Effects of Oscar awards on movie production

Agnani, Betty; Aray, Henry Working Paper Effects of Oscar awards on movie production Economics Discussion Papers, No. 2016-8 Provided in Cooperation with: Kiel Institute for the World Economy – Leibniz Center for Research on Global Economic Challenges Suggested Citation: Agnani, Betty; Aray, Henry (2016) : Effects of Oscar awards on movie production, Economics Discussion Papers, No. 2016-8, Kiel Institute for the World Economy (IfW), Kiel This Version is available at: https://hdl.handle.net/10419/128475 Standard-Nutzungsbedingungen: Die Dokumente auf EconStor dürfen zu eigenen wissenschaftlichen Zwecken und zum Privatgebrauch gespeichert und kopiert werden. Sie dürfen die Dokumente nicht für öffentliche oder kommerzielle Zwecke vervielfältigen, öffentlich ausstellen, öffentlich zugänglich machen, vertreiben oder anderweitig nutzen. Sofern die Verfasser die Dokumente unter Open-Content-Lizenzen (insbesondere CC-Lizenzen) zur Verfügung gestellt haben sollten, gelten abweichend von diesen Nutzungsbedingungen die in der dort genannten Lizenz gewährten Nutzungsrechte. Terms of use: Documents in EconStor may be saved and copied for your personal and scholarly purposes. You are not to copy documents for public or commercial purposes, to exhibit the documents publicly, to make them publicly available on the internet, or to distribute or otherwise use the documents in public. If the documents have been made available under an Open Content Licence (especially Creative Commons Licences), you may exercise further usage rights as specified in the indicated licence. http://creativecommons.org/licenses/by/3.0/ Received January 19, 2016 Accepted as Economics Discussion Paper February 18, 2016 Published February 19, 2016 © Author(s) 2016. Licensed under the Creative Commons License - Attribution 3.0 Discussion Paper No. 2016-8 | February 19, 2016 | http://www.economics-ejournal.org/economics/discussionpapers/2016-8 Effects of Oscar Awards on Movie Production Betty Agnani and Henry Aray Abstract This article tests the effects of Oscar awards on the production of feature films. Time series data for Spain over the 1953–2014 period are used and a production function is estimated assuming that the Oscar effects accrue through the total factor productivity. A lag structure is introduced which allows for a general specification so that the Oscar awards could have constant or diminishing effects over time. The results show that the Oscar wins have significant positive effects on movie production and that some of them have caused structural breaks, while others have vanishing effects over time. The results are fairly robust to the introduction of control variables and different methods of estimation. JEL Z10 L82 C13 Keywords Movie production; Oscar awards; Cobb-Douglas Production Function Authors Betty Agnani, University of Granada, University of Granada, Department of Economics, Campus de la Cartuja S/N, Spain, [email protected] Henry Aray, University of Granada, Department of Economics, Campus de la Cartuja S/N, Spain, [email protected] The authors gratefully acknowledge financial support from the Spanish Ministry of Education and Science through Project MICINN – ECO2011–25737. Citation Betty Agnani and Henry Aray (2016). Effects of Oscar Awards on Movie Production. Economics Discussion Papers, No 2016-8, Kiel Institute for the World Economy. http://www.economics-ejournal.org/economics/ discussionpapers/2016-8 1 Introduction The increasing academic research on the motion picture industry as collected in the surveys by Hadida (2008) and McKenzie (2012) reveals that researchers are typically interested in the exhibition of films, i.e. the financial performance of films or demand for cinema attendance. In fact, most of the articles summarized by McKenzie (2012) have to do with factors explaining thedemandsideorfinancial success of a film such as the role of stars, critics, reviews, awards, nominations, ratings and genres. Thus, many of them estimate demand functions. Hadida (2008) pointed out that many empirical illustrations of film performance are limited to total domestic box office revenues. Regarding production, contribution is, in general, limited to organizational and financial factors related to the production and distribution process. Unlike the traditional literature, we focus on the supply side. Specifically, we are interested in testing what effects Oscar awards may have on the production of Spanish feature films. In spite of the importance normally attributed to the Oscar awards in cinematography, there are no articles in the relevant literature that quantitatively measure their impact on movie production. However, the effects of Oscar nominations and awards have been tested on the financial success of a movie by Nelson et al. (2001), Deuchert et al. (2005), Hennig- 2 Thurau et al. (2007) and Lee (2009), who generally found positive effects. It is well known that each year most Oscars go to US film producers, leaving the remainder of the world with a relatively low number of nominations and awards. Indeed, our interest in testing the effect of Oscar awards outside the US is justified precisely because the countries (other than theUS)thatareawardedOscarsvary across the years. An Oscar award could be more important for the industry of those countries than for the US industry itself. Moreover, winning an Oscar could be interpreted as a positive expectation in general by motion picture producers in such countries. In our specific case, domestic and foreign demand for Spanish films might be expected to rise, which would imply higher expected profits for the domestic industry. Therefore, producers should be prepared to satisfy that increasing demand. Furthermore, winning an Oscar may also be important in that it could attract not only foreign investment to the domestic industry, but also foreign technology. Simonton (2004) pointed out that Oscar awards provided meaningful information about cinematic creativity and achievement. In fact, a country that wins an Oscar could be thought to be endowed with skill factors in the movie industry. Related to this or not, the international trend in movie production shows an increase in filmsmadebymorethanone country as pointed out by Hoskins et al. (1997), who highlight the increase in co-productions between Europe and Canada; and by McCalman (2004), who claims that higher foreign direct investment in the movie production 3 industry leads to increased collaboration between countries. This article is related to Agnani and Aray (2010), who used panel data regression to test the effects of subsidies and international awards on Spanish movie production. They found that awards positively affect the productivity of the movie production industry, while subsidies have no effect. However, this article differs from Agnani and Aray (2010) in four important aspects. First, it focuses specifically on the Academy Awards (Oscars) due to the paramount importance typically attributed to them around the world. Winning an Oscar contributes to the worldwide impact of a film more than any other award by producing not only an increase in box office revenues as pointed out in the above literature, but might also attract funds and technology to the industry through new sponsors, producers or partners. Therefore, this is precisely the contribution of this paper, to look for an Oscar effect on the supply side instead of the demand side. Second, although a production function is also estimated, in this article we use time series data of total feature film production rather than panel data. Third, the time series approach allows us to consider that the impact of an Oscar award on movie production could be constant and persistent or vanishing over time. We therefore specify a sufficiently general model with a lag structure that permits us to determine the decay rates of each Oscar effect. And fourth, we control for the main changes in legislation as suggested by the history of the Spanish cinema industry and for the impacts of television and video. 4 The empirical results can be summarized as follows. Strong and robust evidence supporting the existence of positive Oscar effects on Spanish movie production is found. The general specification proposed in this article suggests that some Oscars might have caused structural breaks in the industry, while others might have had vanishing effects. The rest of the paper is organized as follows. In the following section an overview of the data is presented. In the third section, we specify the econometric model to be estimated. Section 4 presents the main results. The robustness check of the model is shown in Section 5. Finally, conclusions are presented in Section 6. 2OverviewofData Our data were drawn from the Estadísticas de Cine y Audiovisuales (Cinema and Audiovisual Statistics) report published by the Spanish Ministry of Education, Culture and Sport. We concentrate on the 1953-2014 period using annual data on the total production of feature films. Therefore, we include Spanish films and films produced jointly with foreign partners (coproductions). According to the Spanish Ministry of Education, Culture and Sport, a film is considered a "Spanish film" if it is made by a Spanish or European firmlocatedinSpain,whichfulfills the following requirements: 75 percent of the authors (director, screenwriter, director of photography and music 5 composer), players and the rest of the artists, as well as the creative and technical staffmust be Spanish citizens, European Union citizens or citizens of any other European state holding an agreement with the European Union Economic Area, or have a Spanish residency permit or a residency permit of any of these states. In any case, the director of the film must fulfill such requirements. Moreover, the language of the films should be Spanish or any other official language of Spain. The filming, except screenwriting, postproduction and laboratory work must be carried out in the European Union. According to the Spanish Ministry of Education, Culture and Sport, a film is said to be a co-production with one partner whenever the share of the Spanish participation is 20 to 80 percent of the production cost of the film. Moreover, in the case of multiple partners, participation must be 10 to 70 percent. In addition, the participation of artists and technical staff must be proportional to the economic participation. In general, economic participation must not exceed 50 percent. Table 1 shows the basic statistics for the series of number of feature films and firms involved in film production. As can be seen, Spain produces, on average, 108 films per year with a deviation of 46 films. On average, 74 completely Spanish films are produced with a deviation of 33, while there are 34 co-productions on average with a standard deviation of 21. Regarding firmsinvolvedintheproductionoffilms, the mean is 92 with a deviation of 6 67. Figure 1 plots the evolution of Spanish movie production and shows the year (shaded) of Oscar awards. It can be observed that the production of films has fluctuated considerably over time. We plot the total production and disaggregate it into completely Spanish films and co-productions. In general, most of the films produced in Spain were completely Spanish films; a trend that has grown in recent years. In fact, total production and completely Spanish production followed a similar pattern over the sample period, while co-productions followed a different pattern. After joining the European Union, Spain became a more open country, although co-productions have not increased on a par with completely Spanish productions. 7 Years Production (units) 1953 1959 1965 1971 1977 1983 1989 1995 2001 2007 2013 0 50 100 150 200 250 Total Spanish coproductions Figure 1. Evolution of the Spanish Production of Feature Films. According to Figure 1, the series of production of feature films are suspected to have unit roots. Therefore, we perform unit root tests considering the following specifications Yt=C+ρYt−1+µt Log (Yt)=C∗+ρ∗Log (Yt−1)+µ∗ t Where Ytis the production of feature filmsineachperiodt,Cand C∗ are constants, and µtand µ∗ tare random disturbances. Table 1 shows the 8 above, we introduce digital television since it has enlarged the TV supply, especially in terms of movies and, of course, movies produced in Spain. Let uscallDigitalTVadummyvariablethattakesthevalueofonefrom2004 onwards. Finally, εtis a random disturbance. Taking the natural logarithm to equation (1) we obtain Log (Yt)=Log (At)+αLog (Nt) As shown in the previous section, the series Ytand Log (Yt)have a unit root. Therefore, in order to avoid spurious regression, we consider ∆Log (Yt)=∆Log (At)+α∆Log (Nt)(3) Substituting (2) in (3) and rewriting we obtain ∆Log (Yt)=δ+D0 1tB1+D0 2tB2+D0 3tB3+α∆Log (Nt)+εt(4) A drawback to the specification in equation (4) is that the Oscars would have constant effects only in the years after the announcement, but no effects for the rest of the sample period. However, Oscar awards could be expected to have lagged effects since production might react slowly to such awards. Therefore, to overcome this drawback, we incorporate a simple lag structure àlaKoyck (1954) into the equation (4), which allows for a more general model 15 ∆Log (Yt)=δ+ K X k=0 D0 1t−kB1,k +D0 2tB2+D0 3tB3+α∆Log (Nt)+εt(5) where B1,k =ΠB1,k−1=Π2B1,k−2=.... =ΠkB1,0with Πbeing a (n×n) diagonal matrix showing the decay rates of the distributed lag structure and whose value falls in the interval [0,1]. In order to estimate the equation (5) for values of the diagonal components of matrix Πin the interval [0,1], we construct the vector of auxiliary variables, V0 1t=PK k=0 D0 1t−kΠk,12 andrewritetheequation(5)as follows ∆Log (Yt)=δ+V0 1tB1,0+D0 2tB2+D0 3tB3+α∆Log (Nt)+εt(6) Notice that we consider a single lag structure, K,forthefive dummy variables included in D1tand know that the Oscar is awarded on different dates. Thus, it is natural to think that each dummy variable in D1tshould have its own lag structure. However, it is straightforward to see that with the specification of the dummy variables, whenever we consider a lag structure that is equal to or larger than the total lag structure of the Oscar for 1983, we obtain the same vector of the auxiliary variables, V1t.In fact, our model canevenbeinterpretedashavinganinfinite lag structure since we know 12See Appendix. 16 that the dummy variables take the value of zero before the beginning of our sample period. The specification of the equation (6) is general enough since it allows each Oscar award to have a different persistent effect over time due to the parameters included in vector B1,0and the decay rates included in Π. Although the parameters included in B1,0are equal, the Oscar effects could evolve differently over time due to the different decay rates in Π.Conversely, if the decay rates are equal, the effects could evolve differently over time due to the different parameters included in B1,0. It can be also noticed that equation (6) nests the opposite cases of Oscar effects only in the years after the announcement, and fully persistent constant Oscar effects over time. Thus, whenever K=0,wehavethemodelofthe equation (4) with V1t=D1tand B1,0=B1.13 On the other hand, if all the components of the diagonal of the matrix Πare equal to one, the vector of auxiliary variables, V1t, becomes a vector of dummy variables that take the value of one from the period following the year in which the industry wins an Oscar to the end of the sample period, and zero otherwise. In this case, the effects of the Oscars are constant over time and persist forever. Thus, whenever any of the components of the diagonal of the matrix Πis in the interval (0,1),theeffect of the Oscar award associated to that component will be decreasing over time, which is a plausible case, since it recognizes the 13Mathematically, in the specificcaseofK=0and Π=0nxn,wehavean indetermination since we obtain 00. However, we know that whenever K=0,V1t=D1t. 17 lagged effects of an Oscar award, but it also considers that such effects could vanish over time. We have to estimate the parameters in vectors B1,0,B2,B3,δ,andαfor the different combinations of the decay rates included in Π. 4 Estimation Issues Assuming that the industry produces with a combined input of physical capital and labor is actually a very strong assumption and therefore deserves an explanation. When attempting to perform the econometric estimation, we encounter a problem: data for physical capital and labor inputs are not available. In order to overcome thisproblem,wehavetorelyonproxies and make very restrictive assumptions. Therefore, we propose two different measures to proxy Nt. Since it is true that the greater the number of firms in the industry, the higher the physical capital and labor employed in producing films, the number of firms can be thought of as an input that combines the physical capital and labor input in the industry. Let N0 tbe the number of Spanish firms with positive production in the industry in each period t. This assumption is supported by the fact that about 80 percent of the production in the Spanish movie industry is done by firmsthatonlyproduceonefilm. In the history of the Spanish film industry, however, there have been several examples of firmsthatwerecreatedonlytoproduceonespecificfilm or directors who created a firm only to produce their own films. Hence, it is 18 difficult for the variable number of firms to show heterogeneity across firms. Therefore, correlations with the error term might be expected. In order to overcome this problem, we propose the following adjusted variable: N1 t= 3 X i=1 nitexiwifor i=1,2,3(7) where wi= 2014 P t=1995 nit 2014 P t=1995 N0 t for i=1,2,3 And nit is the number of firms that produce xifilms each year. Thus, n1t is the total number of firms that produced only one film (x1=1)inyeart, n2tis the total number of firms that produced between 2 and 4 films in year t(we consider, x2=3)andn3tis the total number of firms that produced more than 5 films (x3=5)inyeart.14 Notice that wi,fori=1,2,3, are weights that were calculated using data for the 1995-2014 period when data for the groups of firms are available. Thus, we use those weights for the entire period of estimation. In line with that, the actual values of n1t,n2tand n3tfor the 1953-1994 period are not available. In order to overcome this problem, we proxy the values for each year toftheperiodasfollows: 14Notice that for n2twe use the central point of the interval of production in this group, that is, x2=3.Forn3twe use the lowest value of the interval of this group, x3=5,since we understand that most of the Spanish firms included in this group are not large enough toproducemorethanfive films on average. 19 nit =wiN0 tfor i=1,2,3and 1953 ≤t<1995 The equation (7) states that the combined input of physical capital and labor is the summatory of three kinds of inputs, thus allowing us to account for heterogeneity in the industry. In fact, each input, nit, is augmented by an exponential factor that is precisely what allows for firms’ heterogeneity when considering their relative weights in the industry. Moreover, we should not neglect the participation of foreign partners that contribute physical capital and human resources to the industry. Therefore, we have to adjust N1 tin order to include the foreign input as follows NC1 t=N1 t(1 + WCt)(8) where WC t=Number of co-productions in period t Totalproductioninperiodt According to (8), foreign partners contribute an additional input equivalent to the share of the number of co-productions on the total film production. We therefore show estimations for the two proxies of the combined input given by N1 tand NC1 t.15 15We also run regressions using N0 tand other alternative measures such as N2 t=N0 te(S3 i=1 xiwi),NC2 t=N2 t(1 + WCt),N3 t=N0 t³P3 i=1 xiwi´and NC3 t= 20 The parameters of equation (6) are estimated using maximum likelihood estimation controlling for heteroskedasticity and autocorrelation by means of a covariance matrix àlaNewey and West (1987) and considering increments of 0.1 in the components of the diagonal of the matrix Πin the interval [0,1]. Since equation (6) has many parameters to be estimated with respect to the available number of observations, we propose an estimation procedure that involves including groups of variables in each step. Thus, we include a constant, the combined input and the Oscar variables in the first step of the estimation procedure. In the second step, we add the variables capturing legislation. In the third step, we delete the non-significant variables controlling for legislation in the second step and add the variables that capture the effects of television and video. Finally, in the fourth step, we deleted the non-significant control variables of the estimation in the third step and reestimated the model including only the significant ones. In using this procedure, the maximum number of parameters that we estimated were 16. However, the total number of parameters that would have to be estimated if we included all explanatory variables at once would be 19. An additional advantageofthisprocedureisthatitallowsshowingtherobustnessofthe Oscar effects as the control variables are included. Thus, for each proxy for the combined input Ntand in each step we carry out 161,051 (11n) regressions and choose the one which provides the highest value of the (1 + WCt)N3 t. Most of the empirical results hold. Available upon request. 21 maximum likelihood function.16 Table 2 shows the estimation of equation (6) for Nt=N1 t.Accordingto the criterion pointed out above and looking at the final step of the estimation procedure (fifth column), we obtain that the components of the diagonal of Πare (0.2,1,1,1,1), i.e. the decay rates for each Oscar are π1983 =0.2 and π1994 =π2000 =π2003 =π2005 =1. All the estimated coefficients of theOscarshavepositivesignsandaresignificant at any conventional level, except the Oscar for To Begin Again. It can be noticed that the Oscar effects are fairly robust to the inclusion of the control variables. The decay rates for the significant ones suggest that the Oscars have constant effects over time and persist forever. Thus, those Oscars can be interpreted as having caused structural breaks in Spanish movie production. The highest Oscar effect is that of Talk to Her (0.1378), followed by the Oscar for Belle Epoque (0.0659), All About My Mother (0.0388) and The Sea Inside (0.0261). Table 3 shows the estimation of equation (6) for Nt=NC1 t.The components of the diagonal of Πare (0.4,0,0.6,0.4,1). The estimated coefficients of the Oscars for Talk to Her and The Sea Inside have positive signs and are significant at any conventional level with decay rates π2003 =0.4 and π2005 =1, respectively. Therefore, the positive effect of the Oscar for Talk to Her vanishes at a moderate rate, while that of The Sea Inside suggests 16GLS regressions were also carried out with very similar results (available upon request). The criterion was the highest R2. Nevertheless, the maximum likelihood estimator is the efficient one since Vit is a vector of generated regressors. 22 astructural breaks in Spanish movie production. The Oscars for To Begin Again and All About My Mother are not significant at any conventional level. And we have found an unexpected result for the Oscar for Belle Epoque since that the estimated coefficient is negative and significant at any conventional level. However, notice that the decay rate is π1994 =0, which suggests that thenegativeeffect arises only in the first year. Again, results are fairly robust to the inclusion of the control variables. Changes with respect to the results in Table 2 can be seen in all decay rates, except for The Sea Inside.The highest Oscar effect in the firstperiodisagainthatofTalk to Her (0.1826), followed by the Oscar for The Sea Inside (0.1281). According to the results of Tables 2 and 3, we can say, in general, that Oscar awards have positive impacts on the production of films. Figure 2 plots the Oscar effects over time for those which are significant with positive signs. In the upper illustration we show the case when Nt=N1 t. Since the decay rates are π1994 =π2000 =π2003 =π2005 =1,theeffects are constant over time. In the lower figure, we show the case when Nt=NC1 t. It is important to note the constant effect of The Sea Inside and the rapidly vanishing effect of Talk to Her since π2003 =0.4. In fact, the median lag of Talk to Her is -0.7565, meaning that 50 percent of the total Oscar effect is conveyed in about one year.17 After that, the rest of the effect vanishes quicklyovertimeandinabout5yearsitisveryclosetozero. 17The median lag of the Koyck model is −log (2) /log (πτ)where τdenotes the years of Oscar wins. 23 1995 1997 1999 2001 2003 2005 2007 2009 2011 2013 0.00 0.05 0.10 0.15 0.20 The Sea Inside Talk to Her All About My Mother Belle Epoque 1995 1997 1999 2001 2003 2005 2007 2009 2011 2013 0.00 0.05 0.10 0.15 0.20 The Sea Inside Talk to Her Figure 2. Oscar effects over time. Our lag structure allowed us to estimate the total Oscar effects over time, which can be calculated as ∞ X k=0 B1,k =(In×n−Π)−1B1,0 where In×nis an (n×n) identity matrix. This can also be understood as a long-term multiplier. 24 5.2 Controlling for Endogeneity The proxies used for the combined input could introduce endogeneity problems. We therefore reestimate only the previous specification by twostage least squares (2SLS). The results are shown in Table (5).20 Some changes can be seen. However, the evidence in favor of positive Oscar effects hold, although they are weaker. When N1 tis used as the combined input, 3 out of the 5 Oscar wins turn out to be significant: All About My Mother and Talk to Her at the 1% level and The Sea Inside at the 10% level. When NC1 tis used as the combined input, all Oscar wins turn out to be significant. Notice also in Table (5) that most of the estimations for the controller are significant in both cases. The Hausman endogeneity test shows unusual results since it turns to be negative, which could be interpreted as meaning that no endogeneity problems were found. Moreover, the results for the Sargan test show that the instruments were valid. 6Conclusions This article tests the effect of awards on movie production. We use the Oscars since they are considered to be the most important awards worldwide. We 20The instruments in the second column were ∆Log ¡N1 t−1¢,Log ¡N1 t−1¢,andtherestof explanatory variables in such a column. Anagously, the instruments in the third column were ∆Log ¡NC1 t−1¢,Log ¡NC1 t−1¢, and the rest of explanatory variables in the same column. 31 estimate a production function using time series data on the production of Spanish feature films and considering a lag structure that allows the Oscars to have constant or diminishing effects over time. We show strong evidence supporting the existence of positive Oscar effects on the productivity of Spanish movie production. The general specification proposed in this article suggests that some Oscars might have caused structural break in Spanish movie production. Strikingly, reforms in the Spanish film industry have generally had no effects or negative effects. The results are fairly robust to different measures of the input of the production function, methods of estimations and the introduction of control variables. 32 Appendix We have that K X k=0 D0 1t−kB1,k = K X k=0 D0 1t−k¡ΠkB1,0¢A.1 Let us define the vector of Oscar, D1t−k, the vector of parameter, B1,0, and the matrix Πas D1t−k= ⎡ ⎢ ⎢ ⎢ ⎢ ⎣ d1t−k d2t−k d3t−k d4t−k d5t−k ⎤ ⎥ ⎥ ⎥ ⎥ ⎦ B1,0= ⎡ ⎢ ⎢ ⎢ ⎢ ⎣ b10 b20 b30 b40 b50 ⎤ ⎥ ⎥ ⎥ ⎥ ⎦ Π= ⎡ ⎢ ⎢ ⎢ ⎢ ⎣ π10000 0π2000 00π300 000π40 0000π5 ⎤ ⎥ ⎥ ⎥ ⎥ ⎦ where d1t−kis a dummy variable for the Oscar won in 1983, d2t−kfor that of 1994, and so on. The same applies for B1,0and Π. Substituting in A.1 we obtain 33 K X k=0 D0 1t−k¡ΠkB1,0¢= K X k=0 £d1t−kd2t−kd3t−kd4t−kd5t−k¤× ⎛ ⎜ ⎜ ⎜ ⎜ ⎜ ⎝ ⎡ ⎢ ⎢ ⎢ ⎢ ⎣ π10000 0π2000 00π300 000π40 0000π5 ⎤ ⎥ ⎥ ⎥ ⎥ ⎦ k⎡ ⎢ ⎢ ⎢ ⎢ ⎣ b10 b20 b30 b40 b50 ⎤ ⎥ ⎥ ⎥ ⎥ ⎦ ⎞ ⎟ ⎟ ⎟ ⎟ ⎟ ⎠ = K X k=0 ¡d1t−kπk 1b10+d2t−kπk 2b20+d3t−kπk 3b30+d4t−kπk 4b40+d5t−kπk 5b50¢ = ⎛ ⎜ ⎜ ⎜ ⎜ ⎜ ⎝ K X k=0 £d1t−kd2t−kd3t−kd4t−kd5t−k¤⎡ ⎢ ⎢ ⎢ ⎢ ⎣ π10000 0π2000 00π300 000π40 0000π5 ⎤ ⎥ ⎥ ⎥ ⎥ ⎦ k⎞ ⎟ ⎟ ⎟ ⎟ ⎟ ⎠ × ⎡ ⎢ ⎢ ⎢ ⎢ ⎣ b10 b20 b30 b40 b50 ⎤ ⎥ ⎥ ⎥ ⎥ ⎦ =ÃK X k=0 D0 1t−kΠk!B1,0=V0 1tB1,0 where PK k=0 D0 1t−kΠk=V0 1t. 34 References [1] Agnani, B. and Aray, H. (2010). Subsidies and Awards in Movie Production. Applied Economic Letters 17(13-15), 1509-1511. [2] Deuchert, E., Adjamah, K., and Florian P. (2005). For Oscar Glory or Oscar Money. Journal of Cultural Economics 29(3), 159-176. [3] Fernández-Blanco, V. and Baños-Pino, J. F. (1997). Cinema Demand in Spain: A Cointegration Analysis. Journal of Cultural Economics 21(1), 57-75. [4] Hadida, A. L. (2008). Motion Picture Performance: A Review and Research Agenda. International Journal of Management Reviews 11(3), 297-335. [5] Hennig-Thurau, T., Houston, M. B. and Walsh, G. (2007). Determinants of Motion Picture Box Office and Profitability: An Interrelationship Approach. Review of Managerial Science 1(1), 65-92. [6] Hoskins, C., McFayden, S., Finn, A. and Jackel, A. (1997). Evidence on the Performance of Canada/Europe Co-Productions in Television and Film. Journal of Cultural Economics 21(2), 129-138. [7] Koyck, L. (1954). Distributed Lag and Investment Analysis. Amsterdam. North Holland. 35 [8] Lee, F. (2009). Cultural discount of cinematic achievement: the academy awards and U.S. movies’ East Asian box office. Journal of Cultural Economics 33(4), 239-263. [9] McCalman, P. (2004). Foreign Direct Investment and Intellectual Property Rights: Evidence from Hollywood’s Global Distribution of Movies and Videos. Journal of International Economics, 62(1), 107-123. [10] McKenzie, J. (2012). The Economics of Movies: A Literature Survey. Journal of Economic Surveys 26(1), 42-70. [11] Nelson, R. A., Donihue, R., Waldman, D.M. and Wheaton, C. (2001). What’s an Oscar Worth? Economic Inquiry 39(1), 1-16. [12] Newey, W. and West, K. (1987). A Simple, Positive Definite, Heteroskedasticity and Autocorrelation Consistent Covariance Matrix. Econometrica 55(3), 703-708. [13] Simonton, D. K. (2004). Film Awards as Indicators of Cinematic Creativity and Achievement: A Quantitative Comparison of the Oscars and Six Alternatives. Creativity Research Journal 16(2-3), 163-172. 36 37 Table 1: Preliminary Data Analysis: 1953-2014 period Basic statistics Total Spanish Co-productions Firms Sample Mean 108.2258 73.9839 34.2419 91.8548 Standard Deviation 45.7245 33.3213 21.1208 66.5291 Unit Root Tests Yt=C+ρY t−1+µt DF -1.1727 -1.1331 -2.2669 1.2375 PP -1.1924 -1.1522 -2.3050 1.2583 Log (Yt)=C∗+ρ∗Log Yt−1+µ∗ t DF -1.9392 -2.145 -2.6194 -0.7921 PP -1.9718 -2.1810 -2.6634 -0.8054 P(ˆτ<−3.22) = 0.025,T=50.P(ˆτ<−3.17) = 0.025,T= 100. Table 2: Estimation of the Equation (6) Using N1 tas the input in the production function Variable Estimation 1 Estimation 2 Estimation 3 Estimation 4 Π1=diag (1,0.8,0,0,1) Π1=diag (0,0.9,1,1,1) Π1=diag (0.9,1,1,1,1) Π1=diag (0.2,1,1,1,1) Constant 0.0091 0.0056 0.0139 0.0136 (0.0151) (0.0159) (0.0152) (0.0138) ∆Log (Nt)0.8097∗∗∗ 0.8143∗∗∗ 0.7824∗∗∗ 0.7889∗∗∗ (0.1325) (0.1524) (0.1667) (0.1589) Oscar_83 -0.0547∗-0.0477∗∗∗ 0.0697∗0.0135 (0.0325) (0.0003) (0.0390) (0.0441) Oscar_94 0.0936 0.0861∗∗ 0.0708∗∗∗ 0.0659∗∗∗ (0.0970) (0.0395) (0.0174) (0.0169) Oscar_00 0.0877∗∗∗ 0.0549∗∗∗ 0.0435∗∗∗ 0.0388∗∗∗ (0.0292) (0.0120) (0.0118) (0.0113) Oscar_03 0.0763∗∗∗ 0.1343∗∗∗ 0.1410∗∗∗ 0.1378∗∗∗ (0.0291) (0.0038) (0.0047) (0.0036) Oscar_05 0.0374 0.0303∗∗∗ 0.0272∗∗∗ 0.0261∗∗∗ (0.0547) (0.0054) (0.0065) (0.0053) MO 1964 0.0121 (0.0640) Miró Act -0.0486∗∗ -0.0392 (0.0236) (0.0516) RD 1282/1989 -0.0253 (0.0207) Act 17/1994 -0.0397 (0.0267) Act 15/2001 -0.1858∗∗∗ -0.1381∗∗∗ -0.1419∗∗∗ (0.0187) (0.0187) (0.0178) RD 1652/2004 -0.0801∗∗∗ -0.0842∗∗∗ -0.0847∗∗∗ (0.0041) (0.0050) (0.0037) Act 55/2007 -0.1630∗∗∗ -0.1253∗∗∗ -0.1309∗∗∗ (0.0094) (0.0098) (0.0092) Video -0.1331∗∗∗ -0.1224∗∗∗ (0.0099) (0.0094) TV 0.0604∗∗∗ 0.0710∗∗∗ (0.0102) (0.0095) Digital TV -0.0186∗∗∗ -0.0158∗∗∗ (0.0057) (0.0050) ˆσ20.0131∗∗∗ 0.0129∗∗∗ 0.0127∗∗∗ 0.0127∗∗∗ (0.0051) (0.0046) (0.0043) (0.0040) R20.6519 0.6583 0.6639 0.6633 H12.0626 (0.1510) 1.4857 (0.2229) 1.7051 (0.1916) 1.7642 (0.1841) Standard errors in parentheses. ***, **, * Significant at 1%, 5% and 10% levels, respectively 38 Table 3: Estimation of the Equation (6) Using NC1 tas the input in the production function Variable Estimation 1 Estimation 2 Estimation 3 Estimation 4 Π1=diag (0.9,1,1,0,1) Π1=diag (0.9,0,1,1,0.4) Π1=diag (0.4,0,0.6,0.4,1) Π1=diag (0.4,0,0.6,0.4,1) Constant 0.0094 0.0029 0.0148 0.0148 (0.0116) (0.0157) (0.0159) (0.0146) ∆Log (Nt)0.7401∗∗∗ 0.7716∗∗∗ 0.7389∗∗∗ 0.7393∗∗∗ (0.1085) (0.1392) (0.1485) (0.1451) Oscar_83 -0.0827∗∗∗ -0.1860∗∗∗ 0.0282 0.0273 (0.0233) (0.0270) (0.0274) (0.0408) Oscar_94 0.0294∗-0.1036∗∗∗ -0.0748∗∗∗ -0.0758∗∗∗ (0.0152) (0.0001) (0.0001) (0.0124) Oscar_00 -0.0701∗∗∗ 0.0251∗∗∗ 0.0283 0.0281 (0.0174) (0.0093) (0.0267) (0.0202) Oscar_03 0.1376∗∗∗ 0.1703∗∗∗ 0.1828∗∗∗ 0.1826∗∗∗ (0.0001) (0.0067) (0.0059) (0.0085) Oscar_05 0.0483∗∗ 0.0901∗∗∗ 0.1281∗∗∗ 0.1281∗∗∗ (0.0219) (0.0202) (0.0058) (0.0079) MO 1964 0.0259 (0.0579) Miró Act 0.0965∗∗∗ -0.0008 (0.0283) (0.0343) RD 1282/1989 0.0392 (0.0296) Act 17/1994 0.0530∗∗ 0.0664∗∗ 0.0668∗∗ (0.0256) (0.0307) (0.0297) Act 15/2001 -0.0834∗∗∗ -0.0482∗∗ -0.0480∗∗ (0.0215) (0.0221) (0.0198) RD 1652/2004 -0.1611∗∗∗ -0.0520∗∗∗ -0.0521∗∗∗ (0.0073) (0.0052) (0.0076) Act 55/2007 -0.0175∗∗ -0.0210∗∗ -0.0208∗ (0.0086) (0.0096) (0.0117) Video -0.1353∗∗∗ -0.1351∗∗∗ (0.0081) (0.0082) TV 0.0773∗∗∗ 0.0768∗∗∗ (0.0081) (0.0088) Digital TV -0.0003 (0.0067) ˆσ20.0147∗∗∗ 0.0143∗∗∗ 0.0142∗∗∗ 0.0142∗∗∗ (0.0048) (0.0049) (0.0048) 0.0048 R20.6095 0.6203 0.6117 0.6250 H15.7366 (0.0166) 2.6934 (0.1008) 3.0919 (0.0787) 3.2278 (0.0724) Standard errors in parentheses. ***, **, * Significant at 1%, 5% and 10% levels, respectively 39 Table 4: Estimation of the Equation (6) introducing a dummy for the following years of Oscar wins Variable Estimation using N1 tEstimation using NC1 t Π1=diag (0.1,0.9,1,1,1) Π1=diag (1,0.1,1,0,0) Constant 0.0137 0.0156 (0.0146) (0.0158) ∆Log (Nt)0.7876∗∗∗ 0.7167∗∗∗ (0.0746) (0.0853) Oscar_83 0.0363 1.2671∗∗∗ (0.0221) (0.0140) Oscar_94 0.0890∗∗ 1.1964∗∗∗ (0.0406) (0.0502) Oscar_00 0.0807∗∗∗ 1.2897∗∗∗ (0.0099) (0.0244) Oscar_03 0.1654∗∗∗ 1.4133∗∗∗ (0.0031) (0.0053) Oscar_05 0.0423∗∗∗ 1.3001∗∗∗ (0.0039) (0.0004) MO 1964 Miró Act RD 1282/1989 Act 17/1994 0.0423 (0.0402) Act 15/2001 -0.1600∗∗∗ -1.3231∗∗∗ (0.0161) (0.0344) RD 1652/2004 -0.1047∗∗∗ 0.0397∗∗∗ (0.0023) (0.0145) Act 55/2007 -0.1512∗∗∗ -1.2745∗∗∗ (0.0085) (0.0256) Video -0.1226∗∗∗ -0.1431∗∗∗ (0.0075) (0.0291) TV 0.0713∗∗∗ -1.1809∗∗∗ (0.0079) (0.0127) Digital TV -0.0123∗∗ (0.0060) Dummy Oscars -0.0234 -1.2394∗∗∗ (0.0189) (0.0022) ˆσ20.0127∗∗∗ 0.0140∗∗∗ (0.0038) (0.0049) R20.6635 0.6289 H18.1031 (0.0044) 11.0266 (0.0009) Standard errors in parentheses. ***, **, * Significant at 1%, 5% and 10% levels, respectively 40