No Meaningful Impact? Outline The Effect of Pre-Release Movie Piracy on Box-Office Sales • Understanding Movie Piracy • Piracy: In dus try an d Acad emi c P erspecti ves • Promotional potential for piracy? Liyy,e Ma, Param Vir Sing g,h, Michael D Smith, and Alan Montg gyomery Carnegie Mellon University • CliConclusions

e-mail: [email protected] web: http: //www.an drew.cmu.e du /user /a lm 3

Hong Kong; 5 July 2010 (7 -8pm) University of Hong Kong Expert Address Room 613 , United Centre , Admiralty , HK 1 2

Pre-Release Piracy: Studio’s Response

• Pirated copies of Lions Gate Entertainment Corp’s horror sequel “Hostel: Part II” The Problem of Movie Piracy appeared a month before theatrical release on 8 June 2008 • In response, Lion Gate’s PresidObdent Tom Ortenberg said “It’s distressing and disappointing, but it will have no meaningful impact on the box office.”

3 4 The Problem Pre-Release Piracy: Director’s onResponse line before the release, and is really hurting us, especially internationally. Piracy will be the death of the film industry, as it killed the music industry, “However, piracy has become worse than ever now,like Spider-man and a stolen 3, workprintit really hurts (with films unfinished like mine, musicwhich no, no soundhave sound effectsfar effectsless and of no , anand VFX) advertising no leaked VFX) out leaked and out production budget. Not only that, critics have

pirated copy, which is inex andcritics while while I it makes haveit makes actually a smaller a smaller dentkn in hugedent moviesin huge movies while I wouldn't dignify them by mentioning them cusable. Some of these and other filmmakers, andown they for will a fewno longeryears, haveand actuallyany access been been REVIEWING toREVIEWING any of my the thefilms” film basedfilm based off the off the

byname name I know, I know who theywho are they as do are the, studios as do the studios,

Director Eli Roth (MySpace blog)

5

The Problem

10 days, Screener 9 days, Cam 5 days, Telesync 3 months, Workprint 2 days, Cam

27 days, Screener 1 month, Screener 2 days, Screener

6 month, Screener 1 month, Workprint 1 month, Screener Movie Piracy will Increase Proposed Solutions

• The initial emphasis has been on music piracy • Security: Metal detectors, Camcording – Small download sizes amenable to download via dialup felony law • However thfhese factors mean movie piracy will • Technological:Watermarking: Watermarking grow: • Managerial: Limit promotional distribution – Increasing network bandwidth • Legal: Lawsuits against movie pirates – Continuing growth in Internet penetration of • Policy: IP Policy Tsar consumer markets – Merger of media, television, and personal computing 9

US Government shut down The Questions nine sites on 30June2010 How much does pre-release movie piracy harm b ox offi ce revenue?

How does this vary over movie lifecycle?

How does this vary by piracy quality?

11 Industry Arguments against Piracy

• Companiesdecrease in now worldwide have to piracy “compete” rate overagainst 4 years pirated would copies of their own IP Piracy • Will produce less and reduce investments • Losseconomic of tax revenuesgrowth and jobs

Industryyp and Academic Perspectives BSA/IDC Economic Economic Impact Impact Study Study (2005) (2005) found found 10% 10%

– Add 1 1 5 million.5 million new new jobs $64b jobs in, taxes$64b and in taxes $400b and in $400b in

– Yield larger benefit for countries with higher piracy rates

13 14

Movie versus Music Sales: Movie versus Music Sales Per Capita

$16,000.00 $60.00

$55.00 $14,000.00 $50.00

$12,000.00 $45.00

$40.00 $10,000.00 Movie Movie $35.00 Music Music $8,000.00 $30.00

$25.00 $6,000.00 $20.00

$4,000.00 $15.00 1990 1995 2000 2005 2010 1990 1995 2000 2005 2010

15 16 Movie versus Music Sales: Product Lifecycle for Music Per Capita and Real $

$75.00 $70.00 $65.00 $60.00 $55.00

$50.00 Movie $45.00 Music $40.00 $35.00 $30.00 $25.00 1990 1995 2000 2005 2010

17 18

Literature: Literature: Effects of Piracy Modeling Movie Box Office

• Impact of Piracy on Music Demand • What drives box office success? – Oberholzer and Strumpf 2007, Liebowitz 2007, Peitz and – Script (Eliashberg et al . 2007) Waelbroeck 2004, Hui and Png 2003, Zentner 2006, Rob and – Advertising (Rennhoff and Wilber 2008) Waldfogel 2006, … – Stars (Elberse 2007) • Impact of Piracy on Video Demand – Distribution (Swami et al. 1999) – Piracy reduces demand for DVDs (Rob and Waldfogel 2007) – Critics’ reviews (Eliashberg and Shugan 1997) and theat rical re ven u e (Danaher and Waldfogel 2008) – User reviews (Dellarocas et al. 2007, Duan et al. 2008) – No impact on catalog DVD sales (Smith and Telang 2009) • Consumer heterogeneity / product differentiation – Digital distribution substitute for piracy (Danaher et. al. 2009) (Eat on an d Lipsey 1989) – Piracy hurts “bad” movies (Chellappa and Shivendu 2005) – Most (all?) piracy papers in environment with both (low quality?) piracy and (high quality?) legitimate product • Impact of Piracy on Software Demand – Pirates may subsequently purchase (Chellappa et al. 2006) Theory: Impact? Hypotheses

House Committee on Foreign Affairs • H1: Pre-release piracy reduces movie box-office ↓ Substitution sales. “During 2007, the entertainment industry generated a trade surplus of $13.6 billion, imagine what those numbers would be if • H2: Pre-release piracy has a higher impact on we could reign in piracy .” Rep Howard. Howard Berman Berman Chairman, Chairman, earlier periods of movie-box office sales than later periods. • H3: Higher quality pre -release piracy has a lower ↓↑ Word of Mouth impact on box-office sales than lower quality leaks “If people see this movie and don’t like it, and they tell their do. friends, and their friends blog about it, and it just spreads throughout the blogosphere, there are a lot of people that don’t even get near a pirated copy of this film, who don’t go to see [[]the movie in theaters] because of this leak” Steve Zeitchik, The Hollywood Reporter, discussing Wolverine Leak.

Empirical Model Empirical Model

• Sawhney and Eliashberg (1996), Krider and • Sawhney and Eliashberg (1996), Krider and Weinberg (1998) Weinberg (1998) – where… – where… Data: Data: Movie Characteristics Piracy/Quality • All movies released Jan 2006 - Jan 2009 • IMDB: budget, star appeal, user rating, release date (USA) • BoxOfficeMojo: weekly box office, distributor, genre, MPAA rating, director appeal, screens, • Yahoo Movies: Movies: critic rating

Data: Data: Piracy Piracy

• Pre-release piracy: =1 if available before US theatrical release • Audio Quality: Avg. audio quality from vcdq • Video Quality: Quality: Avg. video quality from vcdq Results: Final Dataset Homogeneous Decay

• 194 movies, 21 with pre-release piracy

• inc. movie-level random effects • ρ<0 → total revenue loss = 18.2% • τ<0 → faster decay

Results: Results: Heterogeneous Decay Piracy Quality

• inc. movie-level random effects • inc. movie-level • ρ<0 → lower market random effects potential (-14.8%) • ρ2>0 → higher quality • τ<0 → faster decay piracy, relatively higher market potential • τ~0 → no difference in decay rate Endogeneity Results

• Propensity score matching, re-estimation on matched data – Specification 1: Production budget, number of screens predict piracy – Specification 2: Production budget, star & director appeal predict piracy

Conjecture

• Our results suggest that pre-release piracy reduces Promotional Potential for sales. However, the magnitude suggests that under the right conditions positive promotional effects could Piracy? offset n egativ e eff ects of canni baliz ati on. • Develop a Markovian model with diffusion element to model movie sales.

35 36 Modeling Consumer Transitions from Undecided Viewership

Positive Spreader Consider Inactive Negative UdUndec iddided Spreader

RjRejec t

37 38

Transitions from Consider Transition from Spreader

39 40 Low Quality/Low WOM Low Quality/High WOM

41 42

High Quality/Low WOM High Quality/High WOM

43 44 Findings

Under the right conditions movie piracy could be beneficial: 1. The underlying quality of the movie is high Conclusions 2. The quality of the movie is not well signaled through traditional promotions 3. Viewing of the movie is not greatly diminished by viewers of the pirated copy 4. Word-of-mouth effects are strong

45 46

Limitations & Summary Future Work

• Pre-release piracy associated with… • No data on intensity of pre-release piracy – 14% lower box office • UbltditUnable to disentangl e sub btittistitution and word -of- – Harm disproportionately on opening weeks mouth effects • Higher quality pre -release piracy associated with • Small dataset make it difficult to separate effects lower decline in box office than other pre-release across genres, etc. ppyiracy