Exploring the Dynamics of Music Piracy
Trond Nilsen,” Brian Houle,” Douglas Kuzenski,’ Arpan Sheth?
‘Deparment of Systems & Industrial Engineering, University of Washington
? Department of Sociology, University of Washington
3 Information School, University of Washington
. Corresponding author: Industrial and Systems Engineering, University of Washington, Box
352650, Seattle, WA 98195-2650; (206) 913-3331; xorgnz@gmail.com
Abstract
Recently sales of physical music media have declined along with music industry revenues — the
reasons are complex and poorly understood. We present three models exploring claims made in
the conversation over piracy and the music industry's future. We model stakeholder perspectives
rather than the full industry to identify their assumptions and the implications of their views,
including: (1) Music industry labels claim music piracy directly supplants music sales; (2)
Artists claim free music distribution can increase artist revenue through concert sales; and (3)
Some academics claim reduced music sales revenue is attributable to other market changes (e.g.,
increased entertainment competition) rather than piracy. Modeled results suggest that, for each
claim respectively: (1) Music labels' mental models are simplistic and do not reflect current
research findings; (2) Artists can enhance their concert revenues through piracy as free
marketing; and (3) competition with other media explains part of declining music industry
revenues.
1.0 Introduction
The music industry arose in the mid-20th century from the ability to inexpensively mass-produce
vinyl records. Over the years, revenues continued to climb as technology shifted from vinyl to
cassette tapes and then to compact discs. All three utilized a common business model, allowing
the industry to maintain its existing infrastructure. Over the past decade, however, sales of music
via physical media have declined sharply along with music industry revenues. Despite this, more
consumers report spending money on music than ever before.
Unfortunately, the reasons for this decline are complex and poorly understood (Connolly &
Krueger, 2005; Liebowitz, 2006). In response, researchers, and particularly economists, have
begun to look at the effects of file sharing on CD sales. Various effects have been suggested,
though empirical studies are ambiguous and have found different directions (Bhattacharjee et al.,
2006a; Hui & Png, 2001; Oberholzer & Strumpf, 2005; Peitz & Waelbroech, 2005; Rob &
Waldfogel, 2006; Zentner, 2006). Gopal, Bhattacharjee, & Sanders (2004) present a sampling
argument, whereby file sharing has a positive effect on CD sales by allowing potential customers
to hear new artist and music before purchasing them. Conversely, a substitution process would
have an opposite effect, where downloaded music replaces purchased CDs.
Due to ambiguous theoretical predictions, mixed evidence, and opposing interests, various
groups within the industry operate with different mental models of the situation, leading to
differing policies and claims. System dynamics affords the opportunity to evaluate these claims,
establish any assumptions and implications they might embody, and, potentially, recommend
policies that would result in better outcomes.
The dominant view point is that online piracy has offered consumers an alternative way to obtain
music that undermines the industry’s revenues, potentially to the point of destroying it. In their
2009 Digital Music Report, the International Federation of the Phonographic Industry clearly
stated this claim, saying “The vast growth of unlawful file-sharing quite simply threatens to put
the whole music sector out of business” (IFPI, 2009). Liebowitz (2006, Figure 3 p.14), in
reviewing the drop in music sales, shows that the drop in music sales occurred at the same time
as online peer to peer music sharing services began to operate, starting with Napster.
Alternatively, various scholars (e.g., Oberholzer & Strumpf, 2004), artists, and Internet activists
have suggested that the decline in music revenues derive instead from changes in consumer
demands, expectations of digital distribution, and the growth of other forms of entertainment
competing for the consumer’s money (Peitz & Waelbroech, 2004). Some have even gone so far
as to suggest that music piracy can actually be good for the industry, at least with regards artists.
In these arguments, music piracy is considered a form of free marketing that allows artists to sell
more concert tickets and merchandise (Gayer & Shy, 2006; Peitz & Waelbroech, 2005).
Given the complexity of this issue, and the fact that consumer behavior is poorly understood in
this area (Gopal et al., 2002), we did not attempt to create a single authoritative model describing
the industry. Instead, we sought to model the specific claims made by various stake holders. By
doing so, and by examining the implicit assumptions and implications, we discuss the feasibility
of each claim, and make recommendations accordingly.
1.1 Stakeholders
The music industry includes several different groups, each with their own perspectives and
motivations. For clarity, we distinguish between
e Artists — individuals or groups who create and music
e Labels — companies who provide recording, distribution and promotional services to
artists
e Retailers — companies who engage in the legitimate distribution of music.
e Consumers - individuals who acquire music for listening
e Pirates — individuals who seek to profit in some way from the illegal distribution of
music.
1.2 Problem Constraints
To keep the problem tractable, we constrained our investigation and discussion as follows:
e We have primarily focused on the music industry within the United States of America.
Partly this was due to a lack of good data concerning international music sales and piracy,
and partly because we wanted to avoid the much greater variability in consumer behavior
that we expect exists worldwide. Nonetheless, we expect that our findings will be at least
somewhat generalizable.
e We have not delved into issues of ethics or legality, nor do we consider qualitative
aspects of the music such as improved diversity, novelty, and accessibility.
e We focus only on the direct consumption of music by consumers. We have not
considered music licensed for re-use in movies & advertising, for re-use as samples in
music, or for public broadcast or performance.
1.3 Problem Statement
We have selected three specific claims to model:
e The claim by labels that music theft directly supplants music sales, and that pirates can be
converted into consumers who, instead of stealing, will buy the same amount of music
e The claim by some artists that free distribution of music, through piracy or by choice of
the artist, can increase revenue to artists through concert sales, merchandise, and the sale
of other works
e The claim that reduced industry revenue is attributable not to piracy, but to market
changes such as increasing entertainment choices and changes in consumer expectations.
2.0 Model 1 - Relationship between sales & piracy, as understood by labels
Our first model diagrams the mental model of large music labels — namely that pirated music
directly replaces a music sale, and in turn that reducing piracy will result in increased music
sales. In modeling the music industry's view we have relied mainly on their testimony and the
methodology of their cited data sources. For instance, in a recent trial with The Pirate Bay, a
peer-to-peer torrent sharing site, "... John Kennedy, the chief executive of the International
Federation of Phonographic Industries, testified that people would have purchased every music
track they got free file sharing. Kennedy answered an affirmative 'Yes' to Pirate Bay defense
attorneys when asked whether that was true." (Swartz, 2009).
Further, upon reviewing the RIAA's claims of lost revenue due to piracy they note: "One credible
analysis by the Institute for Policy Innovation concludes that global music piracy causes $12.5
billion of economic losses every year." (http://www.riaa.com/physicalpiracy.php). Upon
reviewing the report, data for the estimates of units sold if piracy were nonexistent are based on
".,.absent piracy, consumers of pirated products would substitute legitimate purchases for all or
nearly all of the pirate purchases that they now make." (IPI, 2007, p.7).
Based on these testimonies and analytical approaches, our model reflects how labels conceive of
pirates and consumers. The total population is divided into people who are either pirates or are
consumers, and every album pirated is a lost sale from the total pool of "potential sales." By
making the reasons to be a consumer outweigh the reasons to be a pirate, the flow of people will
be from pirate to consumer (Bhattacharjee et al. 2006b). In this way, the labels will recoup their
lost potential sales and increase revenues. Below we discuss portions of the model separately for
ease of presentation.
2.1 Assumptions
We have assumed that the average popularity of recorded music increases linearly with time.
This assumption matches the predicted increase in sales according to Liebowitz (2006). We also
assume that consumers’ incentives to pirate music began in the year 2000, and have been
increasingly linearly since. These incentives include the spread of piracy technologies and
people’s awareness of these technologies. Finally, we’ve assumed that the music industry’s
reaction to piracy, in the form of anti-piracy actions, will be maintained proportional to the rate
of piracy.
2.2.1 Sub-model 1 Music Production-Consumption
<Total Population>
Av Piracy Time
a Piracy rate ~
Copies
I “= _ll Stolen
Potential <Total Pirate Activity
sales _| Population> Fraction
Production Rat
eee ‘,. “S a aaal
sts “SS sae Rate
Be 4
Av Album . Av Purchase
Production Time Av Populatity Time
Popularky Av Popularity
Change Rate Stock
Figure 1 — Sub-model 1 Music Production-Consumption
This sub-model reflects the mental model of how the music industry defines potential sales lost
to piracy. The potential sales stock reflects the music labels’ perceived value of their products in
the current market. These potential sales are defined by the number of artists producing music,
how quickly they can produce albums, and the number of potential customers in the market, as
well as a linearly increasing coefficient of the average popularity of music. As per the claim this
model is based upon, given that there is no piracy, all potential sales are converted into actual
sales. As piracy increases, though, the potential sales are instead converted into copies stolen.
The change in Potential Sales is measured by the Sales Rate and the Piracy Rate of albums.
2.2.2 Sub-model 2 - Pirates-Consumers
This sub-model focuses on the division of the music industry market into those who are pirates
and those who are consumers. Together these two groups make up the Total Population of the
market. The conversion rate is defined as how quickly people move between the two groups, and
in which direction. This rate depends on the various incentives for becoming a music pirate and
the incentives for becoming a music purchaser, or anti piracy factors and pro piracy factors
respectively. A high anti piracy factor might reflect very tough laws against piracy combined
with advances in technology
to stop piracy. The steady
linear increase in the pro
A piracy factor of our model,
Anti Piracy ‘ ar
ior Plate <== Factors simulated as beginning in the
ti Vv
year 2000, reflects the
constant advances in the
technology of piracy and
consumer access to it. This
Wa Rate Total Population allows our data to roughly
correlate to the sales data of
the RIAA. These two
competing factors are then
used to determine the
direction and magnitude of
v change in the pirates and
Consumers consumers stocks, based on
whether the environment is
Pirates
Pirate Activity
Fraction
mail|
Time
Pro Piracy
Stock
Pro Piracy
Change Rate
Figure 2 — Sub-model 2 Pirates-Consumers relatively favorable or hostile
for pirates.
2.2.3 Full Model
Integrating the two sub-models, gives us the following full model.
Pirates
‘anti Pirate ——PeAnt Piracy
Action Level
Ay Piracy Time
_— Piracy rete me
l & <P |_Stolen a Rate}! Total Population
O: Potential a Pirate Activity
sales | P Fraction
ga Rate l «
ew sires Sold wll |
Artists Af "Se wate, Time
Ay Album APCS Av Purchase
Production Time puny: Time
Consumers
Popularity AV Popularity
Change Rate ne Pro Piracy
Pro Piracy Stock
Change Rate
Figure 3 — Full Model 1
2.3 Model testing
To critically assess the model we followed Sterman's guidelines for model testing (Sterman,
2000, p.852).
2.3.1 Purpose, Suitability, and Boundary
We used our reference mode behavior to determine the appropriate time horizon for the model,
ranging from 1973-2005. The below model boundary chart describes the boundary of the model.
As our goal for model 1 was to explicitly model the assumptions of the recording industry's
argument for the decline in music sales, the exogenous and excluded variables are appropriate.
Moreover, the model excludes many important system feedback loops, which correspond with
omissions in the industry's position.
Table 1. Model 1 boundary chart
Endogenous Exogenous Excluded
Potential sales Population growth Limit to growth in potential sales
Average purchase and piracy time Supply Effect of price
Copies stolen and sold Demand Competition
Anti & Pro Piracy factors
Pirates and consumers
2.3.2 Physical and Decision-Making Structure
A review of the structure of the model showed that it conforms to basic laws and is
dimensionally consistent. The model excludes time delays and other constraints as well as
decision-making behavior because these are not taken into account in industry arguments.
2.3.3 Robustness and Sensitivity to Alternative Assumptions
We varied each parameter in the model using Vensim's simulation feature to ensure that its range
of values generated expected behavior in the stocks and flows. The model appears robust to our
initial testing.
2.3.4 Pragmatics and Politics of Model Use
To develop and test the model we used several estimated parameters and ranges from the
literature, including:
« Potential sales (Liebowitz, 2006)
« number copies stolen (HUI & Png, 2001; Rob & Waldfogel, 2006)
« average period of popularity (Bhattacharjee et al., 2006b)
* conversion rate for anti-piracy factor (Bhattacharjee et al., 2006a)
« number of artists (Gopal, Bhattacharjee, & Sanders, 2004)
« number of albums (Gopal, Bhattacharjee, & Sanders, 2004;Rob & Waldfogel, 2006)
« number pirates (Liebowitz, 2006)
2.4 Results
We have simulated the mental model of the labels regarding how they perceive piracy’s impact
on their business. In the following two graphs, we have assumed that the average piracy time and
average purchase time remains moderate. The conversion time from pirates to consumers and
vice versa also remains moderate. However, we have varied the anti piracy action level from one
extreme value to another. The anti piracy action level describes the magnitude of the industry’s
response to the existence of music piracy.
As we can see from the graphs, as the anti piracy action level is increased, the sale rate has also
been increased.
fom i [oe 12
CA 12
‘Av Piracy Time Conversion Tizne ‘Ay Purchase Time
People sale rate
2,000 People 80,000
400 People
1,500 People 2
200 People = 40,000
bo] | 3
1,000 People Anti Piste Action Level | 2
0. People
1990 1998 2006 2014 0
Time Cem) 1990 1998 2006 2014
Consumers : action level 1 People Time (Year)
Pirates : action level 1 ————— People Ske Hie -acifon'tevsl't
Figure 4 — Model Results — Effect of Piracy on Music Industry
(ous 12
ok i2|
Ay Piracy Time Ay Purchase Time
People sale rate
2,100 People 80,000
100. People
2,000 People 3
50 People = 40,000
a
1,900 People ‘AntiPirste Acton Level | ©!
0 People
1990 1998 2006 2014 6
Tine. Zenr) 1990 1998 2006 2014
Consumers : action level 1 People Time (ear)
Pirates action level 1 ————— People Si Raeantinsioedlt
Figure 5 — Model Results — Efect of Anti-Piracy Measures
2.5 Conclusion
We have modeled the labels’ claims regarding the impact of piracy on music sales. Some of the
assumptions in this mental model do not hold up well to scrutiny. For example, though it is
indicated that music popularity will continue to increase indefinitely, this is impracticable over
any lengthy amount of time. Also, the claim states that as pirates are converted into consumers,
they purchase the same number of albums they would have pirated. This is how our model
calculates the labels’ perceived amount of potential sales. In reality, there does not seem to be
any evidence to support this claim. The rate at which people pirate music does not have a 1 -to-1
correlation to the rate at which they would otherwise purchase music. Our recommendation to
the segment of the music industry maintaining this claim would be to reconsider their
assumptions. By expressing these views to the public at large, their case against piracy is not
strengthened and may, in fact, be compromised.
3.0 Model 2 - Effect of unpaid distribution on artist revenues
Our second model addresses the claim by artists that the free distribution of albums, through
piracy or by choice of the artist, can increase revenue to artists through concert sales,
merchandise, and the sale of other works. Alan Krueger, a Princeton economist, refers to this
claim as the "David Bowie" hypothesis, based on a famous quote:
"Music itself is going to become like running water or electricity... You’d better be prepared for
doing a lot of touring because that’s really the only unique situation that’s going to be left.”
— David Bowie , New York Times, June 9, 2002
In order to investigate this claim, our reference mode data refer to the actual impact of concert
income on total artist revenue at present. Data based on Pollstar from Krueger (2004) shows that
for the top bands concert income is indeed their main source of revenue by a factor of 7.5 to 1.
Furthermore, there is a large disparity between popular superstar-type bands and smaller ones.
Krueger (2004) shows that the top 1% of bands control over 50% of the revenue from concerts.
Given these data, our dynamic hypothesis is that free distribution of music online will generate
increased artist revenue through concerts. To assess the face validity of the hypothesis, we
developed an operational model of artist music production, distribution, and income for an
individual artist/ band.
For ease of understanding, we will discuss this model in stages before integrating it together into
a single model.
Marketing
3.1.1 Sub-model 1 - od
‘ 5 raction
Viable Music
First, let us consider _-_ Giveaways Per
Month
the artist's music
catalog. Instead of
considering the total
amount of music
i
+ Free Distribution
e Viability
Te First Album
Pa pkey
. a Rate Max
created by the artist, S pea ———— — oe
IS jusic
we model only Gatalb9 | production Rate epee
music that is viable -
worth promoting and
performing. Music is
. Av Cost of +
produced according vin Cost of Desired
to the band's desire
to produce music
and the affordability
of music production.
Production
Production ome Affordable |
Production Ratio Desired
Production Rate
Figure 6 — Sub-model 1 Viable Music
Viable music decays over time according to a first order delay. This accommodates the fact that
the longevity of each track is highly variable - some tracks become classics while others are
forgotten almost immediately.
The amount of viable music in an artist's catalog directly affects how much marketing they
employ and how many tracks they give away for free. It is important here to distinguish clearly
between viable music and total music as a band whose music has become non-viable is unlikely
to engage in marketing.
3.1.2 Sub-model 2 - Fans
Marketing:
\
\ +
Growth Rate ~<_
From Marketing +
Marketing
/~ Effectiveness
Free Distribution
Effectiveness
Fans
\
Music
Attractiveness
Fi Non-Fans
Distribution
+ cy p
+ F
* we
Growth Rate ei Za
From Distribution eer
: al
=<
+
we,
—_ Boredom Rate
+ “Av Boredom Time
Figure 7 — Sub-model 2 Fans
3.1.3 Sub-model 3 - Money
Based on
effectiveness
marketing
and general
of an artist's
music, money spent on
marketing leads to a growth in
the number of fans. Similarly,
sampling consumers who come
across the band's
available for free may become
fans. Over time, fans become
bored with a band's music. The
number of fans directly affects
sales of both albums and
concert tickets. Fans also affect
artist motivation, as artists are
assumed to be partly motivated
by having their music be
considered popular.
attractiveness
music
Artists acquire money primarily through sales of their albums and sales of concert tickets,
depending on the profit margins of each of these. Artists may also profit from merchandise sales
at concerts; for simplicity, this is considered part of their concert profits. Money is spent on
either marketing or production.
Marketing
“is eipsrie ture
Investment
Fraction
Ze
ones + Max Investment
are nO Available
+ +
a
Production
Investment.
Fraction
gee to Money
Album Sales
Artist
Motivation Due
Sale Rate
Albums
Album Income —
co
o-=t
invent
- } + Profit per
Maney es Album
jl y
Concert Income
+ +
eee Initial
+ Investment
Profit per
Concert Ticket
Figure 8 — Sub-model 3 Money
Concert Sales
Sale Rate
Concert
Tickets
The amount of money available directly affects spending on both marketing and production
depending on the desired level of reinvestment. Similarly, money contributes to a band's
motivation, and may constrain their production activities.
3.1.4 Sub-model 4 — Production
Thus far, we have discussed the
model's three stocks (viable
music, fans, and money), as well
as the linkages between viable
music and fans, and fans and
money. The rest of the model
concerns the impact of money and
fans on the production of new
music and is somewhat complex.
Firstly, artists are assigned a
‘desired production rate’. This rate
is increased by their motivation
due to money and popularity, and
constrained by the average time it
Viable
Music
Catalin
ge
a
Production Costs
Max
Production
Musle
Copacty
ction Rate“ \
ft "ig
ss,
_"
Artist
% 4
cote St pesred |
Production
i ad
sired
Production Rate
Pa
Ay Cost of
Production
+ affordable
Production Ratio
2
bs \ yt
Mex Investment
avaible ay Production
4, Time
Pd
Pa
Production *
Investment
Fraction
Figure 9 — Sub-model 4 Production
Motivatian Due
to Popularity
Motivation Due
to Money
Motivating
Power of
Populerity
Mativating
Power of
Money
Artist
takes music to be produced. In
addition, it is clamped to a maximum production capacity, for no matter how enthusiastic the
band may be, they cannot produce more than one or two albums at a time.
Next, the maximum amount of funds available for production is calculated and divided by the
cost of desired production, giving us a ratio of affordable production. This is combined with the
desired rate of production to give the actual rate of production. Finally, the cost of actual
production is calculated and applied as expenditure.
3.1.5 Full Model
These four Sub-models integrate together into a single model as shown below with the addition
of a few extra constants to control various auxiliary variables.
Marketing> Marketing
/~ Bffectiveness
te Gowth Rate
FromMaskting “+
7
Free Distrbution \
Marketing fectiveness Hose:
Investment Brees Attractiveness
a ' on-Fane
Fraction, Bans: | gees
Marketing Giveaways Per
Month ) J
a ~ Growth Rate
& \ fen he 4, |
bag
‘ay Viabilty
Tire
a ut ee Distribution <==
BoredornRate
peste RO 7 Max bs ?
Ome Mere » “Bepuetion ‘Seestens
‘pak Music Capacity “
Production Rate No
2
Artist
Motivation Due
to Popularty
Av Cost of + 7
Production Cost of Desired Motivating
Production. Power of
+ - Popularity
+
Production Costs“+ affordable + Motivating
Production Ratio Desired Power of
ie Production Rate % Money Naar
¥ Album Salet
IS + +
Max Investerent Artist +
‘Av Production Motivation Due
{ ‘Available
+
Concert Sales
ae
Production
Investment
Fraction
Sale Rate
‘Concert
Tickets
ine ats
Figure 10 — Full Model 2
3.2 Model testing
We again followed the guidelines issued by Sterman (2000, p.852) for testing model 2. As the
model tests a hypothetical fan population and individual artist, we assigned a time horizon of 1-
100 months for the model to assess both short and long-term trends. We also fixed the
hypothetical fan population at 10,000 for illustrative purposes.
3.2.1 Purpose, Suitability, and Boundary
We varied the time horizon in this model to evaluate the effects of short and long-term views on
the hypothesis modeled since the artist assumption is not explicit. The below model boundary
chart describes the boundary of the model. We were unable to disaggregate the marketing
process of the music industry due to lack of available information or insider knowledge, this is a
recognized limitation of the model. Further, due to lack of systematic data, we were unable to
model the costs and profits of concerts more explicitly.
Table 2. Model boundary chart
Endogenous Exogenous Excluded
Music production Music viability and marketing Less aggregated marketing effects
Music distribution/ marketing Distribution effectiveness Limited number of fans
Artist income Concert profit and sales rate Concert timing with album release
Growth and decay in fans Marketplace and competition
3.2.2 Physical and Decision-Making Structure
A review of the structure of the model showed that it conforms to basic laws and is
dimensionally consistent. The model excludes time delays and other constraints. Further, we
were unable to generate a decision-making structure for music fans due to a lack of consensus on
how individuals select their music (e.g., the economics literature is divisive on this issue). As
Sterman (2000) notes, in order to model mental data information must be elicited through
interviews and other methods beyond the scope of this project. While these are limitations, the
model is still representative of the claim by artists that we are attempting to model.
3.2.3 Robustness and Sensitivity to Alternative Assumptions
We varied each parameter in the model using Vensim's simulation feature to ensure that its range
of values generated expected behavior in the stocks and flows. The model appears robust to our
initial testing, however, given the large number of parameters, it is difficult to assess every
possible scenario.
3.2.4 Pragmatics and Politics of Model Use
To develop and test the model we used several estimated parameters and ranges from the
literature, including:
« Concert sales (Lewis, 2006)
« Profit per album in general (Sandoval, 2007)
« Profit per concert in general (Lewis, 2006)
« Av sales per person per month (Hui & Png, 2002)
« Music Decay Rate (Rob & Waldfogel, 2006)
« Average viability time (Sandoval, 2007)
3.3 Results
To test the face validity of the claim and our dynamic hypothesis, we explicitly modeled two
case scenarios: (1) a small, relatively new and unknown band; and (2) a large, experienced and
popular band. Our rationale in selecting these examples as cases was to assess whether or not
there was a differential effect of free distribution on concert revenue based on band parameters.
3.3.1 Small band
The figure below presents the results for a small band. In selecting parameters, we allowed
"giveaways per month" and "free distribution effectiveness" to remain the same as for the large
band. This is to allow for a direct comparison of the effects of free distribution on concert
revenue between the two cases. Since the band is small, we set that they had no investment in
marketing. Finally, we lowered their parameters relative to the large band for the
following: profit per album, profit per concert ticket, music attractiveness, average viability time,
average cost production, sale rate concert tickets and albums. Again, this was to reflect that the
band is less popular, is likely not producing as quality of music as a large band, and is less able
to demand higher prices for their products.
= wr ni
(too n] (ie 10] Money
Free Distribution Effectiveness ‘Av Boredom Time 1,000 Dollars
ihe 75 60. Dollars/Month
Marketing Efectiveness Ao VaR
oS
Olo n a
Marketing Investment Fraction
11500 lo
Production Investment Fraction 30 Dollars/Month
‘Ay Cost of Production
SS
Olo.o002
ivating Power of Mon =
‘Motivating Power of Money 010.05 Nt
ooo nN Sale Rate Concert Tickets
Motivating PowerofPopulsrity | —==-—} 0 Dollars
ol0.21 h 0 Dollars/Month
Sale Rate Albums
Profit per Album 0 15 30 a oa oa 75 90
ime (Mont
Dla 10 112 le
5:5 ‘Av Production Time Money : garage Dollars
Profit per Concert Ticket Makening gan? Datla
-—S cll Production Costs : garage. —————————-. Dollars
[ato.0001 ml 013,674 10 Album Income : garage Dollars/Month
Music Attractiveness Giveaways Per Month Concert Income : garage ——————————-_ Dollars/Month
Figure 11 — Small Band Revenues
From the results, we see that concert sales are the bulk of the band's income, though it decreases
over time due to their music being less viable. Given that these bands have low production costs
(e.g., by leveraging new electronic technologies), they are also likely to be producing new music
more frequently than a larger, more established band.
3.3.2 Large band
oF Mon
‘i.002 n ] [hie ia] ey
Free Distribution Effectiveness ‘Av Boredom Time 20,000 Dollars
———— 10,000 Dollars/Month
1124 lo
Marketing Effectiveness ‘Av Viability Time
o
Marketing Investment Fraction 5
Max Production Capacie
Production Investment Fraction | 712,300 10 5,000 Dollars/Month
—-——— fv Cost of Production
Olo.0002_ i
Motivating Power of Mone:
——" 010.1 in
Olo. 001 n Sale Rate Concert Tickets
Motivating Power of Popularity 0 Dollars
a [oa] 0 DoltarsMonth
Sale Rate Albums
takpr seen 0 15 30 45 60 75 90
0110.5 J ‘Av Production Time Money : Cusrent Dollars
Profit per Concert Ticket Marketing : Current Dollars
[ Tl Production Costs : Current. ————- Dollars
[oto.o01 Fl [9,674 0 Album Income ; Current Dollars/Month.
Music Attractiveness Giveaways Per Month Concert Income : Curent. ——————— Dollars/Month
Figure 12 — Large Band Revenues
The next figure presents the results for a large band. Parameter selections for this band were the
same as the small band scenario for "giveaways per month" and "free distribution effectiveness"
to permit comparisons. Further, we allowed the large band to have investment in marketing as
they are more popular. Finally, the following parameters were larger relative to the smaller
band: higher profit per album, profit per concert ticket, music attractiveness, average viability
time, average cost production, sale rate concert tickets and albums. These parameters were
increased since the band is more popular and has a more long lived music catalogue due to
having several "hit" songs. Also, they are able to negotiate higher prices for their concerts.
Finally, since they are likely recording in a studio, their production time and costs are also
higher than the small band scenario.
Similar to the small band results, we see that the large band's income is largely driven through
concert sales. However, this income shows a goal-seeking behavior, suggesting that a large band
can maintain a certain level of revenue due to their greater popularity and amount of music that
is viable.
3.4 Conclusions
It appears that the assumption has at least face validity -- free distribution can increase the fan
base through sampling, in turn increasing concert sales and a band’s income. While both big and
small bands can benefit from these effects, increasing adoption of this technology by smaller
bands along with decreased production costs may increase the competition among artists in the
marketplace. We might expect in the future that the percentage of total concert revenue captured
by the biggest bands will decay over time as more artists are able to compete. Indeed, the
literature has documented a greater amount of competition just in billboard listings over time
from smaller, alternative labels (Bhattacharjee et al., 2006b). However, our model is unable to
address these postulations due to excluding the marketplace and competition effects. A logical
extension of the model would be to include marketplace effects to better understand how free
distribution might affect bands differently. Further, future work could treat the population of
fans and pricing as endogenous factors. For instance, as a band grows in popularity over time
and its fan base ages, we would expect increased demand for concerts and therefore increased
prices for tickets (see Krueger, 2004 for a detailed discussion).
Finally, our model assesses the network effects for artist stakeholders only. A more complete
view of the competing interests of other stakeholders would allow us to analyze differential
network effects based on free distribution. For instance, Gayer and Shy (2006) built a model
assessing free distribution effects for artists as well as music publishers. Their results suggest a
conflict of interest, where artists benefit due to other revenue streams (e.g., concerts) while
recording publishers net a lower profit. Including other stakeholders would thus allow us to
endogenously model the role and network effects of interventions such as copyright
enforcement.
4.0 Model 3 - Albums sales affected strongly by non-piracy factors
This model shows the claim made by some academics, such as Oberholzer and Strumpf (2004),
that the causes and effects of music piracy are largely independent of music sales. That is, even
though music piracy may raise or fall dramatically, it will have only a minimal effect on sales.
Academics do cite two main connections between piracy and sales, though: a decrease via
substitution and an increase via sampling. The debate still continues as to the magnitude of these
forces, as well as their relative strength.
Avg Expenditures Deslrabllity of
Buying Non-Music
Entertalnment
Desirability of
Buying Music, + Avg Target Music
+ Expenditures
Avg Target Expenditures
on All Nen-Music
Entertalnment
i)
Es
Avg Music
Expenditures fag
[e 5
Avg Music Buying Music.
Sales Rate on a Budget
Overall Music Sales
Popul arty of Network Effects
_~
Other causes of
music popularity
SS Increase In
Music Spending
Avg Price of +
Album a Avg Number of
Albums Purchased
Per Month
Figure 13 — Sub-model 1 Entertainment Expenditures
over the past ten years, according to NPD data (Matthews, 2009).
4.1.1 Sub-model 1 -
Entertainment
expenditures
First, a consumer's average
expendable income is
considered to be relatively
fixed from individual to
individual. Therefore,
average expenditures on
entertainment must be
divided between
expenditures for music and
expenditures for all other
types of entertainment. The
money allocated to each is
largely dependent on the
relative desirability of
buying music versus other
entertainment related
expenses. This mental
model implies that looking
at music industry revenue in
isolation is avoiding the
larger picture of consumer
spending. Take, for
example, the video game
industry, which has shown
a marked increase in sales
Video games reflect just a single industry competing for consumers! expendable income. Other
competing industries include movies, both in theaters and on DVD, cable and satellite television,
broadband Internet, and various live events. As consumers decide to purchase other types of
entertainment related goods and services, their funds available for music purchases are offset.
As consumers have less money allocated to music purchases, they necessarily buy fewer
albums. This balancing action overcomes any positive network effects of music purchasing that
the industry has typically relied upon.
4.1.2 Sub-model 2 - Music Piracy
In this causal loop diagram,
Ease of Finding Avg number of
+
ae Ratest—— Population
Legal Fears of
Music Online + people sharing the primary causes of piracy
Gt each song are considered separately from
*- es
Network the music industry as a whole.
A vb isu Effects of As more people download
‘2 ay pe oa music illegally, there are more
people sharing each file,
Total Pirating
making it easier to pirate
music. This the Network
Effect of Sharing reinforcing
Plrating loop. Other factors for piracy
Avg Tort Music include the total pirating
Le:
Desired Song _Limited by on: Computers population and legal fears of
+ Downloads Technology piracy, both considered
exogenous to our model.
Avg Size of Disk s
The primary balancing loop in
Song File
this mental model is the
physical and _ technological
Avg Disk wa Nes limitations of computers. In
Space Music following with Moore's law,
Figure 14 — Sub-model 2 Music Piracy the speed and capacity of
computers has been rising
constantly for over two decades. As people are enabled to download a greater quantity of music
and store it on their personal computers, this mental model implies that they will proceed to do
so until they are satisfied with the amount of music they possess. Our causal loop diagram
shows the effect of the amount of available hard drive space dedicated to digital music on music
piracy. Following Moore’s & Kryder’s laws, the amount of hard drive space on new computers
has been increasing exponentially, from hundreds of megabytes in the mid 1990s to terabytes
today.
Though this technical capacity may eventually level off, it has far exceeded the amount needed
to download many thousands of songs illegally, leaving much capacity left over for other typical
19
uses of personal computers. Other technological factors that affect the rate of piracy include
average size of song files and average Internet download speed.
4.1.3 Full Model
wcsilines Desirability of Buying
ug Expert seaty ot
Entertainment
‘Avg Number of
os
Speed y a besiraitty of 5
many Ma, Avg Target Music AVd at Expenditures:
Bendre Al Nore Music
semen ne ice
ss = ee
i ~
a a
=)
Limited by 97 Computers
iinet septic
Poet se Teh * vera Peplaty ppoendume cap
ore.
ey samping fate of i
New Mish
nage Di sped cap
a
‘vg Total Music
Music Spending
aR
Goal Percentage of
‘Avg Disk Space” Dik Space Ful oF
SE svg mater of
‘Albums Purchased
ar Month
Figure 15 — Full Model 3
As can be seen in the full causal loop diagram, this mental model includes two primary
interactions between entertainment expenditures and music piracy. The first is the average
number of purchase substitutions: as people have the files on their computers, this may act as a
substitute for purchases, causing a reduced desire to buy more. The second is the sampling rate
of new music: as people discover music online, the overall popularity of music increases. This
can cause both more pirating, and more purchases of music via the Music Discovery loop.
Similarly, as the Music Sales Network Effects cause more Popularity in Music, people will pirate
more as well. In this model, the effects of piracy can be compared to those of the Music
Streaming Rate (i.e.: listening to music on the radio, on TV, or streaming online). As previously
mentioned, the magnitude of these effects is much debated. It appears that customer behavior
greatly depends on individual preferences and smaller group dynamics.
4.2 Model testing
Since this model is only a CLD, our testing was limited to the critical assessment of its
boundaries and assumptions as listed in the chart below. In particular the decision-making model
of consumers is left exogenous due to limited data; this is a recognized limitation of this model,
and a major reason that it was not developed into a full working model.
20
Table 3. Model boundary chart
Endogenous Exogenous Excluded
. : Effect of sales Il
Piracy Rate Average computer size/speed . SNORE IRAE ON OV SES
piracy
Network Effects on piracy Pirating population Music quality effects on sales
Average entertainment
Hard drive limitations Legal mp3 purchases
expenditure
Budget for music purchases Alternative entertainment Non-music listening population
Desirability of buying music Music Streaming Rate
Overall popularity of music
Effects of piracy on music sales
4.3 Conclusions
The claim that music purchasing and music piracy are only indirectly related, upon being
diagrammed, appears to be a possibility. Record labels recently have committed tremendous
resources to increasing computer users' legal fears of piracy. Though this potentially may have
the effect of reducing the rate of increase of piracy, this policy's impact on music sales is
indeterminate. As music sales depend more directly, according to this model, on how buying
music compares to other alternatives for consumer spending, music labels may better spend their
resources pursuing improving that comparison. By focusing on increasing the value of their
product for consumers, music labels can increase revenue. For example, by including special
promotional items in music packaging, that could raise interest in those products. Also, by
ensuring that a purchased piece of music is able to be played in the various ways that consumers
desire it to be, including at home, on mp3 players, and as ring tones, the value of music might
offset more consumer dollars.
21
5.0 Conclusion
This report outlines our investigation of the problem of music piracy and declining revenues
within the music industry from several perspectives. It discusses three claims made by labels,
independent artists, and academics and internet activists respectively, and attempts to determine
their feasibility and implications.
5.1 Claim 1
By modelling the simple relationship between consumers and pirates as professed by labels, we
see that, unless piracy is stopped, the music industry will collapse. However, mass online music
piracy has existed for almost a decade, and this prediction has not come to pass. Furthermore,
there exist presently a number of new business models within the music industry, many
integrating music with social networking, that are thriving. This suggests that the mental model
on which this model based is flawed. In turn, this implies that the business decisions made by
music labels also rest on uncertain ground.
Our model does not conclusively prove that this understanding is flawed; however, it does make
it clear that this mental model is extremely simplistic, with simplistic implications. This
emphasizes that the situation is far more complex than the statements of various labels and
industry groups would indicate they understand.
5.2 Claim 2
Using our model, we have been able to demonstrate that, when treated as free marketing, piracy
can actually enhance the revenues of artists by increasing their popularity and thus increasing
their revenues from the performance of music. This reveals that piracy is chiefly a problem for
labels — if claim two is indeed correct, it is in the artist’s interests to encourage the distribution of
their music by whatever means possible, including piracy.
One interesting implication of this is that piracy can act to level the playing field for small bands
and independent artists. No longer is massive marketing provided by a label required to promote
an act and compete for listener’s ears; artists with access to the internet are able to market
themselves. This will most probably be good for consumers, too, as it will result in increased
music diversity, increased availability of music, and more fair competition amongst artists
leading presumably to music of higher quality.
5.3 Claim 3
Our model of the third claim, being non-functional, is unable to give us firm conclusions.
Nonetheless, it suggests that declining revenues in the music industry are at least partly due to
factors other than piracy. In particular, music now competes with other entertainment media such
as video games and DVDs for the consumer’s dollar. Piracy may have an impact on sales, but it
is not the only factor, and most likely not the most important.
22
Given this observation, the current strategies employed by labels of aggressively prosecuting
consumers and asserting intellectual property rights may not be the best use of resources, and
may in fact be causing more harm than good by alienating the consumer base. Our model implies
that instead, labels should instead re-examine the quality and value that they offer to consumers,
in order to compete more effectively with other forms of entertainment, Additionally, labels
should continue to leverage synergies that exist between their products and those of other
entertainment providers, and should aggressively explore new distribution mechanisms and
internet driven business models.
Acknowledgments
Special thanks to Bill Harris for his guidance and feedback throughout this research project.
References
Bhattacharjee S, Gopal RD, Lertwachara K, Marsden JR. 2006a. Impact of legal threats on
online music sharing activity: an analysis of music industry legal actions. J Law Econ 49(1) : 91-
114.
Bhattacharjee S, Gopal RD, Lertwachara K, Marsden JR. 2006b. The effect of P2P file sharing
on music markets: a survival analysis of albums on ranking charts. SSRN Working paper.
Gayer A, Shy O. 2006. Publishers, artists, and copyright enforcement. Information Economics
and Policy 18 : 374-384.
Gopal RD, Sanders GL, Bhattacharjee S, Agrawal MK, Wagner SC. 2002. A behavioral model
of digital music piracy. SSRN working paper.
Gopal RD, Bhattacharjee S, Sanders GL. 2004. Do artists benefit from online music sharing?
SSRN working paper.
Hui KL, Png IPL. 2001. Piracy and the legitimate demand for recorded music. SSRN working
paper.
IFPI. Digital Music Report. 2009. Retrieved February 9, 2009, from
http://www. ifpi.org/content/library/DMR2009.pdf.
IFPI. Piracy Report. 2006. Retrieved February 9, 2009, from
http:/Awww.ifpi.org/content/library/piracy-report2006.pdf.
Krueger AB. 2004. The economics of real superstars: The market for rock concerts in the
material world. NBER working paper.
23
Lewis R. 2006, January 10. The Rolling Stones did make the biggest bang. LA Times.
Retrieved March 2, 2009, from http://articles.latimes.com/2006/jan/10/entertainment/et-
ultimate 10.
Liebowitz SJ. 2006. File sharing: Creative destruction or just plain destruction. J Law Econ 49(1)
21-28.
Matthews M. 2008, December. NPD: Behind the Numbers. Gamasutra. Retrieved February 21,
2009, from
http://www.gamasutra.com/view/feature/3906/npd_behind_the_numbers_december_.php.
Oberholzer F, Strumpf K. 2005. Effect of file sharing on record sales: An empirical analysis.
SSRN working paper.
Peitz M, Waelbroech P. 2004. The effect of Internet piracy on music sales: cross-section
evidence. Review of Economic Research on Copyright Issues 1(2) : 71-79.
Peitz M, Waelbroeck P. 2005, April 1. Why the music industry may gain from free downloading
— the role of sampling. Working Paper.
Rob R, Waldfogel J. 2006. Piracy on the high C's: Music downloading, sales displacement, and
social welfare in a sample of college students. J Law Econ 49 : 26-62.
Sandoval G. 2007. Study: Free beats fee for Radiohead's 'In Rainbows'. CNET. Retrieved
March 2, 2009, from http://news.cnet.com/8301-10784_3-9811013-7.html.
Sterman J. 2000. Business dynamics: systems thinking and modeling for a complex world.
McGraw-Hill: Boston MA.
Swartz O. 2009, February 25. Music executive ridiculed at Pirate Bay trial. Wired blog network.
Retrieved March 2, 2009, from
http://blog.wired.com/27bstroke6/2009/02/piratebaywednes.html.
Zentner A. 2006. Measuring the effect of online music piracy on music sales. J Law Econ 49 :
63-90.
24
