. Although involvement is found to exert a strong impact on all three kinds of
traditional movie consumption (rtheater = .41; rbuying = .58; rrental = .35), it does not influence the

15
results. Specifically, the paths from downloading behavior to traditional movie consumption
remain largely unchanged.2 That said, the results suggest that a consumer’s reaction to movie
downloading does not depend upon the consumer’s movie-related involvement.
Study 2 Results: Determinants of Downloading Behavior
The determinants-side of the conceptual model was also tested with PLS in two ways. In
addition to using the absolute numbers of movies downloaded and watched, we again used the
logarithmic transformation of these variables in a separate calculation. Table 3 lists the
standardized path coefficients as calculated by PLS and the respective t-values generated through
a bootstrapping run with 500 subsamples. The correlation between downloading and watching
downloaded movies is slightly smaller in study 2 than in the first study, but still significant and
strong (path coeffient = .443 for the standard model and .685 for the logarithmic model). When
comparing the predictive power of the two models, the variance explanation of downloading
behavior is clearly higher in the logarithmic model, with 26.0 percent compared to 17.2 percent
for movie downloading and 58.2 percent compared to 26.2 percent. We therefore concentrate on
the logarithmic model when discussing the results.
------------------------------------------------------------------------
Insert Table 3 approx. here
------------------------------------------------------------------------

For all five variables, the proposed significant impact was supported by the data, extending
our understanding of the present movie downloading phenomenon. At the same time, we find that
the impact of the variables on downloading behavior differs (a) between the two kinds of
downloading behavior considered in our study and (b) between the variables. Specifically, we

2 When involvement is included, the paths are as follows: Movie downloading on (a) movie theater visits =
-.123; (b) DVD purchases = .017; (c) DVD rentals = -.220; watching downloaded movies on (a) movie theater visits

16
find that the consumer’s movie preferences have the strongest impact on the number of files
downloaded, followed by the consumer’s level of peer-to-peer expertise. Both TV playability of
downloaded files and their sound and picture quality do not determine the extent of file
downloading, but do determine the number of downloaded movies consumers are actually
watching. Apparently, consumers do not found their decision to download movies on the quality
they expect to receive; a finding that might be explained by peoples’ desire to collect hedonic
products (e.g. Belk 1995). Overall, our results help to provide a deeper understanding of movie
downloaders’ motives, supporting the assumption that consumers with an interest in blockbuster
movies are particularly inclined to download movies through peer-to-peer networks.
DISCUSSION OF RESULTS, IMPLICATIONS, AND LIMITATIONS
This paper examines two related aspects of consumer downloading behavior, namely,
downloading movies through peer-to-peer networks and watching downloaded movies, for their
impact on movie consumption through traditional channels, as well as the determinants of such
behavior. Regarding the former, as we find no significant negative correlation between
downloading and legal consumption, our findings contrast with the argument often stated by
industry representatives that downloading substitutes legal movie consumption. The separation of
the two aspects of downloading behavior in this study allows us to explain these results, as we
find a (limited and non-significant) negative impact of technical downloading on movie theater
visits and DVD rentals, which is fully compensated by a positive correlation between watching
downloaded movies and theater visits and DVD rentals. The latter effect which is significant in
the case of DVD rentals supports the “sampling” argument put forth by scholars (Gopal,
Bhattacharjee, and Sanders 2004). Overall, these oppositional effects cancel each other out so

= .113; (b) DVD purchases = -.148; (c) DVD rentals = .289.

17
that the total effect of movie downloading behavior on traditional movie consumption amounts to
zero. As for the determinants of movie downloading, we identify three distinct groups of
downloading “drivers”, namely, consumer, infrastructure, and content characteristics, which help
to understand this wide-spread phenomenon and go beyond the simplistic explanation “because
it’s free” (e.g. MPAA 2004d).
Overall, our findings imply that the movie industry’s reaction to the perceived threat of
movie file sharing, namely, suing its own customers, is misguided. Probably due to the large size
of files to be downloaded and the time required to do so, peer-to-peer downloading of movies
does not correlate significantly with traditional movie consumption and should be treated
accordingly.Considering that the quality of the downloaded material would be clearly better in
the case of legal offerings such as MovieLink or CinemaNow, our results suggest that such
services would be considered more attractive by consumers.
As with every study, our results are limited to a certain extent. First, the study relies on
cross-sectional rather than longitudinal data. Without the latter, no true causal effects can be
established. Rather, our analysis reveals the “ceteris paribus” impact of file sharing on the
respondents’ legal movie consumption, derived from the behavior of comparable respondents. A
longitudinal study would thus be an interesting direction for future research. Second, most
constructs were measured using single-item scales due to the lack of existing established scales.
Despite the conceptual tightness of most concepts under scrutiny in this paper, studies that work
on the development of valid multi-item scales would be welcome. Finally, as our sample covers
respondents from numerous countries and cultures, but does not allow the identification of
cultural differences in downloading, it would be interesting to see such comparisons in other
papers in the future.

18
REFERENCES
Anderson, Eugene W., Claes Fornell, and Sanal Mazvancheryl (2004), “Customer Satisfaction
and Shareholder Value,” Journal of Marketing 68 (Oct) 172-185.
Armstrong, J. Scott, Roderick J. Brodie, and Andrew G. Parsons. 2001. “Hypotheses in
Marketing Science: Literature Review and Publication Audit,” Marketing Letters 12 (May):
171-187.
BayTSP (2004), “Top Pirated Films on P2P,” [available at
http://www.baytsp.com/whatsnew.html].
Belk, Russel W. (1995), Collecting in a Consumer Society, London: Routledge.
Business Week (1997), “Movies: Be Happy, Film Freaks,” [available at
http://www.businessweek.com/1997/21/b3528144.htm].
Chin, Wynne W. (2001), "PLS-Graph Manual version 3.0," Houston: University of Houston.
CNC (2004), "Le téléchargement de films sur Internet," [available at
http://www.cnc.fr/a_presen/r4/telechfilmsInternet.pdf].
FFA (2003), "Brennerstudie 2," [available at
http://www.ffa.de/start/download.php?file=publikationen/brenner_studie2.pdf].
Forbes (2004), "Europe's Net file-swappers unfazed by lawsuits," [available at
http://www.forbes.com/business/commerce/newswire/2004/05/24/rtr1381988.html].
Goodhue, Dale L. and Ronald L. Thompson (1995), "Task-Technology Fit and Individual
Performance," MIS Quarterly 19 (Jun): 213-236.
Gopal, Ram D., Sudip Bhattacharjee, and G. Lawrence Sanders (2004), “Do Artists Benefit from
Online Music Sharing?” Journal of Business, forthcoming
Henning, Victor (2004), “An Empirical Study of the Effects of Peer-to-Peer Filesharing on the
Film Industry,” working paper, Bauhaus-Universität Weimar.
Hess, Thomas, Markus Anding, and Matthias Schreiber (2002), "Napster in der Videobranche?
Erste Überlegungen zu Peer-to-Peer-Anwendungen für Videoinhalte," in Peer-to-Peer, Detlef
Schoder, Kai Fischbach, and René Teichmann, eds. Berlin: Springer.
IFPI (2003), "The Recording Industry - World Sales 2002," [available at
http://www.ifpi.com/site-content/statistics/worldsales.html].
---- (2004a), "The Recording Industry - World Sales 2003," [available at
http://www.ifpi.com/site-content/statistics/worldsales.html].
---- (2004b), "Jahreswirtschaftsbericht 2003," [available at
http://www.ifpi.de/jb/2004/absatz.pdf].
---- (2004c), "Response to the Harvard/North Carolina study released 29 March 2004," [available
at http://www.ifpi.com/site-content/press/20040330n.html].
International Video Federation (2003), European Video Yearbook 2002/2003, Brussels:
International Video Federation.
Levine, David K. and Michele Boldrin (2001), "Why Napster is Right," working paper, UCLA.

19
Levine, David K. and Michele Boldrin (2002), "The Case Against Intellectual Property," in
American Economic Review Papers and Proceedings 92: 209-212
Marcolin, Barbara L., Deborah R. Compeau, Malcolm C. Munro, and Sid L. Huff (2000),
“Assessing User Competence: Conceptualization and Measurement,” Information Systems
Research 11 (Mar): 37-61
MPAA (2004a), "Valenti reports 2003 box office as second highest in history at Showest,"
[available at http://www.mpaa.org/jack/2004/2004_03_23a.pdf].
---- (2004b), „Studios to Begin Suing Illegal Film File Swappers,” [available at
http://www.mpaa.org/ CurrentReleases/2004_11_04_PressRelease.pdf].
---- (2004c), "Respect Copyrights," [available at http://www.respectcopyrights.org].
---- (2004d), "Worldwide Internet Piracy Study." [available at
http://www.mpaa.org/MPAAPress/2004/2004_07_08.pdf].
Oberholzer, Felix and Koleman Strumpf (2004), "The Effect of File-Sharing on Record Sales -
An Empirical Analysis," working paper, Harvard Business School.
Reuters (2000), "Online piracy of films grows rampant," Los Angeles: Reuters.
Sawyer, Alan G. and J. Paul Peter (1983), “The Significance of Statistical Significance Tests in
Marketing Research,” Journal of Marketing Research 20 (May): 122-133.
Smith, J. Brock and Donald W. Barclay (1997), "Effects of Organizational Differences and Trust
on the Effectiveness of Selling Partner Relationships,” Journal of Marketing 61 (Jan), 3-21.
Websense (2003), "Beyond the Music: Peer-to-Peer File-Sharing Web Sites Grow 300 Percent,
Driven by Swapping Movies, Games and more." [available at
http://www.websense.com/company/news/pr/Display.php?Release=03012359].
White, J. Chris, P. Rajan Varadarajan, and Peter A. Dacin (2003), “Market Situation
Interpretation and Response: The Role of Cognitive Style, Organizational Culture, and
Information Use.” Journal of Marketing 67 (Jul): 63-79.
Zaichkowsky, Judith Lynne (1985), “Measuring the Involvement Construct,” Journal of
Consumer Research 12 (Dec): 341-52.

20
FIGURE 1
A Model of Consequences and Determinants of Motion-Picture Downloading Behavior
File sharing
knowledge
Movie-related
preferences
Bandwidth
TV
playability
Technical
quality
Movie
downloading
DVD rentals
Movie theater
visits
DVD sales
Co
ns
um
er
c
ha
ra
ct
er
ist
ics
In
fra
st
ru
ct
ur
e
ch
ar
ac
te
ris
tic
s
M
ov
ie

ch
ar
ac
te
ris
tic
s
Determinants Consequences
Watching
downloaded
movies
Core constructs

21
TABLE 1
Descriptive Statistics and Correlations
Construct M SD N 1 2 3 4 5 6 7 8 9 10 11
1 Movie downloads within last 12
months
23.40 / 29.09 84.55 / 65.74 415 / 547 1
2 Downloaded movies watched
within last 12 months
21.40 / 29.15 57.40 / 58.55 415 / 547 .781** /
.490**
1
3 Movie theater visits 13.26 13.14 415 -.041 -.048 1
4 DVD rentals 14.60 28.05 415 -.014 .039 .154** 1
5 DVD sales 18.94 37.87 415 -.020 -.057 .303** .097* 1
6 File sharing knowledge 2.68 1.05 541 .299** .240** n.a. n.a. n.a. 1
7 Movie-related preferences 3.70 1.29 475 .306** .275** n.a. n.a. n.a. .407** 1
8 Bandwidth 3.73 0.92 542 .238** .168** n.a. n.a. n.a. .331** .311** 1
9 TV playability 1 2.23 1.51 432 .140** .140** n.a. n.a. n.a. .230** .218** .019 1
10 TV playability 2 2.13 1.42 429 .063 .131** n.a. n.a. n.a. .143** .120* .044 -.007 1
11 Technical quality 1 3.36 1.07 486 .242** .219** n.a. n.a. n.a. .214** .415** .011 .196** -.059 1
12 Technical quality 2 3.32 1.09 485 .235** .205** n.a. n.a. n.a. .219** .375** .086 .150** -.049 .785**
NOTE: Numbers before the slash refer to study 1, numbers after the slash to study 2. N includes all participants of the study, i.e. all people who have downloaded
movies, have watched downloaded movies, or have done neither
**. Correlation is significant at the level of 0.01 (two-tailed)
*. Correlation is significant at the level of 0.05 (two-tailed)

22
TABLE 2
Impact of Downloading Behavior on Traditional Consumption Channels
Hypo-
thesis
Effects of On: PLS
estimates
(linear
model)
t-
values
(linear
model)
H
supported
?
PLS
estimates
(non-linear
model)
t-values
(non-
linear
model)
H
supported?
Total
effects
(linear /
non-linear
model)
Movie
down-
loading
Movie
theater
visits
-.007 .07 H1Alt -.052 .51 H1Alt -.023 /
.046
H1
Watching
downloaded
movies
Movie
theater
visits
-.043 .52 H1Alt .122 1.13 H1Alt -.043 /
.122
Movie
down-
loading
DVD
pur-
chases
.063 .49 H2Alt .116 1.00 H2Alt -.020 /
.006
H2
Watching
downloaded
movies
DVD
pur-
chases
-.106 .82 H2Alt -.136 1.29 H2Alt -.106 / -
.136
Movie
down-
loading
DVD
rentals
-.112 .71 H3Alt -.161 1.16 H3Alt -.014 /
.079
H3
Watching
downloaded
movies
DVD
rentals
.126 .80 H3Alt .296 1.95 H3Alt .126 / .296
TABLE 3
Impact of Determinants of Downloading Behavior
Hypothesis Effects of On: PLS
estimates
(linear
model)
t-values
(linear
model)
PLS
estimates
(non-
linear
model)
t-values
(non-
linear
model)
H
supported?
Total
effects
(linear /
non-linear
model)
H4 Consumer
file sharing
knowledge
Movie
downloading
.177 4.01 .146 1.91 +/+ .177 / .146
H5 Consumer
movie
preferences
Movie
downloading
.139 2.32 .333 4.19 +/+ .139 / .333
H6 Bandwidth Movie
downloading
.114 3.10 .119 1.73 +/+ .114 / .110
TV
playability
Movie
downloading
.040 0.80 .034 .70 -/- .040 / .034H7
TV
playability
Watching
downloaded
movies
.114 1.94 .126 2.23 +/+ .132 / .149
Movie
quality
Movie
downloading
.143 2.38 .076 1.20 +/- .143 / .076H8
Movie
quality
Watching
downloaded
movies
.101 2.13 .131 2.07 +/+ .168 / .189

23
APPENDIX
List of Items
Construct Measurement Scale
Movie downloading “Approximately how many movies do you
download per year through P2P networks?”
Metric
Watching downloaded
movies
“Please estimate: per year, how many of the movies
you watched had been downloaded (or burned from
a downloaded file)?”
Metric
Movie theater visits “Please estimate: per year, how many films do you
see in the cinema?”
Metric
DVD rentals “Please estimate: per year, how many films do you
rent on DVD?”
Metric
DVD purchases “Please estimate: per year, how many films do you
buy on DVD?”
Metric
File sharing knowledge “How good is your knowledge of file sharing
networks?”
Ordinal four-point scale
Movie-related preferences “How often do you find and get exactly the film
you want in your file sharing system?”
Ordinal five-point scale
Bandwidth “What bandwidth Internet connection do you have
at home?”
Ordinal five-point scale
TV playability “Where do you watch the movies you have
downloaded?
1. On a TV connected to the computer via a
video cable, or.
2. On TV+DVD after I have burned the file
onto DVD-R or CD-R. ”
Ordinal five-point scale
Technical quality “How do you rate the technical quality of
downloaded films in general?
1. Image quality
2. Sound quality”
Ordinal five-point scale
Involvement “How do you feel about films?
1. Films are very important to me.
2. I have favorite directors and try to see as
many of their films as possible.
3. I actively search for information on new
films that will be released, e.g. by reading
reviews.
4. I regularly read film magazines or browse
film websites.”
Ordinal five-point scale