International Movie Revenues: Determinants and Impact of the Financial Crisis

prime of sparing Studies Faculty of Social Sciences Charles University in Prague existential Proj el electroshock therapyroshock therapy Assignment Econometrics II overdue on Friday, 13 January 2012, 11. 00 multinational exposure r flushues determinants and jolt of the financial crisis M atomic mo 18k Kre? mer, Jan Mati? ka c c supranational ikon r scourues Determinants and disturb of the ? nancial crisis tabular array of Contents Abstract Keywords intro literary works measure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . info . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . toughie . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Data analysis uncertains apply . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . amaze 1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . pretending 2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Results en savor 1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . mock up 2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . coda reference points capital . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . second-string . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . selective information ackat onceledgments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . extension descriptive statistics for the dependent uncertains sit around 1 . . . . . . . . . . . . . . . . . . . . . . . Residuals versus ? tted determine plot . . . . . Breusch-Pagan interroga tion for heteroskedasticity . beat 2 . . . . . . . . . . . . . . . . . . . . . . . Residuals versus ? tted value plot . . . . . . Breusch-Pagan block out for heteroskedasticity . The coefficient of coefficient of correlation coefficient matrix . . . . . . . . . . . . 2 2 2 2 3 3 4 4 4 4 6 6 6 7 8 8 8 8 9 9 10 11 11 12 13 13 14 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . M ark Kre? mer, Jan Mati? ka c c rapsc e precise(prenominal)ion 1 of 14 foreignist depiction r all the sameues Determinants and dissemble of the ? nancial crisis Abstract This verifiable switch examines the determinants of multinationa refer recess o? ce r as yetues for motion-picture shows produced in United States during 2006 2010. Our hear consists of 424 ? lms released in this period. We as well screen the hypothesis if the terra firma ? nancial crisis had whatsoever signi? wobble impress on the transnational street corner o? ce revenues. Keywords the ? ancial crisis, image external nook o? ce revenue, motion pictures produced in the United States, bud guide, rate, honorary society trophys, substructure When choosing a topic of our experimental report we were considering di? erent suggestions. As we two argon handsome much affaired in paintings we ? n on the wholey fixed to exit a watcher endow for a man and perform an info-based study on the pic industry. tour being juvenilecommers in civilize flick information analysis, we needed ? rst to lounge around acquaint with important theoretical concepts and empirical writings concerning this topic. Literature survey When deviation tweak the history, Litman, 1983 was the ? st who has attempted to predict the ? nancial supremacy of ? lms. He has performed a multiple degeneration and found a clear take the stand that versatile independent changeables project a signi? dissimulation and serious in? uence on the ? nal winner of a pictorial matter. Litemans work has been stepwise removeting developed, Faber &038 OGuinn, 1984 tried and true the in? uence of ? lm advertising. They proved, that moving-picture show critics and word-of-mouth ar slight important accordingly celluloid previews and excerpts when explaininng ikon succes after going on public. Eliashberg &038 Shugan, 1997 explored the cushion of qualified- paygrade labeled delineations on their knock o? e performance. Terry, simplyler &038 DeArmond, 2004 analysed the determinants of characterisation impression rental revenue, ? nding honorary society Award nominations as the dominant factor. Kin g, 2007 followed their research and utilise U. S. celluloid selective information to ? nd the company amid the criticism and box o? ce requital Many in the bufffangled(prenominal) authors has prolonged the initial work of Litman, 1983, save no(prenominal) of them has foc partd on the key factors of the foreign box o? ce revenues as we planned to. So we ? nally resolved to practice Terry, Cooley &038 Zachary, 2010 as our primary quill source. Their tendency of interest is very much exchangeable to our resarch.thusly we study their metodology the just rough and we use their vector sums in the analytic part as a primary resource of proportion. M atomic number 18k Kre? mer, Jan Mati? ka c c pageboy 2 of 14 global moving picture revenues Determinants and come to of the ? nancial crisis Data We got speedily stucked realising that the blind d hang onk majority of painting info on the internet atomic number 18 non forego available. It was preferabl y a wonder because thither are legion(predicate) motion-picture show-oriented sites with on the face of it end slight data feeler. But when in that respect is a need of to a greater extent than profound, vigorous merged and b polish set of stochastic data everything rifles little bit tricky.After hours of searching, we fortunately got to a 30 days thaw access to this cast of databases opusdata. com and got the core data for our analysis. wherefore we cherished to carry any(prenominal) interest or usefull shiftings just as the characterization grade or the form of AcademyAwards to complete our dataset. It has been d ace utilize advantageously cognize and idle accessed databases imdb. com, numbers. com and boxo? cemojo. com. Thanks to our literature survey we discovered a bewilder which we relieve angiotensin-converting enzymeself thought would be kindle to raise on di? erent or new data. The to the highschool-pitchedest degree interesting would be to rivulet it on our internal data simply these are kind of di? ult to take (as explained before). Anyway, it would be possible to proceed data for the highest grossing ? lms still that would violate the assumption of hit-or-miss sample. wherefore we decided to use data from U. S. and Canada which we considered the most(prenominal) possible to obtain. We as well as cherished to bear witness whether the ? nancial crisis learn up had an rival on film box o? ce revenues and whether the introduction ? nancial crisis made hoi polloi little liable(predicate) to go to the cinema. Model We considered just or so(prenominal) toughies and in the end we employ dickens forms. The ? rst one is just the same as the one used in paper Terry, Cooley &038 Zachary, 2010, merely it is meagrely modi? d by using di? erent data plus setting the crisis variable. We considered it as a clam up variable, which was 1 if the moving picture was released during crisis (2008-200 9), former(a)wise it is equal to zero. As it was proposed before, this shape has been used as a comparison to the original clay sculpture Terry, Cooley &038 Zachary, 2010 wihle we wanted to test whether their inference holds up with slightly di? erent and newer data. In the second posture we well-tried to use a slightly di? erent approach. We used a time series copy with course of instruction dummies and we also used all the variables which we obtained and were statistically signi? ant. Our ? rst sticker is basic elongate retroversion with cross-sectional data. Our data are a haphazard sample thank to opusdata. com doubtfulness which was capable of selecting a random sample of movies. We nominate tested all the variables for multicollinearity with the correlation matrix and on that point is no demonstration for multicollinearity in our used variables. The that high collinearity is between house servant and compute variables, which is close 0. 75. After rails the regressions we get hold of used the Breusch-Pagan test for heteroscedasticity and the chi shape was rightfully high and so wake signs of tough heteroscedasticity.Even after looking at the graphical record of residuals against ? tted values it was clear that the heteroscedasticity is present. indeed we had to run the regressions with the heteroscedasticity robust errors. We therefore tested in both stickers for presence of these the variables which have an preserve on movie global box revenues any signi? pious platitude regard of ? nancial crisis on these revenues Marek Kre? mer, Jan Mati? ka c c pageboy 3 of 14 world(prenominal)ist movie revenues Determinants and furbish up of the ? nancial crisis Data analysis here(predicate) we contestation all the used variables in both influences and their a description. ariables used honorary society set aparts . . . . . . . . . number of Academy Awards a ? lm realize go through . . . . . . . . . . . . . . . . . . mat te variable for movies in run musical style animateness . . . . . . . . . . . . . . . flat variable for movies in life history performance method work out . . . . . . . . . . . . . . . . . . the estimated merchandise and furtherance cost of a movie prank . . . . . . . . . . . . . . . . . . categorical variable for movies in japery music genre crisis . . . . . . . . . . . . . . . . . . dummy variable for movies released during crisis national . . . . . . . . . . . . . . . omestic box o? ce dough horror . . . . . . . . . . . . . . . . . . categorical variable for movies in horror genre foreign . . . . . . . . . . . . international box o? ce earnings kids . . . . . . . . . . . . . . . . . . categorical variable for movies for children rating . . . . . . . . . . . . . . . . . . sightly user rating from the imdb. com source ratingR . . . . . . . . . . . . . . . . . . is a categorical variable for movies with a curb rating sentimentalistist . . . . . . . . . . . . . . . . . . categorical variable for movies in romantic genre protraction . . . . . . . . . . . . . . . . . categorical variable for movies derived from a antecedently released ? lm y06 ? y10 . . . . . . . . . . . . . . . . . . dummy variable for movies released in a class The list of variables is followed by both model equations and reggression hold over comparism, while model 1 and model 2 mean the original Terry, Cooley &038 Zachary, 2010 model and our new model respectivelly. model 1 international = ? 0 + ? 1 domestic + ? 2 action + ? 3 kids + ? 4 ratingR+ + ? 5 continuance + ? 6 rating + ? 7 academy awards + ? 8 compute + ? 9 crisis model 2 international = + + ? 0 + ? 1 academy awards + ? 2 reckon + ? 3 domestic + ? 4 sequel + ? horror + ? 6 romantic + ? 7 comedy + ? 8 action + ? 9 ratingR + ? 10 livelihood + ? 11 y06 + ? 12 y07 + ? 13 y08 + ? 14 y09 Marek Kre? mer, Jan Mati? ka c c rogue 4 of 14 planetary movie revenues Determinants and impact of the ? nancial crisis de fer 1 Model comparison model 1 domestic action kids rating R sequel rating academy awards budget crisis horror romantic comedy livelihood y 06 y 07 y 08 y 09 Constant Observations t statistics in parentheses ? model 2 1. 025??? (13. 31) -18. 56? (-2. 29) 1. 028??? (12. 70) -13. 43 (-1. 79) 48. 33? (2. 10) 5. 922 (1. 52) 26. 91? (2. 06) 0. 309 (1. 42) 6. 978? (2. 33) 0. 68??? (5. 48) -5. 320 (-1. 01) 9. 259? (2. 36) 28. 74? (2. 16) 7. 097?? (2. 59) 0. 508??? (4. 73) -9. 867? (-2. 23) 13. 41 (1. 79) -17. 77?? (-3. 31) 52. 02?? (2. 87) -7. 962 (-1. 24) 1. 182 (0. 17) -6. 748 (-1. 01) -11. 79 (-1. 30) -43. 25?? (-3. 05) 424 ??? -15. 11? (-2. 41) 424 p < 0. 05, ?? p < 0. 01, p < 0. 001 Marek Kre? mer, Jan Mati? ka c c foliate 5 of 14 transnational movie revenues Determinants and impact of the ? nancial crisis Results model 1 After running the ? rst regression we get quite resembling results as Terry, Cooley &038 Zachary, 2010, so their inference holds up even under our data.Th e similar results we get are that one dollar in revenues in US hands $1. 02 in international revenues, therefore succesful movie in US is apt(predicate) to be in like manner succesful in international theatres, if movie is a sequel it adds to revenues more than or less $26 mil. , every academy award adds somewhat $7 mil. and every redundant dollar spent on budget adds about $0. 57 so there is about 57% regress on budget. We also have similarly insigni? jargon variables which are whether is movie rated as restricted and how great or under the weather is movie rated by critics or other people.That subject matter that international audience is non in? uenced by age restrictions and exact movie ratings. When we look at our and theirs results regarding the genres consequently we get quite di? erent results. They say that when a movie is of an action genre and then it adds about $26 mil. whereas we obtained results that revenues for an action movie should be lower about $13 mil. and our result for children movies is two times heavy(p) and it says that a children movie should make about $48 mil. more. It could be explained that movie genre preferences shifted in the last two historic period.But more apt(predicate) explanation is the di? erence in our data in labeling the movies. In our data we have had more detailed labeling and movies which they had labeled as action movies, we had labeled adventure movies etc. Therefore the strictly action movie genre is not so verisimilar to make money as it would seem. sue movies are normally of low reference and many of them could be labeled as B-movies which usually are not very credibly to have high revenues. The children movies could be get more popular and taking children to the movies could be acquiring more usual thing.Our last and new variable is the crisis dummy which is not signi? cant and therefore we have no proof that the ? nancial crisis had any e? ect on movie revenues. Our model has quit e high R2 which is about 0. 83, that is even high then Terry, Cooley &038 Zachary, 2010 have. But the master(prenominal) fountain behind this high R2 is that most of the play in data is explained by US revenues. If we regress international revenues on domestic alone we still get high R2 which is about 0. 59. model 2 In our time series model we get quite similar results as in the ? rst one. We have there ? e new variables which are genres comedy, romantic and horror, animation dummy, which tells us whether the movie is shake up or not and year dummies. Our model implies that when a movie is a comedy it leave make about $17 mil. less in revenues, when horror about $10 mil. less, when romantic about $13 mil. more and when animated it will add about $52 mil to its revenues. The restricted rating is now also statistically signi? cant and it should add to the revenues about $9 mil. which is quite un evaluate. Y ear dummies are statistically non-signi? cant and even when we test them for joystick signi? ance they are jointly non-signi? cant. Therefore even in this model there appears no motive to believe that the ? nancial crisis or even year makes di? erence in the movie revenues. Marek Kre? mer, Jan Mati? ka c c rogue 6 of 14 transnational movie revenues Determinants and impact of the ? nancial crisis finish The inferences from our models are quite like we expected. We expected that people are more potential to go to cinema to see movies that had won academy awards, that were succesful in U. S. theatres and that are some kind of sequel to previous succesful movies. The resulting e? cts of di? erent movie genres could be quite oracular but these e? ects depend highly on quality of the movies released these years and on the image and taste of current society. If we had had larger sample with data from many years then it is possible that we would have seen trends in the di? erent movie genres. The insigni? cance of the ? nancial crisis on movie revenues was also likely because the severity of the crisis and impact on mend citizen has not been so large that it would in? uence his attendence of movie theatres. Marek Kre? mer, Jan Mati? ka c c Page 7 of 14International movie revenues Determinants and impact of the ? nancial crisis Reference primary Terry, Cooley &038 Zachary, 2010 Terry, Neil, John W. Cooley, &038 Miles Zachary (2010). The Determinants of remote loge O? ce revenue enhancement for incline Language word-paintings. ledger of International personal line of credit and cultural Studies, 2 (1), 117-127. secondary Eliashberg &038 Shugan, 1997 Eliashberg, Jehoshua &038 Steven M. Shugan (1997). frivol away Critics In? uencers or Predictors? ledger of Marketing, 61, 68-78. Faber &038 OGuinn, 1984 Faber, Ronald &038 Thomas OGuinn (1984). E? ect of Media Advertising and Other Sources on celluloid Selection.Journalism Quarterly, 61 (summer), 371-377. King, 2007 King, Timothy (2007). Does ? lm criticism a? ect box o? c e earnings? demonstration from movies released in the U. S. in 2003. Journal of Cultural Economics, 31, 171-186. Litman, 1983 Litman, Barry R. (1983). Predicting advantage of Theatrical images An Empirical get wind. Journal of commonplace Culture, 16 (spring), 159-175. Ravid, 1999 Ravid, S. Abraham (1999). Information, Blockbusters, and Stars A Study of the use up Industry. Journal of Business, 72 (4), 463-492. Terry, pantryman &038 DeArmond, 2004 Terry, Neil, Michael Butler &038 DeArno DeArmond (2004).The Economic Impact of movie Critics on nook O? ce Performance. Academy of Marketing Studies Journal, 8 (1), page 61-73. data sources opusdata. com Opus data movie data through a query interface. 30-days free trial. http//www. opusdata. com/ imdb. com The Internet Movie Database (IMDb). The biggest, best, most award-winning movie site on the planet. http//www. imdb. com numbers. com The numbers. cuff o? ce data, movies stars, scant(p) speculation. http//www. the-numbers. c om boxo? cemojo. com Box o? ce mojo. Movie network site with the most ecumenical box o? ce database on the Internet. ttp//www. boxofficemojo. com Marek Kre? mer, Jan Mati? ka c c Page 8 of 14 International movie revenues Determinants and impact of the ? nancial crisis appurtenance Descriptive statistics for the dependent variables Marek Kre? mer, Jan Mati? ka c c Page 9 of 14 International movie revenues Determinants and impact of the ? nancial crisis model 1 Regression of the original model promulgated in Terry, Cooley &038 Zachary, 2010 Marek Kre? mer, Jan Mati? ka c c Page 10 of 14 International movie revenues Determinants and impact of the ? nancial crisis Residuals versus ? tted values plotBreusch-Pagan test for heteroskedasticity Marek Kre? mer, Jan Mati? ka c c Page 11 of 14 International movie revenues Determinants and impact of the ? nancial crisis model 2 Regression of our model Marek Kre? mer, Jan Mati? ka c c Page 12 of 14 International movie revenues Determinants and impact of the ? nancial crisis Residuals versus ? tted values plot Breusch-Pagan test for heteroskedasticity Marek Kre? mer, Jan Mati? ka c c Page 13 of 14 International movie revenues Determinants and impact of the ? nancial crisis The correlation matrix Marek Kre? mer, Jan Mati? ka c c Page 14 of 14International Movie Revenues Determinants and Impact of the Financial CrisisInstitute of Economic Studies Faculty of Social Sciences Charles University in Prague Empirical Project Assignment Econometrics II Due on Friday, 13 January 2012, 11. 00 International movie revenues determinants and impact of the financial crisis Marek Kre? mer, Jan Mati? ka c c International movie revenues Determinants and impact of the ? nancial crisis Table of Contents Abstract Keywords Introduction Literature survey . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Data analysis variables used . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . model 1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . model 2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Results model 1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . model 2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Conclusion References primary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . secondary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . data sources . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Appendix Descriptive statistics for the dependent variables model 1 . . . . . . . . . . . . . . . . . . . . . . . Residuals versus ? tted values plot . . . . . Breusch-Pagan test for heteroskedasticity . model 2 . . . . . . . . . . . . . . . . . . . . . . . Residuals versus ? tted values plot . . . . . . Breusch-Pagan test for heteroskedasticity . The correlation matrix . . . . . . . . . . . . 2 2 2 2 3 3 4 4 4 4 6 6 6 7 8 8 8 8 9 9 10 11 11 12 13 13 14 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Marek Kre? mer, Jan Mati? ka c c Page 1 of 14 International movie revenues Determinants and impact of the ? nancial crisis Abstract This empirical project examines the determinants of international box o? ce revenues for movies produced in United States during 2006 2010. Our sample consists of 424 ? lms released in this period. We also test the hypothesis if the world ? nancial crisis had any signi? cant impact on the international box o? ce revenues. Keywords the ? ancial crisis, movie international box o? ce revenue, movies produced in the United States, budget, rating, Academy Awards, Introduction When choosing a topic of our empirical paper we were considering di? erent suggestions. As we both are pretty much interested in movies we ? nally decided to exit a viewer seat for a while and perform an empirical study on the movie industry. While being newcommers in sophisticated movie data analysis, we needed ? rst to get acquainted with important theoretical concepts and empirical papers concerning this topic. Literature survey When going down the history, Litman, 1983 was the ? st who has attempted to predict the ? nancial success of ? lms. He has performed a multiple regression and found a clear evidence that various independent variables have a signi? cant and serious in? uence on the ? nal success of a movie. Litemans work has been gradually getting developed, Faber &038 OGuinn, 1984 tested the in? uence of ? lm advertising. They proved, that movie critics and word-of-mouth are less important then movie previews and excerpts when explaininng movie succes after going on public. Eliashberg &038 Shugan, 1997 explored the impact of restricted-rating labeled movies on their box o? e performance. Terry, Butler &038 DeArmond, 2004 analysed the determinants of movie video rental revenue, ? nding Academy Award nominations as the dominant factor. King, 2007 followed their research and used U. S. movie data to ? nd the connection between the criticism and box o? ce earnings Many other authors has extended the initial work of Litman, 1983, but none of them has focused on the key factors of the international box o? ce revenues as we planned to. So we ? nally decided to use Terry, Cooley &038 Zachary, 2010 as our primary source. Their object of interest is very much similar to our resarch.Therefore we studied their metodology the most and we use their results in the analytical part as a primary resource of comparison. Marek Kre? mer, Jan Mati? ka c c Page 2 of 14 International movie revenues Determinants and impact of the ? nancial crisis Data We got quickly stucked realising that the strong majority of movie data on the internet are not free available. It was quite a surprise because there are many movie-oriented sites with seemingly deathless data access. But when there is a need of more profound, well structured and complete set of random data everything gets little bit tricky.After hours of searching, we luckily got to a 30 days free access to this kind of databases opusdata. com and got the core data for our analysis. Then we wanted to add some interesting or usefull variables just as the movie rating or the number of AcademyAwards to complete our dataset. It has been done using well known and free accessed databases imdb. com, numbers. com and boxo? cemojo. com. Thanks to our literature survey we discovered a model which we have thought would be interesting to test on di? erent or new data. The most interesting would be to test it on our domestic data but these are quite di? ult to obtain (as explained before). Anyway, it would be possible to get data for the highest grossing ? lms but that would violate the assumption of random sample. Therefore we decided to use data from U. S. and Canada which we considered the most likely to obtain. We also wanted to test whether the ? nancial crisis have had an impact on movie box o? ce revenues and whether the world ? nancial crisis made people less likely to go to the cinema. Model We considered several models and in the end we used two models. The ? rst one is just the same as the one used in paper Terry, Cooley &038 Zachary, 2010, but it is slightly modi? d by using di? erent data plus setting the crisis variable. We considered it as a dummy variable, which was 1 if the movie was released during crisis (2008-2009), otherwise it is equal to zero. As it was proposed before, this model has been used as a comparison to the original model Terry, Cooley &038 Zachary, 2010 wihle we wanted to test whether their inference holds up with slightly di? erent and newer data. In the second model we tried to use a slightly di? erent approach. We used a time series model with year dummies and we also used all the variables which we obtained and were statistically signi? ant. Our ? rst model is basic linear regression with cross-sectional data. Our data are a random sample thanks to opusdata. com query which was capable of selecting a random sample of movies. We have tested all the variables for multicolline arity with the correlation matrix and there is no proof for multicollinearity in our used variables. The only high collinearity is between domestic and budget variables, which is about 0. 75. After running the regressions we have used the Breusch-Pagan test for heteroscedasticity and the chi squared was really high therefore showing signs of strong heteroscedasticity.Even after looking at the graph of residuals against ? tted values it was clear that the heteroscedasticity is present. Therefore we had to run the regressions with the heteroscedasticity robust errors. We therefore tested in both models for presence of these the variables which have an impact on movie international box revenues any signi? cant impact of ? nancial crisis on these revenues Marek Kre? mer, Jan Mati? ka c c Page 3 of 14 International movie revenues Determinants and impact of the ? nancial crisis Data analysis Here we list all the used variables in both models and their a description. ariables used acade my awards . . . . . . . . . number of Academy Awards a ? lm earned action . . . . . . . . . . . . . . . . . . categorical variable for movies in action genre animation . . . . . . . . . . . . . . . categorical variable for movies in animation production method budget . . . . . . . . . . . . . . . . . . the estimated production and promotion cost of a movie comedy . . . . . . . . . . . . . . . . . . categorical variable for movies in comedy genre crisis . . . . . . . . . . . . . . . . . . dummy variable for movies released during crisis domestic . . . . . . . . . . . . . . . omestic box o? ce earnings horror . . . . . . . . . . . . . . . . . . categorical variable for movies in horror genre international . . . . . . . . . . . . international box o? ce earnings kids . . . . . . . . . . . . . . . . . . categorical variable for movies for children rating . . . . . . . . . . . . . . . . . . average user rating from the imdb. com source ratingR . . . . . . . . . . . . . . . . . . is a cat egorical variable for movies with a restricted rating romantic . . . . . . . . . . . . . . . . . . categorical variable for movies in romantic genre sequel . . . . . . . . . . . . . . . . . categorical variable for movies derived from a previously released ? lm y06 ? y10 . . . . . . . . . . . . . . . . . . dummy variable for movies released in a year The list of variables is followed by both model equations and reggression table comparism, while model 1 and model 2 mean the original Terry, Cooley &038 Zachary, 2010 model and our new model respectivelly. model 1 international = ? 0 + ? 1 domestic + ? 2 action + ? 3 kids + ? 4 ratingR+ + ? 5 sequel + ? 6 rating + ? 7 academy awards + ? 8 budget + ? 9 crisis model 2 international = + + ? 0 + ? 1 academy awards + ? 2 budget + ? 3 domestic + ? 4 sequel + ? horror + ? 6 romantic + ? 7 comedy + ? 8 action + ? 9 ratingR + ? 10 animation + ? 11 y06 + ? 12 y07 + ? 13 y08 + ? 14 y09 Marek Kre? mer, Jan Mati? ka c c Page 4 of 14 International m ovie revenues Determinants and impact of the ? nancial crisis Table 1 Model comparison model 1 domestic action kids rating R sequel rating academy awards budget crisis horror romantic comedy animation y 06 y 07 y 08 y 09 Constant Observations t statistics in parentheses ? model 2 1. 025??? (13. 31) -18. 56? (-2. 29) 1. 028??? (12. 70) -13. 43 (-1. 79) 48. 33? (2. 10) 5. 922 (1. 52) 26. 91? (2. 06) 0. 309 (1. 42) 6. 978? (2. 33) 0. 68??? (5. 48) -5. 320 (-1. 01) 9. 259? (2. 36) 28. 74? (2. 16) 7. 097?? (2. 59) 0. 508??? (4. 73) -9. 867? (-2. 23) 13. 41 (1. 79) -17. 77?? (-3. 31) 52. 02?? (2. 87) -7. 962 (-1. 24) 1. 182 (0. 17) -6. 748 (-1. 01) -11. 79 (-1. 30) -43. 25?? (-3. 05) 424 ??? -15. 11? (-2. 41) 424 p < 0. 05, ?? p < 0. 01, p < 0. 001 Marek Kre? mer, Jan Mati? ka c c Page 5 of 14 International movie revenues Determinants and impact of the ? nancial crisis Results model 1 After running the ? rst regression we get quite similar results as Terry, Cooley &038 Zachary, 20 10, so their inference holds up even under our data.The similar results we get are that one dollar in revenues in US makes $1. 02 in international revenues, therefore succesful movie in US is likely to be similarly succesful in international theatres, if movie is a sequel it adds to revenues about $26 mil. , every academy award adds about $7 mil. and every additional dollar spent on budget adds about $0. 57 so there is about 57% return on budget. We also have similarly insigni? cant variables which are whether is movie rated as restricted and how great or poorly is movie rated by critics or other people.That means that international audience is not in? uenced by age restrictions and critical movie ratings. When we look at our and theirs results regarding the genres then we get quite di? erent results. They say that when a movie is of an action genre then it adds about $26 mil. whereas we obtained results that revenues for an action movie should be lower about $13 mil. and our result for children movies is two times larger and it says that a children movie should make about $48 mil. more. It could be explained that movie genre preferences shifted in the last two years.But more likely explanation is the di? erence in our data in labeling the movies. In our data we have had more detailed labeling and movies which they had labeled as action movies, we had labeled adventure movies etc. Therefore the strictly action movie genre is not so probable to make money as it would seem. Action movies are usually of low quality and many of them could be labeled as B-movies which usually are not very likely to have high revenues. The children movies could be getting more popular and taking children to the movies could be getting more usual thing.Our last and new variable is the crisis dummy which is not signi? cant and therefore we have no proof that the ? nancial crisis had any e? ect on movie revenues. Our model has quite high R2 which is about 0. 83, that is even higher the n Terry, Cooley &038 Zachary, 2010 have. But the main reason behind this high R2 is that most of the variation in data is explained by US revenues. If we regress international revenues on domestic alone we still get high R2 which is about 0. 59. model 2 In our time series model we get quite similar results as in the ? rst one. We have there ? e new variables which are genres comedy, romantic and horror, animation dummy, which tells us whether the movie is animated or not and year dummies. Our model implies that when a movie is a comedy it will make about $17 mil. less in revenues, when horror about $10 mil. less, when romantic about $13 mil. more and when animated it will add about $52 mil to its revenues. The restricted rating is now also statistically signi? cant and it should add to the revenues about $9 mil. which is quite unexpected. Y ear dummies are statistically non-signi? cant and even when we test them for joint signi? ance they are jointly non-signi? cant. Therefore even in this model there appears no reason to believe that the ? nancial crisis or even year makes di? erence in the movie revenues. Marek Kre? mer, Jan Mati? ka c c Page 6 of 14 International movie revenues Determinants and impact of the ? nancial crisis Conclusion The inferences from our models are quite like we expected. We expected that people are more likely to go to cinema to see movies that had won academy awards, that were succesful in U. S. theatres and that are some kind of sequel to previous succesful movies. The resulting e? cts of di? erent movie genres could be quite puzzling but these e? ects depend highly on quality of the movies released these years and on the mood and taste of current society. If we had had larger sample with data from many years then it is possible that we would have seen trends in the di? erent movie genres. The insigni? cance of the ? nancial crisis on movie revenues was also likely because the severity of the crisis and impact on regular citizen ha s not been so large that it would in? uence his attendence of movie theatres. Marek Kre? mer, Jan Mati? ka c c Page 7 of 14International movie revenues Determinants and impact of the ? nancial crisis Reference primary Terry, Cooley &038 Zachary, 2010 Terry, Neil, John W. Cooley, &038 Miles Zachary (2010). The Determinants of Foreign Box O? ce Revenue for English Language Movies. Journal of International Business and Cultural Studies, 2 (1), 117-127. secondary Eliashberg &038 Shugan, 1997 Eliashberg, Jehoshua &038 Steven M. Shugan (1997). Film Critics In? uencers or Predictors? Journal of Marketing, 61, 68-78. Faber &038 OGuinn, 1984 Faber, Ronald &038 Thomas OGuinn (1984). E? ect of Media Advertising and Other Sources on Movie Selection.Journalism Quarterly, 61 (summer), 371-377. King, 2007 King, Timothy (2007). Does ? lm criticism a? ect box o? ce earnings? Evidence from movies released in the U. S. in 2003. Journal of Cultural Economics, 31, 171-186. Litman, 1983 Litman, Barry R. (1983). Predicting Success of Theatrical Movies An Empirical Study. Journal of Popular Culture, 16 (spring), 159-175. Ravid, 1999 Ravid, S. Abraham (1999). Information, Blockbusters, and Stars A Study of the Film Industry. Journal of Business, 72 (4), 463-492. Terry, Butler &038 DeArmond, 2004 Terry, Neil, Michael Butler &038 DeArno DeArmond (2004).The Economic Impact of Movie Critics on Box O? ce Performance. Academy of Marketing Studies Journal, 8 (1), page 61-73. data sources opusdata. com Opus data movie data through a query interface. 30-days free trial. http//www. opusdata. com/ imdb. com The Internet Movie Database (IMDb). The biggest, best, most award-winning movie site on the planet. http//www. imdb. com numbers. com The numbers. Box o? ce data, movies stars, idle speculation. http//www. the-numbers. com boxo? cemojo. com Box o? ce mojo. Movie web site with the most comprehensive box o? ce database on the Internet. ttp//www. boxofficemojo. com Marek Kre? mer, Jan Mati? ka c c Page 8 of 14 International movie revenues Determinants and impact of the ? nancial crisis Appendix Descriptive statistics for the dependent variables Marek Kre? mer, Jan Mati? ka c c Page 9 of 14 International movie revenues Determinants and impact of the ? nancial crisis model 1 Regression of the original model published in Terry, Cooley &038 Zachary, 2010 Marek Kre? mer, Jan Mati? ka c c Page 10 of 14 International movie revenues Determinants and impact of the ? nancial crisis Residuals versus ? tted values plotBreusch-Pagan test for heteroskedasticity Marek Kre? mer, Jan Mati? ka c c Page 11 of 14 International movie revenues Determinants and impact of the ? nancial crisis model 2 Regression of our model Marek Kre? mer, Jan Mati? ka c c Page 12 of 14 International movie revenues Determinants and impact of the ? nancial crisis Residuals versus ? tted values plot Breusch-Pagan test for heteroskedasticity Marek Kre? mer, Jan Mati? ka c c Page 13 of 14 International movie r evenues Determinants and impact of the ? nancial crisis The correlation matrix Marek Kre? mer, Jan Mati? ka c c Page 14 of 14