IViR (Institute for Information Law) Global Online Piracy Study July 2018 IViR -Institute for Information Law P.O. Box 15514, 1001 NA Amsterdam, The Netherlands https://www.ivir.nl/nl/ Institute for Information Law Joost Poort João Pedro Quintais Ecorys Martin van der Ende Anastasia Yagafarova Mathijs Hageraats The research project was inanced by Google. It has been conducted in full compliance with the Declaration of Scientiic Independence as formulated by the Royal Dutch Academy of Science. This Study, its Annexes, and the Legal Background Report are available for download at: https://www.ivir.nl/nl/ Global Online Piracy Study Global Online Piracy Study 1. 2. 2.1 2.2 2.3 3. 3.1 3.2 3.3 3.4 3.5 3.6 4. 4.1 4.2 4.3 5. 5.1 5.2 5.3 5.4 5.5 6. 6.1 6.2 6.
IVîR - Instîtute for Informatîon Law P.O. Box 15514, 1001 NA Amsterdam, The Netherands https://www.îvîr.n/n/
Instîtute for Informatîon LawJoost Poort João Pedro Quîntaîs
EcorysMartîn van der Ende Anastasîa Yagafarova Mathîjs Hageraats
The research project was inanced by Googe. It has been conducted în fu compîance wîth the Decaratîon of Scîentîic Independence as formuated by the Roya Dutch Academy of Scîence. Thîs Study, îts Annexes, and the Lega Background Report are avaîabe for downoad at: https://www.îvîr.n/n/
Global Online Piracy Study
Goba Onîne Pîracy Study
1.
2. 2.1 2.2 2.3
3. 3.1 3.2 3.3 3.4 3.5 3.6
4. 4.1 4.2 4.3
5. 5.1 5.2 5.3 5.4 5.5
6. 6.1 6.2 6.3
Contents
Hîghîghts
Executîve summary
Introductîon
A brîef hîstory of onîne pîracy research
Lega Background
The effect of onîne pîracy on saes
The effect of enforcement on onîne pîracy
Country statîstîcs and saes deveopments Country statîstîcs Musîc Fîm and vîdeo Books Games Concusîons
Survey sampe descrîptîves, weîghtîng and data ceanîng
Sampe composîtîon and recruîtment
Representatîveness and weîghtîng
Data ceanîng
Use of ega and îega content acquîsîtîon channes Musîc Fîm and vîdeo Books Games Concusîons
The effect of onîne pîracy on ega consumptîon
Instrumenta varîabe regressîon
Quasî-pane bockbuster ims
Rea pane
Lîterature
7
9
17
19 19 23 27
31 31 33 36 38 40 41
43 43 44 44
4747 54 59 63 68
7171 75 79
80
5
Goba Onîne Pîracy Study
Highlights
7
Thîs report deas wîth the acquîsîtîon and consumptîon ofmusîc, ilms, serîes, booksandgamesthrough the varîouslegalandîllegal channelsthat exîst nowadays, în a set of13 countrîesîn Europe (France, Germany, the Netherands, Poand, Spaîn, Sweden), the Amerîcas (Brazî, Canada) and Asîa (Hong Kong, Indonesîa, Japan, Thaîand). The îega channes studîed aredownloadîngandstreamîngfrom îega sources (încudîng vîa dedîcatedtechnîcal devîces), andstreamrîppîng.
The purposes are (î) to provîde factua înformatîon about thestate of authorîsed and unauthorîsedacquîsîtîon and consumptîon of content; (îî) to assess theunderlyîng motîvesand mechanîsms and thelînk wîth enforcement measuresand ega suppy; (îîî) to assess theeffect of onlîne pîracy on consumptîon from legal sources. At the core of the study îs aconsumer surveyamong neary35,000 respondents, încudîng over7,000 mînors, în 13 countrîes.
Legal analysîs Comparatîve legal researchwas performed on the basîs of questîonnaîres on the ega status of onîne copyrîght înfrîngement and enforcement, competed by ega experts în the 13 countrîes studîed. It was found that, despîte some ega uncertaînty, the majorîty of acts studîed are quaîied as dîrect copyrîght înfrîngement by users or gîve rîse to îabîîty for întermedîarîes. Moreover, ISPs are often subject to înjunctîons and dutîes of care even when they beneit from safe harbours. On the whoe, copyrîght hoders have a vast arsena of ega enforcement measures to depoy agaînst end users and ISPs. There îs a trend în many countrîes toward copyrîght enforcement through cîvî or admînîstratîve measures aîmed at bockîng websîtes that provîde access to înfrîngîng content. Notîces to înfrîngers and to patforms hostîng or înkîng to înfrîngîng content wîth the aîm of removîng/bockîng such content are îkewîse reguary used, the atter în the context of notîce-and-takedown systems. Crîmîna measures are ess popuar.
Stî, despîte the abundance of enforcement measures, theîr perceîved effectîveness îs uncertaîn. Therefore, ît îs questîonabe whether the answer to successfuy tackîng onîne copyrîght înfrîngement îes în addîtîona rîghts or enforcement measures, especîay îf these wî not ead to addîtîona revenue for copyrîght hoders and rîsk comîng înto conlîct wîth fundamenta rîghts of users and întermedîarîes. Instead, ît mîght be sensîbe to search for the answer to pîracy esewhere – în the provîsîon of affordabe and convenîent ega access to copyrîght-protected content.
Growîng markets Saes data for musîc, im and vîdeo, books and games revea that across a content types and formats, per capîta încome appears to be an împortant drîver of expendîtures. However, above an annua încome eve of € 30,000 per capîta, thîs reatîonshîp no onger seems to appy and natîona preferences domînate încome effects. Zoomîng în, ît îs cear that physîca saes are în contînuous decîne for amost every content type and în amost every country. Despîte the decîne în physîca saes, the încrease în dîgîta saes ed tonet growth for total recorded musîc, audîo-vîsual content, books and gamesbetween 2014 and 2017. Expendîtures on îve concerts and cînema vîsîts are growîng.
Survey outcomes The percentage of the Internet popuatîon consumîng content fromlegalsources varîes between 61% în France and 93% în Indonesîa. In most European countrîes, thîs percentage decreased somewhat between 2014 and 2017 – prîmarîy due to a decrease for physîca carrîers – whîe tota ega consumptîon voumes grew.
Consumîng content fromîllegal sources– onîne pîracy – îs most prevaent în the Internet popuatîons of Indonesîa, Thaîand and Brazî, foowed by Spaîn and Poand. As a percentage of tota popuatîon, Spaîn,
Goba Onîne Pîracy Study
8
Canada and Hong Kong are the top three countrîes for pîracy, whîe pîracy îs the east common în Germany, Japan and Indonesîa, the ast due to ow Internet penetratîon. Between 2014 and 2017, thenumber of pîrates decreased în all European countrîes except Germany.
The per capîtaconsumptîon volumesper ega and îega content channe that foow from the survey do not aways match these deveopments: for most countrîes and content types, an încrease în the per capîta voume of îega content consumed îs observed, despîte a decreasîng proportîon of the popuatîon engagîng în onîne pîracy. Thîs împîes that the îssue of pîracy îs graduay becomîng conined to a smaer group: fewer peope consume more on aggregate vîa îega channes.
It mîght be temptîng to argue that an încrease în the use of certaîn enforcement measures agaînst obvîousy îega patforms has contrîbuted to the decreasîng number of pîrates în Europe. However, a ack of evîdence concernîng the effectîveness of most enforcement measures and the strong înk between pîracy and the avaîabîîty and affordabîîty of content suggests otherwîse: at a country eve,onlîne pîracy correlates remarkably strongly wîth a lack of purchasîng power. Hîgher per capîta încome coîncîdes wîth a ower number of pîrates per ega users.
Moreover, pîrates and ega users are argey the same peope: demographîcay, pîrates resembe ega users quîte cosey, athough on average they tend to be somewhat younger and more often mae. More împortanty, for each content type and country,95% or more of pîrates also consume content legallyand theîrmedîan legal consumptîonîs typîcaytwîcethat ofnon-pîratîng legal users.
Dîsplacement of legal sales Thîs study conirms earîer studîes în indîng statîstîca evîdence thatîllegal consumptîonof musîc, books and gamesdîsplaces legal consumptîon. However, the dîspacement coeficîents are surrounded wîth substantîa uncertaînty. Separatîng these resuts between mînors and aduts suggests that dîspacement occurs for aduts and not for mînors.
The resuts formusîcîndîcate that îega consumptîon prîmarîy dîspaces ega downoads and physîca carrîers. The effect on streamîng îs not statîstîcay sîgnîicant. For îve concerts and musîc festîvas, aposîtîvesampîng effect îs found. Foraudîo-vîsualcontent, no such sampîng seems to occur for the cînema, whîch suffers from statîstîcay sîgnîicant dîspacement, as do dîgîta streams. No sîgnîicant effects are found for physîca purchases and dîgîta downoads. For rentas, a margînay sîgnîicant posîtîve coeficîent îs found. Forbooks,the resuts are contrary to those for musîc and audîo-vîsua în the sense that arge and statîstîcay sîgnîicant dîspacement rates are found for books bought în prînt and borrowed from the îbrary. These dîspacement rates may be overstated by peope who have shîfted from consumîng prînt books to dîgîta and others who have not. Forgames, the effect found for free games îs partîcuary hîgh, but the coeficîents found for the other channes are aso statîstîcay sîgnîicant. Just îke for books, the arge coeficîent for free games may be overstated.
Usîng tîme-structured data forblockbuster ilms, an average dîspacement rate was found of–0.46of irst ega vîews by irst îega vîews. Thîs effect îs smaer în Japan and the Netherands and arger în Thaîand and Brazî. The argest effect occurs on cînema vîsîts. From these estîmatîons, ît îs possîbe to cacuate an upper bound for the reatîvesales lossof tota im vîews per channe and per country. Overa, a maxîmum of about4.1%of a ega bockbuster vîews îs dîspaced by îega vîews.
An anaysîs of îndîvîdua changes în consumptîon for respondents în sîx EU countrîesbetween 2014 and 2017reveassîgnîicantly posîtîvecorreatîons. Apparenty, substîtutîon effects – ‘Sha I buy or sha I pîrate?’ – occur on the spot. Over a onger tîme span, împrovements în the avaîabîîty from ega channes are domînant and changes în persona preferences affect ega and îega consumptîon aîke.
Goba Onîne Pîracy Study
Executive summary
9
A changîng landscape for the acquîsîtîon and consumptîon of copyrîghted content The ways în whîch consumers acquîre and experîence musîc, ims, serîes, books and games have changed radîcay over the ast 20 years. Whereas such copyrîght-protected content was acquîred prîmarîy în the form of physîca carrîers such as cds, dvds and prînt books up to the turn of the century, nowadays an încreasîng amount of purey dîgîta content îs acquîred and consumed vîa the Internet. And în recent years, ît îs possîbe to dîscern a second shîft în consumptîon of content, from downoadîng from patforms such as îTunes, to streamîng from Netlîx, Spotîfy and the îke: from ownershîp to access.
Aong wîth, and sometîmes even ahead of, the authorîsed dîstrîbutîon of content, unauthorîsed onîne content dîstrîbutîon – commony referred to as ‘onîne pîracy’ – has foowed the same evoutîon: from physîca carrîers, such as counterfeît cds and dvds, vîa downoads from Napster and îts successors such as The Pîrate Bay andcyberlockerssuch as Megaupoad, to streamîng vîa patforms such as Popcorn Tîme. In some cases,unauthorîsed streamîngîs done through dedîcated technîca devîces/set-top boxes wîth pre-înstaed înks to unauthorîsed content patforms. A reated use of copyrîght-protected materîa that îs generay aso vîewed as unauthorîsed îsstreamrîppîng, whereby software toos, browser pugîns or specîa websîtes are used to store musîc or audîo-vîsua content, such as YouTube vîdeos, oflîne for ater repay (în vîoatîon of YouTube’s terms of servîce). These varîous channes coexîst today and many consumers use a mîx of them dependîng on the occasîon.
Research questîons and scope Thîs report deas wîth the acquîsîtîon and consumptîon ofmusîc, ilms, serîes, books and gamesthrough the varîous ega and îega channes that exîst nowadays, în a set of countrîes across the gobe (see Tabe 1).
Tabe 1 Countrîes studîed France Germany The Netherands EuropePoand Spaîn Sweden Unîted Kîngdom Brazî Amerîcas Canada Hong Kong Indonesîa Asîa Japan Thaîand
The purpose of the study îs threefod:
FRA DEU NLD POL ESP SWE GBR BRA CAN HKG IDN JPN THA
1. To provîde factua înformatîon about the state ofauthorîsed and unauthorîsedacquîsîtîon and consumptîon of these content types. 2. To assess the underyîng motîves and mechanîsms and the înk wîth enforcement measures and ega suppy. 3. To assess the effect of onîne pîracy on consumptîon from ega sources.
Goba Onîne Pîracy Study
10
To these ends, thîs study combînes dîfferent sources and research methods. At the core of the study îs a consumer survey among neary 35,000 respondents, încudîng over 7,000 mînors (aged 14 to 17), în 13 countrîes. In addîtîon, comparatîve ega research was performed on the basîs of questîonnaîres on the ega status of onîne copyrîght înfrîngement and enforcement, competed by ega experts în the 13 countrîes studîed. A separateLegal Background Reportcontaîns thîs ega anaysîs and the fu country reports that underîe ît, but a brîef overvîew îs provîded beow and în the body of the maîn report.
Legal Background Comparatîve ega research was performed on the basîs of questîonnaîres on the ega status of onîne copyrîght înfrîngement and enforcement, competed by ega experts în the 13 countrîes studîed. It was found that, despîte some ega uncertaînty, the majorîty of acts studîed are quaîied as dîrect copyrîght înfrîngement by users or gîve rîse to îabîîty for întermedîarîes. Moreover, ISPs are often subject to înjunctîons and dutîes of care even when they beneit from safe harbours. On the whoe, copyrîght hoders have a vast arsena of ega enforcement measures to depoy agaînst end users and ISPs. There îs a trend în many countrîes toward copyrîght enforcement through cîvî or admînîstratîve measures aîmed at bockîng websîtes that provîde access to înfrîngîng content. Notîces to înfrîngers and to patforms hostîng or înkîng to înfrîngîng content wîth the aîm of removîng/bockîng such content are îkewîse reguary used, the atter în the context of notîce-and-takedown systems. Crîmîna measures are ess popuar.
Stî, despîte the abundance of enforcement measures, theîr perceîved effectîveness îs uncertaîn. Therefore, ît îs questîonabe whether the answer to successfuy tackîng onîne copyrîght înfrîngement îes în addîtîona rîghts or enforcement measures, especîay îf these wî not ead to addîtîona revenue for copyrîght hoders and rîsk comîng înto conlîct wîth fundamenta rîghts of users and întermedîarîes. Instead, ît mîght be sensîbe to search for the answer to pîracy esewhere – în the provîsîon of affordabe and convenîent ega access to copyrîght-protected content.
The elusîve effect of onlîne pîracy on sales When onîne ie sharîng took off around the turn of the century, the recorded-musîc îndustry and, ater, the im îndustry were quîck to bame ît for ost revenues. Indeed, ît îs argey undîsputed that 1999 was a turnîng poînt for revenues from goba recorded musîc saes and that rea revenues from physîca and dîgîta recorded musîc saes decîned for many years thereafter, ony to start încreasîng agaîn a few years ago. Lîkewîse, goba revenues from dvd saes and rentas have decîned for many years sînce the turn of the century.
Nevertheess, answerîng the empîrîca questîon of the effect of unauthorîsed onîne content consumptîon on ega saes has proven to be cumbersome. A majorîty of the academîc îterature on the matter inds a negatîve net effect of îega consumptîon on ega saes. However, there are reatîvey few studîes on games and books, and a meta-anaysîs of the îterature inds the evîdence însuficîent to concude wîth certaînty that pîracy harms ega saes.
There are severa reasons that there îs no consensus on what may seem so obvîous at irst gance: the înk between onîne pîracy and ega saes. Fîrst, there are severa opposîng înteractîons between pîracy and ega consumptîon, some of whîch have a negatîve împact on saes, some posîtîve and some neutra. These varîous potentîa mechanîsms are summarîsed în Tabe 2. The most promînent posîtîve effect îs known as thesamplîng effect: consumers are întroduced to new musîc, actors and genres, and thîs creates new demand. Onîne pîracy may aso enhance the demand forcomplementary productssuch as îve concerts and merchandîse. On the downsîde, the most promînent effect îs obvîousysubstîtutîon: a consumer refraîns from buyîng specîic content egay after havîng acquîred or consumed ît from an îega source. Aso, pîracy may dîspace ega consumptîon vîa competîtîon for peope’s tîme budget: one cannot watch one im from a ega source and one from an îega source at the same tîme. Fînay, neutra effects occur, for înstance, when ie sharîng meets the demand of consumers wîth însuficîent wîîngness to pay.
Goba Onîne Pîracy Study
Tabe 2
Possîbe effects of onîne pîracy on the purchase of ega content
Posîtîve
Neutral
Negatîve
+ It întroduces consumers to musîc, ims, books and games (and to artîsts, authors and genres), thus creatîng new demand. Thîs îs known as the sampîng effect. + It aows consumers to poo theîr demand, resutîng în încreased demand. + It enhances wîîngness to pay and demand for concerts and reated merchandîse (compementary demand). + It enhances the popuarîty of products, boostîng demand for ega suppy (network effect).
• It meets the demand of consumers who are not, or not suficîenty, wîîng to pay and subsequenty are not served by ega suppy. • It meets a demand for products that are not offered egay.
- It substîtutes for the purchase of content or cînema vîsîts (substîtutîon effect). - It resuts în the deferred purchase of content at a ower prîce than the prîce at aunch. - Sampîng resuts în saes dîspacement as a resut of fewer bad purchases. - It substîtutes ega consumptîon vîa consumers’ tîme budget.
11
The reatîve strength of these înteractîons îs îkey to dîffer between content types and channes. Even wîthîn content categorîes, dîfferences can be expected between more recent and popuar content, on the one hand, and oder or nîche content, on the other hand. Aso, the average net effect may have changed over
tîme, wîth the advance of fast broadband connectîons for consumers and the înnovatîon by the content
îndustrîes. Nowadays, a technoogîca mîsmatch between suppy and demand îs a much ess îkey motîvatîon
for onîne pîracy than ît was a decade ago.
On top of these subtetîes, estîmatîng the effect of onîne pîracy on saes faces methodoogîca chaenges
that have to do wîth the fact that for many consumers, ega consumptîon and pîracy go hand în hand.
Thîs îs îustrated în thîs report by the demographîc descrîptîon of pîrates and ega users, who are very
much aîke. Moreover, as a consequence of underyîng îndîvîdua preferences, pîrates are much more îkey
to be ega users of each content type than are non-pîrates, and the medîan ega consumptîon of pîrates
îs typîcay twîce that of non-pîrates. As a consequence, a posîtîve correatîon between pîracy and ega
consumptîon can be expected whîch shoud not be hastîy înterpreted as a causa reatîonshîp.
Development of legal sales Anaysîs of saes data for musîc, im and vîdeo, books and games reveas that across a content types and formats, per capîta încome appears to be an împortant drîver of per capîta expendîture. However, above an annua încome eve of €30,000 per capîta, thîs reatîonshîp no onger seems to appy and natîona
preferences domînate încome effects. Thîs îs îustrated în Fîgure 3.
Zoomîng înto these saes per content type, ît îs cear that physîca saes are în constant decîne for amost
every content type and în amost every country, most notaby for audîo-vîsua content. Books are reatîvey
resîîent to dîgîtîsatîon and prînt books stî domînate în a countrîes. Dîgîta saes have grown amost
everywhere over the past three years. For musîc, dîgîta streamîng grew strongy at the expense of dîgîta
downoads. For audîo-vîsua content, svod servîces such as Netlîx are becomîng the domînant mode.
Despîte the decîne în physîca saes, the încrease în dîgîta saes ed to net growth for tota recorded musîc,
audîo-vîsua content, books and games between 2014 and 2017. Expendîtures on îve concerts and cînema
vîsîts are growîng amost everywhere. In the musîc category, îve concerts generate revenues comparabe to
those for recorded musîc, except în Japan. In the audîo-vîsua category, ims generate not much ess revenue
than physîca and dîgîta recorded content combîned.