Thearical Feaure Film Trade in he Unied Saes, Europe, and Japan since he 1950s: An Empirical Sudy of he Home Marke Effec David Waerman Dep. of Telecommunicaions Indiana Universiy 1229 E. 7 h S. Bloomingon, IN 47401 waerman@indiana.edu Sang-Woo Lee Division of Telecom and Broadcasing Policy Korea Informaion Sociey Developmen Insiue Juam-Dong, Kwachun, Kyunggi-Do, 427-710, Korea Leesw726@kisdi.re.kr Revised Augus 19, 2005 We are graeful o paricipans in he TPRC Conference on Communicaion, Informaion, and Inerne Policy and in seminars a Indiana Universiy and Norhwesern Universiy for commens. We are especially indebed o Krishna P. Jayakar and Weiing Lu, who made valuable conribuions o he compilaion and analysis of he daabase for his sudy. 1
Absrac Thearical Feaure Film Trade in he Unied Saes, Europe, and Japan since he 1950s: An Empirical Sudy of he Home Marke Effec We es a home marke model of inernaional rade in media producs using a movie indusry daabase covering six major counries (he Unied Saes, Japan, Germany, Ialy, France, and he U.K) over he 1950-2003 ime period. In suppor of he model, we find a consisenly posiive relaionship over ime beween domesic heaer box-office marke shares and various measures of domesic movie spending or domesic movie aendance. Based on hese resuls, we aribue declining domesic film producion indusries in Europe and Japan afer abou he 1970s, along wih growing dominance of he world film marke by he Unied Saes, o a relaively rapid growh of domesic consumer spending on movies in he U.S. 2
I. Inroducion The apparen endency for relaively large and wealhy counries, noably he Unied Saes, o dominae world rade in elecronic media producs has been widely aribued o a home marke effec. Tha is, counries having relaively high domesic consumer spending on a given media produc should end o have relaively high expors of ha produc o oher naions. In his paper, we es he home marke heory in an aemp o explain long erm shifs in he balance of hearical feaure film rade among he Unied Saes, Japan, and four European counries from 1950 o 2003. As we deail below, he early par of his period, generally from he 1950s o he 1970s, was one of relaive prosperiy and high heaer box office marke shares for domesic European and many oher film producion indusries ouside he Unied Saes. The ables hen urned; while film producion in he Unied Saes seadily expanded unil a leas he early 1990s, indigenous producion indusries in Europe, Japan, and many oher counries declined, many of hem becoming heavily dependen on sae subsidy. In Wesern Europe for example, European produced films as a whole were repored o earn 60% of he box office in 1968, bu only 27% in 2003; he marke share of American produced films in Europe rose over his same period from 35% o 72%. 1 We seek o undersand hese rends over ime using he home marke model. We begin in Secion II below wih an exposiion of he home marke heory, along wih a brief review of previous work. In Secion III, we discuss he consrucion of our daabase and se ou basic descripive relaionships ha emerge from i. Specificaion of our economeric models and resuls follow in Secion IV and V, followed in Secion VI by a discussion and conclusion. 1 European Union (1994); European Audiovisual Observaory (2004) 3
II. Theory and Previous Research The basic oulines of a home marke heory o explain he hisorical endency for he U.S. or oher larger counries o dominae inernaional rade in movies, elevision, or oher producs was advanced by Pool (1977). More sysemaic economic models based on he home marke effec were advanced by Hoskins & Mirus (1988); Waerman (1988); and Wildman & Siwek (1987, 1988). The logic of hese media-specific explanaions is imbedded in more general heories published in he inernaional rade lieraure. A prevalence of home marke effecs ha is, a endency for counries wih relaively high home demand o accoun for larger proporions of expors--was firs noed by Linder (1961), and hen given a rigorous heoreical basis in he form of new economic geography models. Krugman (1980) and Helpman and Krugman (1985) showed ha imperfec compeiion in differeniaed produc indusries ha exhibi economies of scale would lead o a home marke effec if ranspor coss are significan. Tha is, such firms will end o locae in relaively large markes in order o minimize ranspor coss while also bes realizing scale economies. More recen auhors have exended hese models o show circumsance under which counry-o-counry demand differences or differen assumpions abou coss and compeiion can lead o similar resuls (Feensra, Markusen & Rose, 2001; Head, Mayer & Reis, 2002; Davis & Weinsein, 2003). Weder (1996) offers a axonomy of demand condiions, including variaions in income, ases, and climae, ha can also generae home marke effecs. The media produc specific, home marke explanaions for media produc rade dominance by he U.S. or oher large counries offered by he above cied auhors provide he heoreical basis for our empirical analysis o follow. According o hese 4
heories, relaively large and/or wealhy counries end naurally o have a large consumer demand base for movies. Unless inernal or exernal economies of scale in movie producion are exremely high, however, ha advanage would no in iself lead he U.S. or oher major counries o be dominan producers and exporers or movies. Because hey have minimal ranspor coss, movie producion could poenially ake place anywhere. Addiional assumpions abou demand and echnology lead o a large counry effec in he movie case. One assumpion is ha here is a home bias in consumpion, which hese auhors label as a culural preference or culural discoun. Tha is, all oher hings equal, audiences are assumed o prefer movies ha are produced in heir naive languages or ha reflec heir own culural values. Thus, domesically produced movies, using naive acors and acresses and oher agens of producion, are preferred by ha counry s audiences over movies having foreign origins, oher hings equal. The implicaion of he culural discoun assumpion is ha producers in high domesic demand counries effecively have larger poenial markes. In a wo counry world, for example, he large counry effecively has a large slice of a large marke, and a small slice of a small marke, while he small counry has a small slice of a large marke and a large slice of a small marke. The nex sep leading o a large counry effec involves he high seup (ie, producion) cos, low marginal disribuion cos characerisics of media producs, and he endogenous naure of hose seup coss. As Shaked and Suon (1987) and Suon (1991) have shown in a more general conex, indusries having relaively large poenial markes ha are faced wih hese cos condiions will end o produce larger variey and higher qualiy producs. I hen follows ha wih free rade, he movies produced in counries having relaively high levels of consumer movie spending, 5
like he U.S., will end o dominae rade and o have relaively large box-office marke shares in boh heir home and foreign markes. The culural discoun assumpion upon which he media-specific explanaion of large counry dominance depends seems o have widespread, hough anecdoal, empirical suppor (e.g., Tracey, 1985; Mills, 1985; Wildman, 1995). A similar home bias in consumpion of oher producs has been commonly observed (Wolf, 2000). Noable is a growing lieraure aemping o explain a srong home bias in he purchase of securiies (French & Poerba, 1991); among explanaions offered are familiariy, language and culure (Grinbla & Keloharju, 2001). Several empirical ess of home marke models published in he broader inernaional rade lieraure have been made using manufacuring indusry daa, wih generally supporive, hough mixed resuls. Lundback and Torsensson (1998) and Davis and Weinsein (1999, 2003) generally find posiive evidence for home marke effecs across counries or across producs. Using U.S.-Canadian daa, Head and Reis (2001) find suppor for eiher an increasing reurns model or a consan reurns model, bu inerpre he weigh of evidence o suppor he laer, suggesing ha home marke effecs do no predominae. Among empirical ess of media-specific rade models, Marvasi (1994) repored ha among oher facors, counry size had a significan effec on expors of a broad range of culural producs. In a laer empirical sudy of facors influencing moion picure rade, Marvasi (2000) found suppor for a facor endowmen heory of rade and some evidence of counry size effecs. Jayakar and Waerman (2000) repored crosssecional evidence in he movie case ha larger size, GDP, and spending on movies were significanly relaed o domesic box office share. Oh s (2001) analysis of 1988 6
hrough 1994 box office daa from 14 counries found significan relaionships beween he box office marke share of domesic films and gross domesic produc (GDP), box office revenue, and some measures of culural disance from he Unied Saes. Waerman and Jayakar (2000) and Lee (2002) primarily used graphical analysis o relae rends over ime in movie box-office marke shares o consumer movie spending in Ialy and he U.S. and in Japan and he U.S., respecively, producing resuls ha are consisen wih he home marke inerpreaion for American dominance. Waerman (2005) and Waerman & Lee (2005) repor graphical and saisical analysis ha aemp o explain he rend oward U.S. dominance of he box office in five counries: Japan, Germany, Unied Kingdom, France and Ialy, since he 1970s, and how he movie spending rends are relae o diffusion raes of elevision and video media. III. Daa and Descripive Trends The empirical analysis of his paper covers a broader range of movie rade among six counries: he Unied Saes, Japan, Germany, UK, France and Ialy. We use a home marke model o explain domesic box office marke shares in hese six counries and Japan over he 1950 o 2003 ime period, as a funcion of rends in he levels of domesic movie spending. Ideally, of course, we would include a larger variey of counries, bu long erm ime series daa were generally unavailable excep for his counry group. In 2003, however, he six subjec counries of our sudy garnered 73% of world box office revenues, wih he Unied Saes alone accouning for 43% of ha oal. 2 Thus, he six subjec counries accouned for he bulk of he enire world marke for movies. 2 Screen Diges (2004, Sepember), pp. 270-271. 7
Marke Shares Table 1 repors domesic hearical box office marke shares over ime ha we were able o assemble for each of he six subjec counries from 1950 o 2003. These daa originae from quasi-governmenal film organizaions, he European Audiovisual Observaory, and rade associaions. The daa indicae a general prosperiy of naional (domesically produced) films unil abou he mid-1970s in he four European counries, bu hen seady decline ino he 1990s. Japan shows he same general paern, even hough he srengh of is domesic producion indusry began o ebb earlier. Over ime, Japan and France have susained he sronges domesic film indusries excep for he U.S., even hough he marke shares of domesic movies have recenly fallen below half in all five of hese counries. The U.S. daa are less complee, bu hey indicae very high domesic marke shares since a leas he early 1980s, wih a sligh decline eviden in he 1990s. 3 Daa repored for he marke shares of U.S films in he oher five counries is less complee bu demonsraes he large exen o which he decline in domesic film producion has been mirrored by a rise in he shares of U.S. films. In mos cases, he U.S. accouns for he overwhelming proporion of impored films in hese counries. 4 Consumer Spending We define primary movie spending o consis of oal consumer spending on he hree main media heaers, pay TV (premium subscripion movie channels, plus pay-per-view movie services), plus videocassees and DVD ha primarily exhibi 3 Alhough sysemaic earlier daa are no available, saisics published in Variey and repored by Guback (1969) sugges ha he marke shares of foreign films in he Unied Saes were subsanially higher in he 1958-64 period han afer 1980. See also Waerman, Lee, and Lu (2005). 4 Lange & Newman (2003), p. 20. 8
hearical movies wih direc paymen suppor. In 2002, hese spending caegories accouned for abou 88% of all U.S. disribuor revenues from he domesic exhibiion of hearical movies (25% heaers, 54% video, 9% pay elevision). 5 The remaining 12% came from sales o basic cable and broadcas neworks and saions and are no accouned for by our analysis. A difficuly of adding he primary spending caegories ogeher is ha differen wholesale markup raes may apply and hese revenues may be received by disribuors a differen poins in ime. Hisorical daa on consumer spending for hese media were assembled from a variey of sources and are shown as a fracion of GDP for each of he six counries in Table 2. As hese daa show, he U.S. generally railed he oher 5 counries in movie spending in early years, bu in abou he mid-1970s, he U.S. ook a lead ha wih lile excepion, i has mainained since. A rough correspondence of hese long erm rends wih he long erm shifs in box-office marke shares suggess ha hey may be saisically relaed, a possibiliy we es formally below. IV. Empirical Models In heir wo counry comparaive saisic model, Wildman and Siwek (1988) show ha if producers are monopolisically compeiive wih free enry, and here is free rade, hen under cerain assumpions abou demand and culural discoun facors, ha higher movie spending by consumers in counry A implies ha producers in A will offer a greaer number of movies and make higher invesmens in hem. Tha resul furher implies ha afer rade akes place, domesic marke shares in counry A will be greaer han domesic shares in counry B. Counry A s movies, ha is, will be relaively 5 Paul Kagan Associaes (2002, Augus), p. 4 9
dominan in inernaional rade. I furher follows ha he raio of domesic movie spending in counry A o ha of counry B should be posiively correlaed o he raio of oal revenues for domesically produced movies, or equivalenly, he raio of domesic marke shares, in counries A and B. If he home marke heory is correc, and his logic can be generalized o a muli-counry world, hen changes in he marke shares of domesic vs. U.S. movies should be posiively explained by shifs over ime in he relaive amouns of home marke consumer spending on movies in hose counries. Ouside he Unied Saes, governmen regulaion or influence resrics he proporion of American programs on mos publicly and privaely operaed elevision sysems. In he heaer box-office (as well as video) cases, however, governmen quoas or oher resricions have been largely absen or ineffecual during pos-world War II period, so we can be reasonably confiden ha hose rends reflec marke forces. 6 We specify five basic saisical models as follows: ( 1) ε XSHAR Ei, = α i + β ( BOXSPENDSHAREi, 1) + i, ( 2) ε XSHAR Ei, = α i + β ( TOTALSPENDSHAREi, 1) + i, ( 3) ε XSHAR Ei, = α i + β ( BOXADMISSIONSHAREi, 1) + i, ( 4) ε XSHAR Ei, = α i + β ( BOXPERGDPSHAREi, 1) + i, (5) XSHAR Ei, = α i + β ( TOTALPERGDPSHAREi, 1) + ε i, where i indicaes counries. Variable definiions are as follows: (All financial variables are specified a prevailing exchange raes.) 6 Guback (1969); Schiller (1969, 1992); Seagrave (1997); Thompson (1985) 10
Variable i BOXSPENDSHARE i TOTALSPENDSHARE i BOXADMISSIONSHARE i BOXPERGDPSHARE i TOTALPERGDPSHARE i Definiion he box-office marke share of counry i s domesically produced movies in counry i. Counry i s share of aggregae heaer boxoffice spending by he combined group of six counries in he model Counry i s share of aggregae heaer boxoffice, premium pay TV, pay-per-view, video renals and video sales spending by he combined group of six counries in he model Counry i s share of aggregae heaer admissions accouned for by he combined group of six counries in he model The raio of counry i s heaer box-office spending as a % of GDP o ha for he combined group of six counries in he model.. The raio of counry i s heaer box-office, premium pay TV, pay-per-view, video renals and video sales spending as a % of GDP o ha for he combined group of six counries in he model.. These five models es he home marke heory in wo basic ways. Models (1) (3) reflec he hypohesis ha he marke share of domesic movies in counry i will be posiively explained by is share of aggregae primary movie spending in all six counries combined. Elaboraing on Model (1), for example, he marke share in counry i of movies produced in i is assumed o be linearly relaed o counry i s share of consumer movie spending on movies in all six counries combined. Comparably for Models (2) and (3) The models hus incorporae he implici assumpion ha all five 11
non-i counries can freely rade wih couny i. Essenially, he five oher counries are reaed as a single eniy in hese models. These models hus ignore he influence of movies produced ouside of he six counries in he daa se. As noed above, however, he overwhelming proporion of all box-office receips hroughou he period was accouned for by films produced in one of hese six counries. All models lag he independen variable by one year on he assumpion ha i akes approximaely his amoun o ime for observed consumer spending o resul in a finished movie produc. I may be, however, ha he correc lag is longer or shorer. Model (2) makes use of he oal primary movie spending daa, bu his model has he drawback ha he acual revenues disribuors receive from he hree primary media accrue a differen poins in ime, and he fracion of oal consumer spending hey receive varies from one echnology o anoher. Model (1) avoids hese problems by including only heaer spending, under he implici assumpion ha heaer spending is a consan fracion of all movie spending in he six counries. Similarly for Model (3), which is based only on heaer admission couns. The laer model has he advanage ha i avoids cross-counry financial comparisons a curren exchange raes. Turning o Models (4) and (5), hese respecively specify he oal box office spending and he oal primary movie spending relaionship in per GDP erms. The laer model hus relies on financial comparisons, bu has he advanage of implicily avoiding he problems of irrelevan influences on exchange raes over ime. We wrie he general esimaion model as: y i, = α i + X i, β + ε i, whereα is a se of dummy variables represens he effecs specific o he six differen i counries, and ε i, varies independenly across ime and across counries. 12
Because we are using a panel daa se, however, we need o deermine if OLS, fixed effecs, or random effecs esimaion is mos appropriae. The fixed effecs model is wrien as y i ' = α + β X + ε, i i i while for he random effecs case: y i ' = α + β X + u + ε i i i where u i is an error erm measuring he degree o which he inercep of he ih crosssecional uni differs from he overall inercep, andε i is he usual error erm. If he counry-specific erm, α, does no vary over i, OLS provides consisen and efficien i esimaes. However, if counries have he same average impac on marke share, subjec o an addiional error erm ha differs for each individual counry, he random effecs model is more appropriae; ha is, α = α + u, where heα s represen independen random variables wih he same mean (α ) and variance ( σ ). i i Because culural, poliical, or oher facors may make Japan an unusual case in some sense, he models are esimaed boh wih and wihou Japanese marke share observaions included. Similarly, because of he exremes represened by he U.S., he models are also esimaed boh wih and wihou he U. S. domesic marke share observaions. i ' 2 u V. Resuls Tables 3 and 5 show he random effecs esimaes for he oal of 20 models esimaed. Tables 4 and 6 show es resuls for he appropriaeness of fixed v. random effecs models. Firs, given all of he confidence levels of he F-saisics in Tables 4 and 13
6 esing he join significance of he counry effecs in each model, he evidence is in favor of a counry specific effec in he daa. The Lagrange muliplier es (Breusch and Pagan, 1980) gives he confidence level a which he hypohesis, σ 2 u = 0 may be rejeced. Since p-values of he Lagrange Muliplier es for hese models are saisically significan, he random-effecs model is more appropriae han OLS. Finally, Hausman s es shows ha here is no significan correlaion beween u i and he independen variables, implying ha he random-effecs specificaion is appropriae. We also esimaed all models using fixed effecs, bu since hese resuls were very similar, only he random effecs esimaions are repored. All models repored were correced for auocorrelaion wih he Prais Winsen mehod. All parameer esimaes in Tables 3 and 5 for he domesic marke share models are saisically significan and in he hypohesized direcions a eiher he.01 or.05 significance levels. Tha is, domesic box-office shares are posiively relaed o domesic movie spending, movie spending per GDP, and movie admissions, relaive o hose of rading parner counries. Resuls were very close including or excluding Japan. Significance of coefficiens was, however, was generally weaker, however, in models wihou he U.S. Overall, hen, he saisical resuls of a variey of differen specificaions indicae consisen suppor for he exisence of home marke effecs in inernaional rade of hearical feaure films. VI. Discussion and Conclusion We have explained long erm rends in box-office marke shares in six major counries based on a home marke effec. From he 1950s o abou he 1970s, when domesic box office marke shares in counries ouside he U.S. were relaively high, 14
consumer spending on movies in hose counries was also relaively high. Since his period, he siuaion has roughly reversed; lower domesic marke shares in counries ouside he U.S. since he 1970s have been accompanied by relaively high levels of U.S. spending on movies. Our analysis hus implies ha declining domesic film indusries ouside he Unied Saes since he 1970s, accompanied by a rising dominance of U.S. movies worldwide, may be explained by a relaively slow growh in consumer movie spending in hese counries, relaive o ha of he Unied Saes. Tha faser spending growh in he Unied Saes has served o increase he economic resources of American movie producers in comparison o heir foreign counerpars, plausibly providing he foundaion for he Hollywood sudios o make a larger number of higher producion value films han compeing producers could offer. The models we esed have obvious limiaions. Firs, we have generalized he logic of a wo counry model o apply o muliple counries in our saisical analysis. I is no obvious, for example, how simulaneous spending growh or decline in a number of differen counries rading wih a cerain single naion, for example, should affec he overall rade balance for ha individual counry. Furhermore, he exreme simpliciy of he models allows no more han an explanaion of broad rends. The causes of shifs in movie box-office shares are surely complex, involving a wide variey of culural, sociological, poliical, and echnological facors for which we had no sysemaic daa. 7 7 Among hese are: he influence of American adverising and general fascinaion wih American and Americna producs (Pells, 1997; Sorlin, 1991; Tracey, 1985; Tunsall, 1977), prevalence of he English language (Wildman & Siewek, 1988; Sraubharr, 1991), aggressive behavior by he Moion Picure Associaion of America (MPAA) and he Unied Saes governmen o promoe American movie company ineress (Guback, 1969; Jarvie, 1992; Punam, 1998; Schiller, 1969; Seagrave, 1997; Thompson, 1985), and inadequae or misdireced proecionis or subsidizaion policies in oher counries (Dale, 1997; Ilo, 1996; Finney, 1996). 15
Whaever he complee explanaion, however, he economic realiy remains ha over he long erm, hearical film invesmens ha are reasonably responsive o markeplace forces in any counry mus have a commercial suppor base of icke buyers, pay elevision subscribers, and video reners and purchasers o reurn hose invesmens. The home marke model implies ha a given counry s domesic suppor base is disproporionaely imporan o mainaining a viable hearical film indusry. 16
References Breusch, T., & Pagan, A. (1980). The LM es and is applicaions o model specificaion in economerics. Review of Economic Sudies, 47, 239-254. Dale, M. (1997). The Movie Game: The Film Business in Briain, Europe, and America. Wellingon House Davis, D., & Weinsein, D. E. (1999). Economic Geography and Regional Producion Srucure: An Empirical Invesigaion. European Economic Review, 43, 379-407. Davis, D., & Weinsein, D. E. (2003). Marke Access, Economic Geography and Comparaive Advanage: An Empirical Tes. Journal of Indusrial Economics, 59(1), 1-23. European Union. (1994, 1999). Repor of he Think Tank on he Audiovisual Policy in he European Union. Brussels: Auhor. European Audiovisual Observaory. Saisical Yearbook, annual. Srasbourg, France: Auhor. European Audiovisual Observaory (2004). Focus 2004: World Film Marke Trends. Srasbourg, France. Feensra, R., Markusen, J., & Rose, A. (2001). Using he graviy equaion o differeniae among alernaive heories of rade. Canadian Journal of Economics, 34, 430-47. Finney, Angus (1996) The Sae of European Cinema. London: Cassell. French, K., & Poerba, J. (1991). Invesor Diversificaion and Inernaional equiy markes. American Economic Review, 81(2), 222-226. Frank, B. (1992), A Noe on he Inernaional Dominance of he U.S. in he Trade in Movies and Television Ficion, Journal of Media Economics 5 (1992): 31 38. 17
Grinbla, M., & Keloharju, M. (2001). How Disance, Language, and Culure Influence Sockholdings and Trades, Journal of Finance, 56(3), 1053-1074. Guback, T. H. (1969). The Inernaional Film Indusry. Bloomingon, IN: Indiana Press. Head, K., & Reis, J. ( 2001). Increasing reurns versus naional produc differeniaion as an explanaion for he paern of U.S.-Canada rade. The American Economic Review, 91(4), 858-876. Head, K., & Reis, J. (2002). On he pervasiveness of home marke effecs. Economerica, 69(275), 371-390. Helpman, E., & Krugman, P. (1985). Marke Srucure and Foreign Trade: Increasing Reurns, Imperfec Compeiion and he Inernaional Economy. Cambridge: MIT Press. Hoskins, C., & Mirus, R. (1988). Reasons for he US dominance of he inernaional rade in elevision programmes. Media, Culure, and Sociey, 10, 499-515. Ilo, T. (1996), Budges and Markes: A Sudy of he Budgeing of European Film, Rouledge Inernaional Moneary Fund (2001). Inernaional Financial Saisics (CD Rom). Washingon, D. C.: Join IMF World Bank CD-Rom Projec. Inernaional moion picure almanac. (2000). New York: Quigley. Jarvie, I. (1992). Hollywood s Overseas Campaign: The Norh American Movie Trade, 1920-1950. Cambridge, NY: Cambridge Universiy Press. Jayakar, K., & Waerman, D. (2000). The Economics of American Movie Expors: An Empirical Analysis. Journal of Media Economics, 13(3), 153-169. Jouhou Media Hakusyo (Informaion Media Whiebook), annual. Tokyo: Densu Insiue. 18
Krugman, P. R. (1980). Scale Economies, Produc Differeniaion, and he Paern of Trade. American Economic Review, 70(5), 950-959. Lange, A., & Newman, S. (2003). Focus 2003: World Film Marke Trends. Rerieved November 07, 2003 from www.obs.coe.in. Lee, S. (2002),. An Economic Analysis of he Movie Indusry in Japan. The Journal of Media Economics, 15(2), 125-139. Linder, S. B. (1961). An Essay on Trade and Transformaion. New York: John Wiley Lundback, E. J. & Torsensson, J. (1998). Demand, Comparaive Advanage and Economic Geography in Inernaional Trade Evidence from he OECD. Review of World Economics, 134 (2), 230-249. Marvasi, A. (1994). Inernaional Trade in Culural Goods: A Cross-secional Approach. Journal of Culural Economics, 18, 135-148. Marvasi, A. (2000). Moion Picures Indusry: Economies of Scale and Trade. Inernaional Journal of Economics of Business, 7(1), 99-114. Mills, P. (1985). An Inernaional Audience. Media, Culure and Sociey, 7, 487-501. Moion Picure Producers Associaion of Japan (2005). Saisics of Film Indusry in Japan. Rerieved May 15, 2005 from hp://www.eiren.org. Oh, J. (2001). Inernaional rade in film and he self-sufficiency raio. Journal of Media Economics, 14, 31-44. Paul Kagan Associaes (2002, Augus 9). Moion Picure Invesor. p. 4. Pells, R. (1997). No Like Us. Basic Books. Pool, I., (1977), The changing flow of elevision. Journal of Communicaion, 27(2), 139-149. 19
Punam, D. (1998), Movies and Money, Knopf. Schiller, H. (1969, 1992). Mass communicaion and American empire. Boulder, CO: Wesview Press. Screen Diges, monhly, London: Auhor. Seagrave, K. (1997). American Films Abroad: Hollywood's Dominaion of he World's movie screens from he 1890s o he presen. N.C. : McFarland. Shaked, A., & Suon, J. (1987). Produc Differeniaion and Indusrial Srucure. Journal of Indusrial Economics, 36(2), 131-146. Sorlin, P. (1991). European cinemas, European socieies, 1939-1990. New York: Rouledge. Sraubhaar, J. D. (1991). Beyond Media Imperialism: Asymmerical Inerdependence and Culural Proximiy. Criical Sudies in Mass Communicaion, 8, 39-59. Suon, J. (1991). Sunk Cos and Marke Srucure. MIT Press. Thiermeyer, M. (1994). Inernaionalisierung von Film und Filmwirschaf. Koln. Thompson, K. (1985). Exporing enerainmen: America in he world film marke, 1907-1934. London: Briish Film Insiue. Tracey, M. (1985). The poisoned chalice? Inernaional elevision and he idea of dominance. Daedalus, 114, 17-56. Tunsall, J. (1977). The Media are American. London: Consable. Veronis, Suhler & Associaes (1999, 2000). Communicaions Indusry Forecas. New York, N. Y.: Veronis, Suhler & Associaes. Waerman, D. (2005). Hollywood's Road o Riches, Harvard Universiy Press, forhcoming. 20
Waerman, D., & Jayakar, K. (2000). The Compeiive Balance of he Ialian and American Film Indusries. European Journal of Communicaion, 15(4), 501-528. Waerman, D., & Lee, S. (2005). Deerminans of U. S. Box Office Marke Shares in he EUJ5, 1950-2003. In D. Waerman, Hollywood's Road o Riches. Harvard Universiy Press, forhcoming. Waerman, D., S. Lee, & W. Lu (2005). Comparaive Analysis of Movie Indusries and Trade in he Unied Saes and he EUJ5 Counries: Saisical Daa, 1950-2003. In D. Waerman, Hollywood's Road o Riches, Harvard Universiy Press, forhcoming. Waerman, D. (1988). World Television Trade: The Economic Effecs of Privaizaion and New Technology. Telecommunicaions Policy, 12(2), 141-151. Weder, R. (1996). How domesic demand shapes he paern of inernaional rade, World Developmen. Wildman, S. (1995). Trade liberalizaion and policy for media indusries: A heoreical examinaion of media flows. Canadian Journal of Communicaion, 20, 367-388. Wildman, S and Sephen E. Siwek (1987), The Privaizaion of European Television: Effecs on Inernaional Markes for Programs, Columbia Journal of World Business 22 (Fall): 71 76 Wildman, S., & Siwek, S. (1988). Inernaional Trade in Films and Television Programs. Cambridge, MA: Ballinger. Wolf, H. (2000). Inranaional Home Bias in Trade. Review of Economics and Saisics, 82(4), 555-563. 21
Table 1. Domesic Marke Shares (%) in he Six Subjec counries, 1950-2003 Year U. S. France Germany Ialy UK Japan 1950 n. a. 45.9 n. a. 23.9 28.9 n. a. 1951 n. a. 47.4 n. a. 27.9 27.0 n. a. 1952 n. a. 49.5 n. a. 33.1 27.7 n. a. 1953 n. a. 48.4 n. a. 34.9 30.6 n. a. 1954 n. a. 48.8 n. a. 36.2 32.8 n. a. 1955 n. a. 48.9 47.3 34.8 29.9 65.8 1956 n. a. 50.3 47.1 34.8 32.6 67.5 1957 n. a. 51.9 47.7 30.0 37.1 69.1 1958 n. a. 49.5 47.7 32.6 39.2 76.1 1959 n. a. 51.5 47.0 35.9 40.9 77.4 1960 n. a. 53.3 41.0 41.2 41.2 78.3 1961 n. a. 52.7 32.6 41.9 45.0 77.2 1962 n. a. 51.4 29.0 47.0 41.2 73.1 1963 n. a. 49.3 30.4 45.9 47.7 68.8 1964 n. a. 50.9 27.0 45.2 47.1 66.3 1965 n. a. 55.0 25.8 47.0 n. a. 66.7 1966 n. a. 53.2 25.9 58.9 n. a. 63.2 1967 n. a. 54.4 24.7 53.2 n. a. 61.3 1968 n. a. 52.5 37.0 54.0 n. a. 64.4 1969 n. a. 48.5 39.3 61.7 42.7 64.1 1970 n. a. 52.6 39.2 60.3 39.8 59.5 1971 n. a. 56.1 36.1 65.1 43.7 51.3 1972 n. a. 55.9 32.7 64.8 41.5 51.9 1973 n. a. 61.8 26.3 62.5 36.9 55.9 1974 n. a. 56.2 26.5 62.0 27.0 51.1 1975 n. a. 52.4 12.9 59.1 29.5 44.4 1976 n. a. 52.5 11.4 60.8 22.5 48.4 1977 n. a. 47.3 11.4 50.8 19.7 50.8 1978 n. a. 46.5 12.8 42.8 26.1 48.6 1979 n. a. 51.1 16.0 36.3 17.2 53.5 1980 n. a. 46.9 9.3 43.5 10.0 55.0 1981 95.0 49.6 18.7 44.1 13.0 54.5 1982 98.0 53.3 11.3 46.1 16.0 51.1 22
1983 96.1 49.3 14.1 39.5 20.0 52.6 1984 96.8 49.6 16.8 33.0 17.0 48.6 1985 97.7 44.7 22.7 31.8 14.0 51.6 1986 93.6 43.7 22.1 31.6 12.0 49.8 1987 98.3 36.1 17.2 34.1 9.8 48.1 1988 98.8 39.1 23.4 28.5 14.8 49.7 1989 98.0 34.3 16.7 21.7 10.0 46.6 1990 98.6 37.5 9.7 21.0 7.0 41.4 1991 98.4 30.6 13.6 26.8 5.5 41.9 1992 98.7 35.0 9.5 24.4 6.8 45.1 1993 95.8 35.1 7.2 17.3 2.5 35.8 1994 94.8 28.3 10.1 23.7 8.8 40.1 1995 95.9 35.2 6.3 21.1 10.2 37.0 1996 95.7 37.5 15.3 24.9 12.8 36.3 1997 92.4 34.5 16.7 32.9 28.1 41.5 1998 94.6 27.6 8.1 24.7 14.1 30.2 1999 91.7 32.4 14.0 24.1 17.8 32.0 2000 93.3 28.5 12.5 17.5 21.4 31.8 2001 93.1 39.0 16.2 19.4 11.7 39.0 2002 93.9 34.0 11.9 21.9 15.4 27.0 2003 95.1 34.8 17.5 22.0 11.9 33.0 Sources: EAO yearbook (various issues); EAO Focus (2002, 2003, 2004); Guback (1969), Moion Picure Producers Associaion of Japan(2005); Screen Diges (various issues), Thiermeyer (1994). See also Waerman, Lee, and Lu (2005) for specific deails. 23
Table 2. Primary Movie Spending as a proporion of GDP in he Six Subjec Counries, 1950-2002 Year U. S. France Germany Ialy UK Japan EURO- 4 Non-US combned 1950 0.481 0.257 0.350 n. a. 0.803 n. a. n. a. n. a. 1951 0.402 0.271 0.343 0.680 0.742 n. a. 2.438 n. a. 1952 0.379 0.266 0.354 0.726 0.698 0.523 2.423 2.945 1953 0.362 0.274 0.383 0.742 0.646 0.614 2.407 3.021 1954 0.337 0.283 0.414 0.770 0.616 0.601 2.420 3.020 1955 0.298 0.280 0.402 0.778 0.555 0.654 2.313 2.966 1956 0.264 0.260 0.405 0.709 0.506 0.656 2.145 2.801 1957 0.242 0.257 0.398 0.643 0.429 0.627 1.968 2.595 1958 0.222 0.243 0.376 0.588 0.367 0.627 1.797 2.424 1959 0.198 0.223 0.321 0.584 0.285 0.539 1.610 2.149 1960 0.183 0.223 0.252 0.521 0.251 0.455 1.428 1.883 1961 0.173 0.201 0.210 0.488 0.222 0.378 1.294 1.671 1962 0.149 0.192 0.182 0.459 0.202 0.346 1.183 1.530 1963 0.152 0.183 0.156 0.424 0.182 0.307 1.095 1.401 1964 0.143 0.170 0.141 0.415 0.176 0.257 1.045 1.302 1965 0.145 0.163 0.128 0.406 0.174 0.230 1.017 1.246 1966 0.135 0.150 0.117 0.389 0.156 0.198 0.947 1.145 1967 0.133 0.139 0.108 0.351 0.145 0.175 0.877 1.052 1968 0.141 0.128 0.096 0.338 0.134 0.153 0.836 0.989 1969 0.131 0.115 0.090 0.320 0.125 0.134 0.781 0.915 1970 0.137 0.111 0.080 0.271 0.115 0.112 0.715 0.827 1971 0.123 0.106 0.074 0.283 0.106 0.097 0.688 0.786 1972 0.128 0.109 0.070 0.298 0.093 0.083 0.697 0.781 1973 0.110 0.104 0.066 0.275 0.079 0.082 0.633 0.716 1974 0.127 0.104 0.062 0.263 0.083 0.087 0.641 0.728 1975 0.131 0.107 0.061 0.261 0.068 0.088 0.629 0.717 1976 0.115 0.103 0.053 0.215 0.061 0.088 0.547 0.634 1977 0.123 0.096 0.055 0.160 0.059 0.082 0.492 0.574 1978 0.126 0.097 0.058 0.137 0.071 0.079 0.488 0.567 1979 0.127 0.093 0.061 0.117 0.064 0.071 0.463 0.534 1980 0.126 0.108 0.070 0.111 0.091 0.083 0.505 0.588 24
1981 0.151 0.122 0.077 0.109 0.112 0.091 0.571 0.660 1982 0.190 0.131 0.077 0.108 0.118 0.101 0.623 0.723 1983 0.221 0.130 0.082 0.097 0.138 0.114 0.668 0.782 1984 0.243 0.126 0.081 0.084 0.145 0.114 0.679 0.793 1985 0.261 0.118 0.083 0.082 0.159 0.118 0.703 0.821 1986 0.277 0.155 0.085 0.087 0.17 0.124 0.775 0.899 1987 0.303 0.167 0.091 0.078 0.178 0.148 0.817 0.965 1988 0.324 0.168 0.093 0.070 0.180 0.157 0.836 0.993 1989 0.345 0.201 0.090 0.095 0.224 0.144 0.955 1.101 1990 0.350 0.208 0.097 0.078 0.228 0.130 0.961 1.091 1991 0.355 0.220 0.093 0.082 0.237 0.129 0.987 1.116 1992 0.350 0.244 0.078 0.095 0.263 0.121 1.030 1.151 1993 0.350 0.259 0.088 0.121 0.281 0.119 1.098 1.217 1994 0.352 0.260 0.088 0.110 0.282 0.122 1.092 1.214 1995 0.352 0.263 0.086 0.099 0.296 0.128 1.096 1.223 1996 0.361 0.269 0.088 0.098 0.299 0.121 1.115 1.236 1997 0.354 0.270 0.084 0.105 0.304 0.141 1.118 1.259 1998 0.364 0.282 0.090 0.115 0.321 0.153 1.172 1.325 1999 0.368 0.274 0.086 0.114 0.311 0.140 1.153 1.293 2000 0.366 0.292 0.089 0.126 0.356 0.143 1.229 1.372 2001 0.381 0.303 0.107 0.132 0.426 0.163 1.350 1.513 2002 0.401 0.315 0.126 0.128 0.483 0.173 1.452 1.625 Sources: EAO yearbook (various issues); EAO Focus (2002, 2003, 2004); Inernaional Moion Picure Almanac (2000); Jouhou Media Hakusyo (various issues); Moion Picure Producers Associaion of Japan (2005); Screen Diges (various issues), Thiermeyer (1994); Veronis, Suhler, and Associaes (1999). See also Waerman, Lee, and Lu (2005) for specific deails. 25
Table 3. Esimaes of Random Effecs Model: Domesic Box-office Share Models (incl. U.S): 1950-2003 Six Counries Six Counries excep Japan Models Model 1 Model 2 Model 3 Model 4 Model 5 Model 1 Model 2 Model 3 Model 4 Model 5 Dependen Variable Independen Variables (-saisic) (-saisic) (-saisic) (-saisic) (-saisic) (-saisic) (-saisic) (-saisic) (-saisic) (-saisic) 29.9805 20.8788 31.3884 31.7235 25.8734 25.9871 19.1254 30.2168 30.6237 Consan 30.0899 (5.726) c (5.555) c (3.635) c (3.502) c (3.885) c (4.314) c (4.248) c (3.249) c (2.852) b (3.487) c BOXSPEN DSHARE 0.7323 (4.294) c 0.8075 (4.939) c TOTALSP 0.7468 0.7993 ENDSHAR (4.491) c (5.056) c E BOXADMI 1.2736 1.0989 SSIONSH (7.873) c (7.413) c ARE BOXPERG DPSHARE 0.6696 (2.963) c 0.5804 (2.788) c TOTALPE 0.6167 0.5254 RGDPSHA (2.961) c (2.766) c RE 26
No. of observaion s 270 270 253 264 264 222 222 222 219 219 Noe: b p<0.05; c p<0.01 Table 4. Variance of Error Terms in Random Effecs Models (incl. U.S.): 1950-2003 Six Counries Six Counries excep Japan Models Model 1 Model 2 Model 3 Model 4 Model 5 Model 1 Model 2 Model 3 Model 4 Model 5 Var [u] 1.194 1.505 6.714 5.378 3.751 2.008 2.469 4.261 6.455 3.596 Var [e] 20.540 20.552 21.044 20.482 20.433 20.678 20.693 19.753 20.303 20.348 F-es* 12.81 c 14.21 c 49.73 c 14.98 c 13.33 c 20.78 c 23.20 c 46.93 c 17.09 c 15.38 c Lagrange Muliplier 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 Tes** Hausman s Tes*** 0.0103 0.0385 0.1945 0.057 0.063 0.1004 0.2251 0.6029 0.0449 0.018 Noe 1: c p<0.01 Noe 2: * The F-saisics for esing he join significance of he counry effecs; ** Significance level a which he hypohesis σ 2 u = 0 is rejeced; *** Probabiliy of reaining random-effecs model over fixed-effecs model, using Hausman s es of appropriaeness of assumpion. 27
Table 5. Esimaes of Random Effecs Model: Domesic Box-office Share Models (no incl. U.S.) II: 1950-2003 Six Counries excep U. S. Six Counries excep U. S. and Japan Models Model 1 Model 2 Model 3 Model 4 Model 5 Model 1 Model 2 Model 3 Model 4 Model 5 Dependen Variable Independen Variables (-saisic) (-saisic) (-saisic) (-saisic) (-saisic) (-saisic) (-saisic) (-saisic) (-saisic) (-saisic) Consan 27.0558 26.8755 18.9182 22.8487 24.0041 22.2685 22.8242 17.7679 18.0149 20.2589 (5.253) c (4.955) c (3.109) c (3.173) c (3.223) c (4.577) c (3.455) c (2.526) b (2.856) c (2.806) c BOXSPEN DSHARE TOTALSP ENDSHAR E BOXADMI SSIONSH ARE BOXPERG DPSHARE TOTALPE RGDPSHA 0.7567 (2.185) b 0.9923 (2.675) c 0.8455 (2.520) b 0.9936 (2.831) c 1.7279 1.3163 (7.062) c (5.467) c 0.6609 0.6294 (2.692) c (2.804) c 0.5866 0.5166 (2.586) c (2.454) b 28
RE No. of observaion s 248 248 231 242 242 200 200 200 197 197 Noe: b p<0.05; c p<0.01 Table 6. Variance of Error Terms in Random Effecs Model (no incl. U.S.): 1950-2003 Six Counries excep U. S. Six Counries excep U. S. and Japan Models Model 1 Model 2 Model 3 Model 4 Model 5 Model 1 Model 2 Model 3 Model 4 Model 5 Var [u] 0.731 1.228 8.061 2.253 2.339 0.701 2.941 6.423 0.796 1.483 Var [e] 21.753 21.707 22.166 21.901 21.808 22.247 22.206 21.181 22.042 22.032 F-es* 4.82 c 5.80 c 25.60 c 5.90 c 5.32 c 6.95 c 7.94 c 17.40 c 5.17 c 4.69 c Lagrange Muliplier 0.000 0.000 0.000 0.0000 0.000 0.000 0.000 0.000 0.0000 0.0000 Tes** Hausman s Tes*** 0.1677 0.1501 0.6022 0.6094 0.8305 0.2146 0.3553 0.8552 0.2472 0.5088 Noe 1: c p<0.01 Noe 2: * The F-saisics for esing he join significance of he counry effecs; ** Significance level a which he hypohesis σ 2 u = 0 is rejeced; *** Probabiliy of reaining random-effecs model over fixed-effecs model, using Hausman s es of appropriaeness of assumpion. 29
30