Kevin Potter
University of the Arts London，Central Saint Martins，London，UK
Abstract
With hundreds of millions of box office revenue, dazzling stars, and countless fans, the huge profits of the film industry seem to be more likely to cause a public sensation, but multiple data studies express that more than half of movie is unprofitable. Income diversification has become the guarantee for the development of the film industry. Besides the box office, movie merchandise has become an important part of movie revenue. Predicting movie merchandise exploitability can help investors to reduce investment risks and make better decisions. This paper adopts the empirical research method, collects data from hundreds of movies, and produces a merchandise dataset. Through data analysis, we found three core influencing factors: genres, series, and box office, then build a prediction model to support film investment. This model can quantify the exploitability of movie merchandise at the film planning stage and has an accuracy of 88.6%, which has a good generalization ability.
Keywords: Film investment support, Movie merchandise, Prediction model
1. Introduction
The film industry is typically characterized by high investment costs, high investment risks, volatile profits, and unpredictability. Vogel’s estimated profit data (Vogel, 2020) shows that, despite the high profit that certain movies earn, a vast majority of films fail to break even (Davidson, 2012). 78% of movies lose money and only 22% are profitable (De Vany and Walls, 2004). What is more, the sharp rise in film production costs, the high remuneration of stars, and the continuous increase in advertising and marketing costs have made the investment cost of films higher. The box office of some films has even exceeded the cost several times, but they still lose money, such as Star Wars: Return of the Jedi, Forrest Gump, and Harry Potter and the Order of the Phoenix have also struggled to make money. When the market is saturated and competitive (Gong, Young and Van der Stede, 2011), it becomes more difficult to obtain a return on investment by simply relying on the box office. As Canalichio and Pete (2018) said, for those willing to change, the go “solo” approach is no longer sustainable and will lead to an inevitable decline. Therefore, the diversification of income sources is a problem that must be solved in the development of the film industry.
In the early 20th century, film companies have begun to pay attention to the important value of movie merchandise. Disney is a pioneer in the licensing industry and the value they create lies not in the tickets they sell at the box office but in the licensable products they create for future generations of consumers (Epstein, 2005). This “merchandise tie-in” strategy (Taxel, 1982; Elliott, 2014) was followed by the major studios. While only a small portion of studio films contain licensable characters, those are enough to exert enormous influence over retail sales. “Since these characters almost invariably outlive the movies that spawn them, the stream of licensing fees from them can enrich the studios’ clearinghouses for many decades” (Epstein, 2005). For example, Spider-Man, Batman, Superman, and other characters in movie merchandise can form a complementary relationship with their films. These products were still the most popular in the market for a long time and even inspired studios to make further films in each series.
It can be said that movie merchandise is not only an “icing on the cake” business for the film industry but also an important part of improving movie revenue and providing long-term growth momentum for films (Lubbers and Adams, 2004). Compared with other income sources of films, the “blue ocean market” of movie merchandise is the future trend of the development of the film industry and the key to film business transformation.
Figure 1. Movie revenue and expenditure.
In the traditional movie revenue estimation, the estimation of the box office can judge the overall income of a film. Nowadays, the revenue structure of a film includes box office, advertising, home entertainment, Internet and social media, movie merchandise, etc. (Fig. 1) The research on the prediction of the box office has been quite rich (Neelamegham and Chintagunta, 1999; Gunter, 2018). For advertising (Ewing, Du Plessis and Foster, 2001; Barnett and Cerf, 2017), home entertainment (Sebok, 2007), DVD leasing (Chopra and Veeraiyan, 2017; Chao, Hegarty and Fray, 2016), Internet and social media (Castillo et al, 2021; Feng et al,2020) have also made some achievements. However, there is a lack of analysis and research on movie merchandise, which makes it difficult to evaluate the overall revenue of a film. Therefore, the prediction of movie merchandise is a thorny problem that must be solved. Only by solving the prediction problem of movie merchandise can anyone judge whether a film will be profitable or not, and reduce investment risks.
Based on the transformation dilemma of the film industry and the development bottleneck of movie merchandise, this paper aims to build a scientific and effective evaluation system and advance the evaluation of movie merchandise in the planning stage of the film, to reduce the investment risk of the market, improve operational efficiency and increase investors’ confidence.
The objectives of this study are to address the following questions: first, find the influencing factors of movie merchandise; second, quantitative analysis of these factors to find new variables; third, find the weight relationship between these variables and movie merchandise.
2. Literature review
In the film Steamboat Willie (1928), Mickey Mouse achieved great success and helped Disney open the door of the market for movie merchandise. Since Star Wars, movie merchandise has become an important part of the Hollywood film business.They divided new consumers for the film and outlined the future direction of the entertainment industry, helping these characters break through geographical restrictions and become part of “global products” and “global culture”.
The way that movie merchandise increases movie revenue has been widely used by many film companies, but academic research lags far behind the practice. Most of the research focuses on combing the development history of movie merchandise, describing the market scale, analyzing classic cases, and emphasizing its importance. Few scholars use quantitative research methods to solve specific problems of exploiting movie merchandise. Combined with our research purpose, this part will mainly analyze previous literature including the definition, data research, influencing factors, and research methods.
2.1. The definition of movie merchandise
The concept closely related to movie merchandise is the movie franchise. Franchising means that the franchisor grants its trademark, product, patent, operation and management mode, and other intangible assets to the franchisee in the form of a franchise contract. In the field of film studies, franchising is not only related to series of films but also included movie merchandise, electronic games, home entertainment, music, books, theme parks, and other formats.
When we sort out the concept of movie merchandise, it is necessary to distinguish the scope from other franchise rights. Movie merchandise is “licensed goods with film characters and elements as the core”, including film-related peripheral toys, clothing, luggage, and other goods.
Movie merchandise has natural advantages in the continuation of films in time and space. It allows films to have “chemical reactions” with other industries to reinforce and complement other media. It also can help films get more “exposure” in various scenes, improving the scope of influence and communication ability. It has changed the entertainment form of films based on visual experience. These virtual images can be “materialized” into various commodities in life. Consumers can touch and feel films beyond the screen story, accompanying them for a long time. It has broken the shackles of time, eliminated cultural barriers and national barriers, and tapped new market segments, targeting consumers who may not be a part of them. It also extends the lifeline of a film. It makes a film no longer a short-term profitable project, but a long-term business that can be continuously mined and consumed.
2.2. Data research
Movie merchandise involves a wide range of fields, such as the film industry, the licensing industry and different retail businesses. Because of the large number of participants from all walks of life, it becomes more difficult to collect related data.
Movie merchandise is a part of the licensing industry. In all global licensors, film and entertainment companies have an absolute advantage in the licensing industry. According to the annual report of License Global (2022), four of the top ten global brand licensing companies were media and entertainment companies. In addition to the leader the Walt Disney Company, other companies such as Warner Media/Warner Bros., NBCUniversal/Universal Brand Development, and Paramount Global/Paramount Consumer Products and Experiences Consumer Products also occupied a considerable share in the licensing market. The Global Movie Merchandise Market Research Report (2020-2026) showed that “the global movie merchandise market size is projected to reach USD 34,970 million by 2026, from USD 29,290 million in 2020, at a CAGR of 3.0% during 2021-2026.”
These data show the overall development situation and market scale of movie merchandise. From its growth rate, we can see that this is a rapidly developing emerging industry with unlimited potential. In addition to the overall data, there is a lack of specific data to point out the sales of each movie’s merchandise. There are some reasons for this phenomenon:
• The exploitation of movie merchandise requires film copyright. Film copyright licensing is a complex business, which may belong to different companies. Each company has its authorization standards. Many companies regard this business as a “confidential business” and are unwilling to disclose their data, resulting in market opacity.
• Movie merchandise relates to a wide range of fields. It is difficult to make a comparative analysis from the perspective of the overall industry, resulting in statistical difficulties.
• The authorization time of movie merchandise can occur at any time and the cycle is very long and changes in real time, which leads to difficulties in data collection and analysis.
Based on the difficulties of data collection, this paper adopts the dimensionless method to collect data on the Internet platform and carries out a comparative analysis to find the horizontal relationship between them.
2.3. Research on influencing factors
The exploitability of movie merchandise is affected by both internal and external influencing factors. External influencing factors of the film are uncontrollable because they may change at any time. Most of these aspects are qualitative research, from which it is difficult to collect data and quantitative statistics. Internal influencing factors are the core and key to analyzing the exploitability of movie merchandise. Once established, it is difficult to adjust. In this case, this part will focus on analyzing internal influencing factors, which are mainly composed of the main cast and crew, genre, box office, reviews, main content, series, and so on. Before viewing, the consumption of movie merchandise connects with an appealing combination of the cast and crew, famous stars, genres, age rating, and release window. Series films are a guarantee for watching. Familiarity with the front films and curiosity about new content are important factors for audiences to enter the cinema again. Furthermore, serial films achieve longevity in evolving social-cultural contexts, through social salience and ongoing consumer engagement to achieve brand longevity (Preece, Kerrigan and Oreilly, 2019).
The content of the film will be mainly discussed by the media and audience. The value of the film is the core element that can attract audience consumption. Audiences not only participate in the imagination of the film world but also resonate with the values that the film adheres to. They will make an overall judgment based on the value system built by the film (Bassi, 2010), including the dual dimensions of the personal and social drive.
Audiences’ feedback on a film is directly reflected in the box office (Kerrigan, 2009) and reviews. Films with high box office, good reputations, and high reviews will naturally gain high popularity, which will also promote the consumption of movie merchandise.
2.4. Research methods
The research methods of movie merchandise can be divided into qualitative and quantitative methods. Qualitative research is a common research method in this field. It mainly discusses “why it is so important”. Many scholars use the case study to explore famous movie merchandise. Murray (2002) tried to explore the symbiotic relationship between the Harry Potter films and related commodities, and analyze the potential of movie merchandise for sustainable brand growth. Wasko and Shanadi (2006) tried to extract the law of film exploitability through the analysis of the Lord of the Rings films; Kapell and Lawrence (2006) pointed out that the success of Star Wars permitted sequels and prequels, as well as mountains of licensed merchandise.
Because these kinds of literature studied the already successful cases of merchandise, they have sufficient resource advantages and even formed a monopoly trend in the merchandise market. Therefore, their typicality is strong and their universality is weak, so it is difficult to generalize them to other films.
Since movie merchandise involves art, investment, management, commerce, legal affairs, taxation, cultural policy, consumer culture, and other disciplines, the research perspective is also significantly broad. Wasko et al. (1993) explained the process of commercialization of American films and believed that the marketing activities of movie merchandise have become a part of film production, and these marketing activities have opened up new market segments; Bettig (2018) used the critical academic analysis of political economy to put forward that the marginal cost of movie merchandise licensing is almost zero, but capitalists can benefit from the economies of scale of producing and distributing licensed merchandise. Turow (2017) emphasized that movie merchandise usually reflects the larger trend of the retail industry. To keep pace with the trend of choosing multi-channel retail, not only manufacturers but also retailers enter the film industry. These studies enrich the research methods and perspectives of movie merchandise even if movie merchandise is the “marginal” research field of these disciplines.
Although the market scale of movie merchandise is large and its value to the film industry is also high, it is still a new topic in the field of science or engineering. Some scholars have realized the importance of quantitative research for the film field and have adopted this research method in the prediction of the box office, establishment of film brands, consumer purchase intentions, and so on.
Since the success of movie merchandise is inseparable from the success of films, the prediction model of the box office plays an important role in the quantitative research of movie merchandise. Early box office prediction systems focused on the creative sphere, the scheduling and release pattern, and the marketing effort (Litman, 1983). Actors, characters, and stories, as well as kudos from reviewers and industry associations, are key factors (Basuroy, Chatterjee and Ravid, 2003).
Due to the rapid development of the Internet and the popularity of social media, audiences are more and more involved in discussions and comments on films. The data is more and more sufficient. A large number of studies began to use machine learning and data mining methods to predict the box office. Online platforms such as Wikipedia (Mestyán, Yasseri and Kertész, 2013), YouTube (O’Callaghan et al., 2015), Twitter, Facebook and Microblog (Du et al., 2014) are all helping to better optimize the prediction of the box office.
If a film can create brand value, it will also help the development of movie merchandise. Kohli, Yen, Alwi and Gupta (2021) show that the factors influencing the construction of film brands are popularity, sequels and emotional bonding. This is the reason why franchising/merchandising activities and timelessness are highlighted as key moderators.
2.5. Summary of literature review
The analysis of the concept, development history, data research, influencing factors, and research methods of movie merchandise can help us establish a theoretical framework to better understand the purpose of this study and the knowledge gaps that need to be filled.
Movie merchandise is film-related goods, rather than products of broadcasting or services. From its development history, it has gradually become an indispensable part of increasing the long-term income of films. It also has a good data performance in the licensing market over the years, but when it comes to the merchandise sales of each film, the data is extremely scarce. Data needs to be collected and the data set further improved.
In the past, qualitative research was the mainstream research method. This research can be used as a conceptual elaboration, but it is difficult to directly guide the actual operation of movie merchandise. In a sense, movie merchandise is not the product of film art, but the product of the film industry. They are largely linked to disciplines other than commerce, finance, taxation, legal affairs, statistics and other arts. Therefore, the integration of the quantitative research method can more scientifically analyze the exploitable value of movie merchandise. It is necessary to sort out, summarize and deduce the data of a large number of actual cases, obtain reliable parameters and basis, and build a complete movie merchandise prediction model. This method can improve the development efficiency and management process of merchandise effectively, reducing the uncontrollable factors and investment risks, to maximize the return on investment.
Movie merchandise involves a wide range of fields and has many influencing factors. It needs to integrate internal and external influencing factors. Due to the high uncertainty of external influencing factors and cannot be directly used in the exploitable evaluation of movie merchandise. Internal influencing factors are the core value of the film itself and the key factor. However, there are too many internal influencing factors. To predict the value of movies efficiently and accurately, it is necessary to conduct in-depth research and even quantitative analysis on each influencing factor, to find out the most important internal influencing factors and the weight relationship between them.
3. The exploitability evaluation experiment of movie merchandise
To predict the market in advance and evaluate the exploitable value of movie merchandise, it needs to build an effective value evaluation system.
Figure 2. Movie merchandise prediction model
When a film is in the planning stage, we can obtain the data of the film, using the model and its score to judge whether the movie merchandise is worth exploiting. This can help investors make investment predictions quickly and accurately, thereby reducing investment risks (Fig 2).
• This study first collected a large number of films and merchandise-related data from the Internet and used the dimensionless method to score the exploitability of movie merchandise, which was sorted into a data set of movie merchandise.
• Then we used multiple sets of models to perform quantitative analysis of film data and found the key internal influencing factors of the exploitability of movie merchandise.
• To predict more accurately, many models are used to carry out a large number of experiments and then calculate the weight relationship between these influencing factors to build a new model.
• Finally, we used these key internal influencing factors and the new model to predict the exploitability of movie merchandise.
3.1. Experimental design
Through a large number of research analyses, this experiment summarized a set of scientific and effective processes and methods.
Figure 3. Experimental flowchart of movie merchandise exploitability evaluation.
3.1.1. Experimental process
Figure 3 shows the specific experimental process of this study:
• The first step is data processing. A large quantity of movie data was collected from the online media platform. It mines analyzes and cleans up the data. And then it concludes with the labeling and scoring of the internal influencing factors and the obtaining of the data set.
• The second step is data analysis, which analyzes the characteristics of the data, finds the relationship between the data, and obtains the core internal influencing factors that affect the exploitability of movie merchandise.
• The third step is experimental design: extracting the main features of the data to find out the relationship between these features and merchandise sales, and getting a general model.
• The fourth step is to conduct experiments. To verify the reliability and validity, the model is repeatedly tested, and the parameters and weights are further adjusted to obtain a specific model.
• The fifth step is third-party inspection. The final model is submitted to the third-party inspection to obtain more objective suggestions and continuously improve the final model.
3.1.2. Formula
We first tried to build a Movie Merchandise Evaluate (MME) model. Assumption: movie X = {x1, x2, …, xn}. Movie feature data Θ = {Θ1, Θ2, …, Θn}. Movie X has characteristic data Θ. The correlation function is M (xi, {Θ1, Θ2, …, Θn}).Movie X has characteristic data Θ. The set s of is expressed as s = {(xi, Θ) | xi ∈ x, Θ = M (xi, {Θ1, Θ2, …, Θn})}.
The evaluation model has an evaluation function for evaluating and scoring films. The evaluation function is recorded as f(x). The form is f(x) = w1f1(x,Θ1) + w2f2(x,Θ2) + … + (1- w1 – w2 – …-wi-1) fi (x, Θi).
Among Θi is the character data type of movie x. Fi (x, Θi) is the evaluation scoring formula of feature Θi of movie x. Wi is one characteristic of film x. The weight of Θi is w1 + w2 + …+ wn = 1. The evaluation model has an evaluation function to evaluate merchandise exploitability corresponding to the score of the film. The evaluation function is recorded as g(x). The form is g(x) = f(x)–ε. When G(x) >0, f(x) > ε, film x is exploitability.
3.1.3. Experimental design description
The experiment is to find out variables w and n in the formula f(x), the functional form of f1、f2…fn and ε in g(x).
• What kind of movie feature Θ will influence the exploitability of movie merchandise?
• It needs to find out the influence degree of these features on the exploitability effects of movie merchandise through experiments, to confirm the value of these weights W.
• It needs to find out a reasonable value of ε.
3.2. Data processing
To better judge the exploitability potential of a film, this paper collects data from 300 European and American films from 1982 to 2021.
Figure 4. The number of films from 1982 to 2021
If the interval is 10 years, there are 12 films from 1982 to 1991, 56 films from 1992 to 2001, 96 films from 2002 to 2011, and 136 films from 2012 to 2021 (Fig 4). Among them, there are 133 series films and 167 non-series films.
After selecting these films, we begin to give them labels. The main labels include genres, time, MPAA (Motion Picture Association of America), rating, director, writer, stars, language, country of origin, filming locations, production companies, gross worldwide, series, and amount of licensed merchandise in franchise rights.
The data on films is from IMDb (Internet Movie Database), which is currently the largest movie database in the world, covering all kinds of film and television information.
The data of licensed merchandise from film franchises are on public online sites. However, this data is not complete. It needs to be comprehensively considered in combination with overseas movie merchandise sales platforms such as Amazon, e-bay, lost University, and Menkind. After that, all sales of movie merchandise will be scored. The main scoring categories include toys, stationery, daily necessities, clothing and accessories, and luggage. Each category ranges from 0 to 5 and the total score is 25.
4. Experimental analysis of the exploitability of movie merchandise
Based on 100 effective evaluation data and the central limit theorem, the sales score of movie merchandise finally obtains the following four groups of main relationships.
4.1. Genres and merchandise
To explore whether genres will influence the sales of merchandise, we labeled these sampled films to find the logical connection between them. A genre is a specific form of film, which is distinguished according to the characteristics of style, theme, and value, to facilitate the identification and organization of story materials. The main genres on IMDb include adventure, comedy, animation, family, action, fantasy, drama, Sci-Fi, romance, mystery, thriller, crime, documentary, thriller, musical, history, war, western, and sport.
Each movie will be labeled with 3 to 5 genre labels. For example, the movie genre labels of Spider-Man are action, adventure, and Sci-Fi. Through the collection of type data, this paper summarizes the films into different labels, each label of a film will be included, and the film genre data under different labels will be repeatedly calculated. Accordingly, the scores of movie merchandise corresponding to films under each label will also be accumulated. The sum is the movie merchandise score of this film genre. The higher the score means the better the sales of movie merchandise and vice versa. The following features were found:
Figure 5. Proportion relationship between sales scores and genres of movie merchandise
In Figure 5, film characteristics of adventure, action, comedy, and Sci-Fi have higher scores, accounting for 24%, 19%, 12%, and 9% respectively. The sales of these types of merchandise are better than the others.
Figure 6. Movie merchandise sales score and the exploit difficulty of genres
In Figure 6, the film genres on the left side are easier to exploit with merchandise than those on the right side, which shows a hill-like downward trend from left to right. This indicates that the exploitability of merchandise is concentrated in the genre area on the left.
In summary, film genres are strongly related to the merchandise. Films with adventure, action, comedy, and Sci-Fi characteristics are easier to exploit with the merchandise. While films with history, sports, fantasy, war, documentary, and other characteristics are not the mainstream of merchandise development.
4.2. Series films and merchandise
To find out the relationship between series and non-series films with merchandise sales, two cases are distinguished (in Fig. 7): series films are marked with 1 (in orange) and non-series films with 0 (in green). The following information is obtained in combination with the sales score of movie merchandise:
Figure 7. Relationship between movie merchandise sales score and series/non-series films
As can be seen from Figure 7, the yellow area represents series films, and the green part represents non-series films. Films with high scores of movie merchandise are concentrated in the left area, in which series films account for a large proportion. Among the top 20 films with the most successful merchandise, series films accounted for more than 90%. Only a few series of films did not drive the sales of merchandise, such as Home Alone and Mission: Impossible. From here it can be seen that series films are directly related to the exploitability of movie merchandise, and the success probability of series films in exploiting movie merchandise is significantly higher than that of non-series films.
4.3. Box office and merchandise
The box office reflects the popularity of the market and the consumption enthusiasm of the audience. To further understand the impact of the box office on the sales of movie merchandise, this paper makes statistics on the worldwide gross of each film from IMDb, taking $100 million as the unit of measurement. In Fig. 8, the left side of the figure below represents the score of movie merchandise, and the blue part represents the level of the score. The right side is the numerical value of the box office, and the orange part indicates the level of the box office.
Figure 8. Relationship between movie merchandise sales score and box office
As can be seen from Figure 8, there is a linear relationship between the score of movie merchandise and the box office. The left part of the figure shows a double high trend, indicating that films with high merchandise sales scores also have a high box office. In the right part of the figure, except for the film Avatar, there are films with a double low trend and poor merchandise sales scores.
Figure 9. The relationship between movie merchandise sales score and movie ratings
In Figure 9, the orange line represents the box office. Movies are arranged in order of the box office from left to right, with a high box office on the left and a low box office on the right. The blue peaks represent scores of merchandise sales; the higher the height, the better the sales. It can be seen from the picture that the peaks on the left side are significantly higher than those on the right, and most of these films have grossed more than $450 million worldwide (not adjusted for inflation).
Therefore, a $450 million box office is the basic condition for exploiting merchandise, but it does not mean that the higher the box office is, the more potential for exploitation. For example, Avatar and Titanic grossed $2.84 billion and $2.2 billion at the box office but did little merchandise.
In summary, it can be seen from Figure 8 and Figure 9, the box office has a linear relationship with the movie merchandise. Except for a few films, most merchandise with high ratings also do relatively well at the box office. $450 million is the critical value for exploiting movie merchandise. If the box office is lower than this value, it indicates that this film does not have enough audience recognition and has little development value but if the box office is higher than this value, it still needs to be considered with other factors comprehensively.
4.4. Ratings and merchandise
To explore the impact of film ratings on the exploitability of movie merchandise, a comparison is made between the merchandise sales score and rating. The rating data is based on a weighted average of voter turnout from IMDb users on a scale of 10.
Figure 10. The relationship between movie merchandise sales score and movie ratings
In Figure 10, the value on the left represents the movie’s score out of 10. The green part shows the movie rating, from high to low and from left to right. The value on the right represents the sales score of movie merchandise. Figure 10 shows that the merchandise sales score is irregular (the blue curve). This results in the opposite phenomenon to that found in the past analysis: it is not the high movie rating that leads to better sales of movie merchandise. There is no direct correlation between the two.
A movie rating is an evaluation of the overall perception of a film. The score is affected by the number of film audiences. Sometimes a film with a low score but a large number of audiences is still more popular than another film with a high score but a small number of audiences. Ratings are influenced by audience emotions and have little correlation with the ability to exploit movie merchandise. In some movie series, there may be a low score in one or several films. For example, in the tetralogy of Superman films, Superman (1978) got 7.4 points, but Superman JV· The Quest for Peace (1987) only got 3.7 points. However, as the series is a whole, the development potential of its merchandise is less affected by a single film.
4.5. Experimental Results
Four conclusions can be drawn from the above analysis:
First, the genre of a movie has a direct impact on the exploitability of movie merchandise. Among all genre labels, movies with the characteristics of adventure, action, comedy, and Sci-Fi are more suitable for exploiting merchandise, while other genres are not as popular as these genres in marketing. Since each movie has multiple labels, the dominant genre of the movie needs to be prioritized and other genre labels serve as auxiliary references.
Second, the series is one of the important references for exploiting movie merchandise (Filson & Havlicek, 2018). The more series indicates the higher market recognition, which shows the greater the development value of the merchandise. Films such as Star Wars, Harry Potter and Spider-Man have nurtured their markets over decades, creating long-term emotional connections with audiences that guarantee the profits of movie merchandise.
Third, the box office is a reference factor in merchandising, but not decisive. The box office of blockbuster merchandise is usually more than $450 million, but films that exceed this number are not rare, while few have been able to take off in the merchandise market. The main reason is that the limited release time of the film makes it difficult to ensure the continuous attention of the audience. However, the box office can reflect the short-term market heat to a certain extent and has a certain reference function for exploiting movie merchandise.
Fourth, movie rating has no direct impact on the exploitability of movie merchandise. In addition to the phenomenon of counterfeit ratings, the rating level is influenced by audience emotions, viewing habits, personal preferences, and other factors. Therefore, the subjectivity of movie rating is strong and the controllability is weak. When the audience rating reaches a certain amount, it can be used as the evaluation standard of the film market response and the auxiliary reference of the exploiting value of movie merchandise.
5. The GSBO model: the exploitability evaluation model of movie merchandise
Through experiments and analysis, this paper obtained the GSBO evaluation model of the MME model. There’s movie X. X= {x1,x2,…,xn}. Film feature data Θ. Θ = {Θ1, Θ2, Θ3}. Θ1, Θ2, and Θ3 represent the film genre, series, and box office respectively. The correlation function of movie X with characteristic data Θ is: M(xi,{ Θ1, Θ2, Θ3}). The set S of movie Xwith characteristic data Θ is expressed as: S = {(xi, Θ) | xi ∈ x, Θ = M (xi, {Θ1, Θ2, Θ3})}.
The evaluation model has an evaluation function for marking movie merchandise. The evaluation function is denoted as f(x)and expressed as: f(x) = w1 G(x, Θ1) + w2 S(x,Θ2) + w3 BO(x,Θ3). Among the rest G (x, Θi) is the evaluation scoring formula of the genre (G) of film x, S (x, Θi) is the evaluation scoring formula of series (S) of film x, BO (x, Θi) is the evaluation scoring formula of Box Office (BO) of film x.
Based on the above evaluation experiments, the GSBO evaluation model is obtained to judge the exploitability of movie merchandise. In this model, G, S, and BO represent the genre, series, and box office respectively.
First, adventure, action, comedy, and Sci-Fi are the most easily exploited merchandise of genre (G) factors. Since each film has three genres of characteristics in the genre data, so the adventure, action, comedy, and Sci-Fi categories receive 1 point, while the other categories receive 0 point.
Second, in the series (S) factor, sequels are made based on the success of the previous film. If the first movie is a flop, sequels are almost impossible. Therefore, in the series, it takes a “hit” movie to satisfy the conditions for movie merchandise. The expression for this is: S (x, Θ2) = S (if Θ2 = Series then output = 1 else output = 0).
Finally, the box office (BO) factor can only be used as a reference value compared to the genre and series factor. Since the development of movie merchandise may have started at the film planning stage, the box office of the film can only be considered by referring to the box office prediction system of the film companies or similar films. BO (x, Θ1) = BO (if Θ2 > = 4.5 then output = 1 else output = 0).
The prediction model of movie merchandise has the function of predicting the exploitability value of movie merchandise in the planning stage of the film (Table 1). The evaluation function is denoted as G(x) and expressed as G(x) = f (x) – 3.
| THE EXPLOITABILITY OF MERCHANDISE | G(x) |
| UNDEVELOPABLE | 0 |
| DEVELOPABLE | 1 |
| RECOMMENDED | 2 |
| HIGHLY RECOMMENDED | 3 |
Table 1 Movie merchandise exploitability and box office scores
Its formal expression is as follows:
| Algorithm 1: GSBO Model |
| Input: G, G: Genre of film, it is a list and it contains four units, the list set (GL) is as follows: [(Adventure, Action, Comedy), (Adventure, Comedy, Sci-fi), (Adventure, Action, Sci-fi), (Action, Comedy, Sci-fi)]; Input: S: Is the film Movie Series? Value: [yes, no]. Input: BO: the estimated Box Office of the film (Unit: USD 100 million). |
| If BO > 4.5 then If (G |
Figure 11. The GSBO quantitative analysis model flow
The GSBO model is the first model to quantitatively analyze the exploitability of movie merchandise in the film industry. This model predicts the exploitability of movie merchandise by extracting three core feature data (Fig. 11). In Figure 11, the GSBO model uses a very simple process and can quickly complete the development calculation. According to the GSBO model and process, 300 films will be sampled for testing.
5.1. Experiment I G: S: BO = 3: 3: 3
In Experiment I, since the weight of the influence factors could not be accurately judged, we set the weight of genre, series, and box office as 3:3:3, and each weighted score was 3 points.
In the genre factor, if a film met adventure, comedy, action and Sci-Fi simultaneously, the score was 3; if not, the score was 0. Because we only selected three genres in a film, there was no condition that all four characters were met.
In the series factor, the series film got 3 points and the non-series got 0 points.
In the box office factor, over $450 million box office got 3 points and under it got 0 point.
In this experiment, the cumulative scores of genres, series and box office were compared with the sales score of merchandise. If the cumulative score of the three items was 6 or above it was qualified, while those below 6 were considered abnormal. In this test, 252 films were correct and 48 films were wrong, with an accuracy rate of 84%.
5.2 Experiment II G: S: BO = 3: 3: 3
Based on the observation and analysis of the first experiment, it was found that there was a big gap between different films. For example, in series factor, two and eight series got the same score, which caused a margin of error, as well as in terms of genre and box office. Therefore, in Experiment II, the weight of influence factors was further classified and refined.
The weight of genre, series and box office was still set at 3: 3: 3, with 3 points for each weight.
In the genre factor, meeting the requirements of adventure, comedy, action and Sci-Fi at the same time got 3 points; meeting the requirements of 2 of them got 2 points; meeting the requirements of 1 of them got 1 point; failing to meet the requirements got 0 points.
In the series factor, the number of 6 or more series was 3 points; the number of 5 or more series was 2 points, only 1 series was 1 point, and the number of non-series was 0 points.
In the box office factor, 3 points were scored for those over $900 million, 2 points were scored for those over $450 million, 1 point was scored for those over $450 million and 0 points were scored for those over $200 million.
In this experiment, the cumulative scores of genres, series and box office were compared with the sales score of merchandise. If the cumulative scores of the three items were 5 or above it was qualified, while those below 5 were considered abnormal. In this test, 263 films were correct and 37 films were wrong, with an accuracy rate of 87.6%.
5.3 Experiment III G: S: BO= 3: 2: 2
After Experimental II, it was found that the weight of the series and box office was too much, leading to inaccurate predictions of some films. Therefore, further experimental improvement was carried out. The weight of series and box office was lowered, while the weight of genre remained unchanged, and the weight of genre, series and box office was set at 3: 2: 2.
In the adjusted series factor, 6 series or above got 2 points, greater than or equal to 5 and less than or equal to 2 scores 1, and the non-series film got 0 point.
In the box office factor, films with a gross of $450 million or above got 2 points, less than or equal to $450 score but greater than or equal to $200 million got 1 point, and less than $200 million got 0 point.
In Experiment III, the cumulative scores of genre, series, and box office were compared with the sales score of merchandise. If the cumulative score of the three items was 5 or above it was qualified, while those below 5 were considered abnormal. In this test, 266 films were correct and 34 films were wrong, with an accuracy rate of 88.6%.
6. Test analysis
From the comparison of the three groups of experiments in Table 2, it can be seen that Experiment III achieved better results than Experiment I and Experiment II. Experiment III shows that a more accurate prediction effect can be obtained by calculating G, S, and BO attributes according to a certain weight rather than equal proportion. As can be seen in Figure 12, after three experiments, the accuracy of the evaluation is gradually increasing.
| Experiments | Correct number（Total number: 300） | Accuracy |
| Experiment 1 | 252 | 84% |
| Experiment 2 | 262 | 87.6% |
| Experiment 3 | 266 | 88.6% |
Table 2 Accuracy comparison of three experiments
Figure 12. Accuracy of three experiments
These three attributes play a key role in the exploitability of movie merchandise, but their influences on movie merchandise are slightly different from each other. Therefore, it is necessary to adjust the weight to get a more accurate result.
The parameters of Experiment III were used as parameters of the GSBO model in this paper. The final formal expression of the model is as Algorithm 2:
| Algorithm 2: GSBO Model |
| Input: G, G: Genre of film, it is a list and it contains four units, {X1= Adventure; X2 = Comedy; X3 = Action; X4 = Sci-fi}{P(Xi): P(X1; X2; X3; X4) | i = 3: G = 3; i = 2: G = 2; i = 1: G = 1; i = 0: G = 0}Input: S, S: Is the film Movie Series?Switch (Series quantity) {case (Series quantity ≥ 6): S = 2; case (5 ≥ Series quantity ≥ 2): S = 1; case (2 > Series quantity): S = 0; case (Series quantity = 0): S = 0}Input: BO, BO: the estimate Box Office of the film (Unit: USD 100 million). |
| Switch (Box Office) {case (Box Office ≥ 4.5): BO = 2; case (4.5 > Box Office ≥ 2): BO = 1; case (Box Office > 2): BO = 0}.Sum = G + S + BO. IF the Sum ≥ 5,then movie merchandise will success. |
7. Model validation
In order to verify the effectiveness of the GSBO model, a third party is entrusted to use the model to forecast merchandise. The results are shown in Figure 7. These two companies collected a total of 110 films and correctly predicted 94, with an overall accuracy rate of 85.4%. (Table 3)
| Company | The number of test films | Correct quantity | Accuracy |
| A | 60 | 52 | 86.6% |
| B | 50 | 42 | 84% |
| Grand total | 110 | 94 | 85.4% |
Table 3: relationship between the number of test films and test accuracy
Figure 13. Comparison of test quantity and accuracy of company A and B
The third-party verification agency shall independently complete the selection of films used in the effectiveness evaluation of the model and the evaluation of films Figure 13. Calculate the variance of three groups of data (the third experiment data, data of the company A and the company B).
Eq. (A.1)
Eq. (A.2)
S2 = 2.306513 Eq. (A.3)
Calculate the standard deviation of three groups of data:
Eq. (A.4)
S = 1.88326. Eq. (A.5)
Although the three groups of data were provided by different companies, they still achieved a good result of variance of 2.31 and standard deviation of 1.88. This also shows that GSBO method has strong generalization ability, and can provide calibrated data prediction and evaluation ability under different data conditions.
8. Application of the GSBO model
Through experimental analysis, this study found the key influencing factors of movie merchandise and also found the weight relationship between these influencing factors, so as to create the GSBO model to predict the exploitability of movie merchandise.
In the planning stage of the film, the GSBO model can help predict the exploitability of movie merchandise. In the GSBO model, only three factors of film genre, series and box office estimation are needed to obtain the score of movie merchandise exploitability.
Figure 14. The application of the GSBO model
If the score is greater than or equal to 5 points, it has the exploitable value of movie merchandise; if it is less than 5 points, it is unexploitable Figure14. For example, the upcoming movie The Wandering Earth II, which is in the series of The Wandering Earth. The genre of this film is Sci-Fi, adventure and disaster. If the estimated box office is $450million, it will get a score of 4 in the GSBO model, which is not suitable for exploiting movie merchandise.
9. Conclusion
The movie business is a business of extremes. The winners live in palaces and the losers live in slums. High risk and unstable profits are the main characteristics of this industry. How to make money in this extreme industry is a tricky question. In the past, the main revenue of films depended on the box office. This is why when analyzing the movie revenue, investors only need to calculate the profit and loss of the box office to predict the investment risk.
However, with the substantial increase in film production, distribution and marketing costs, it is difficult to recover costs by relying on the box office alone. Many blockbusters have even made very high box office, but they still suffer losses. This urges the film industry to solve the problem of diversified income. In this case, family entertainment, advertising, the Internet and movie merchandise have become more and more important. After in-depth research on the film industry, we found that we need to build an effective method for evaluating and predicting the overall revenue of a movie, including multiple revenue methods, not just calculating the revenue of the box office.
In order to be able to make accurate forecasts of these revenues, we require a general statistical model to capture the important properties of revenue. The quantitative analysis method proposed in this paper for the first time effectively solves the problem of accurate estimation of emerging movie revenue methods such as movie merchandise. This way can help film producers form a clear understanding of the future revenue in the early planning stage, to make dependency planning and reduce the investment risk.
To realize the quantitative analysis of merchandise prediction, this paper first collected a large number of online public data and produced a set of movie merchandise quantitative analysis data. Secondly, it quantitatively annotated the movie merchandise data set and scored the merchandise of each movie. Third, through the quantitative analysis of the data set, we found the key factors that affect the sales of the movie merchandise in the film data collected through the Internet. Fourth, after many experiments, we found a quantitative method for the three key factors of the film genre, series and box office, and calculated the weights of these factors. Fifth, we used the analyzed film factors and the weights between these factors to build a movie merchandise sales prediction model: the GSBO model. Finally, we proved the effectiveness and generalization of the movie merchandise sales prediction model through third-party verification.
Using this evaluation method can help investors make a more scientific and effective investment prediction of the overall income of the film before its release, and make a reasonable investment decision. It can increase confidence and reduce the loss of film investment.
This is very important for the development of the film industry. Assuming that there are 63% (X) movies unprofitable and 37% movies profitable (Vogel, 2020; De Vany and Walls, 2004; Stephen, 2016), among these movies with investment unprofitable, if 80% of these films (N) will use the GSBO model (the accuracy is 88.6%) to improve the income of movie merchandise, 44.6% of movies could be saved from losses to profits. The calculation formula is as follows: Z(44.6%) = X(63%) × N(80%) × GSBO(88.6%).
Further, assuming that the number of movies released worldwide in a year (A) is 3500 (The Numbers, 2022), the average investment cost (C) is $100 million (Mueller, 2021), and the average loss rate is 20% (L) among films (X) that may lose 63%. If 80% of movies (N) use the GSBO model (the accuracy is 88.6%) to do a good job in movie merchandise, the loss of $32,158 million in the film industry will be saved. The calculation formula is as follows: Z(31258.08) = A(3500) × C(100) × X(63%) × L(20%) × N(80%) × GSBO(88.6%).
In this paper, we only study the exploitability of movie merchandise. In the follow-up study, we will further calculate the proportion of movie merchandise in movie revenue in detail. At the same time, we will make a more accurate analysis of different genres of films, such as animation, comedy, Sci-Fi and other film genres, to build a complete prediction system.
References
Baek, H., Oh, S., Yang, H. D., & Ahn, J. (2017). Electronic word-of-mouth, box office revenue and social media. Electronic Commerce Research and Applications, 22, 13-23.
Barnett, S. B., & Cerf, M. (2017). A ticket for your thoughts: Method for predicting content recall and sales using neural similarity of moviegoers. Journal of Consumer Research, 44(1), 160-181.
Bassi, F. (2010). Experiential goods and customer satisfaction: An application to films. Quality Technology & Quantitative Management, 7(1), 51-67.
Basuroy, S., Chatterjee, S., & Ravid, S. A. (2003). How critical are critical reviews? The box office effects of film critics, star power, and budgets. Journal of marketing, 67(4), 103-117.
Bettig, R. V. (2018). Copyrighting culture: The political economy of intellectual property. Routledge.
Canalichio, P. (2018). License to Operate—The Future of the Licensed Brand. In Expand, Grow, Thrive. Emerald Publishing Limited.
Castillo, A., Benitez, J., Llorens, J., & Luo, X. R. (2021). Social media-driven customer engagement and movie performance: Theory and empirical evidence. Decision Support Systems.
Chao, C. N., Hegarty, N., & Fray, I. (2016). Impact of movie streaming over traditional DVD movie rental—An empirical study. Journal of Industrial and Intelligent Information Vol, 4(2).
Chopra, S., & Veeraiyan, M. (2017). Movie rental business: blockbuster, netflix, and redbox. Kellogg School of Management Cases.
Davidson, A. (2012). How Does the Film Industry Actually Make Money?. New York Times, 26.
De Vany, A. S., & Walls, W. D. (2004). Motion picture profit, the stable Paretian hypothesis, and the curse of the superstar. Journal of Economic Dynamics and Control, 28(6), 1035-1057.
Ewing, M. T., Du Plessis, E., & Foster, C. (2001). Cinema advertising re-considered. Journal of Advertising Research, 41(1), 78-85.
Feng, N., Feng, H., Li, D., & Li, M. (2020). Online media coverage, consumer engagement and movie sales: A PVAR approach. Decision Support Systems
Filson, D., & Havlicek, J. H. (2018). The performance of global film franchises: installment effects and extension decisions. Journal of Cultural Economics, 42(3), 447-467.
Gong, J. J., Young, S. M., & Van der Stede, W. A. (2011). Real options in the motion picture industry: Evidence from film marketing and sequels. Contemporary accounting research, 28(5), 1438-1466.
Gunter, B. (2018). Predicting movie success at the box office. Springer.
Kapell, M., & Lawrence, J. S. (Eds.). (2006). Finding the force of the Star wars franchise: fans, merchandise, & critics (Vol. 14). Peter Lang.
Kerrigan, F. (2009). Film marketing. Routledge.
Kohli, G. S., Yen, D., Alwi, S., & Gupta, S. (2021). Film or film brand? UK consumers’ engagement with films as brands. British Journal of Management, 32(2), 369-398.
Litman, B. R. (1983). Predicting success of theatrical movies: An empirical study. Journal of popular culture, 16(4), 159.
Ma, H., Kim, J. M., & Lee, E. (2019). Analyzing dynamic review manipulation and its impact on movie box office revenue. Electronic Commerce Research and Applications, 35.
McMurtry, L., & Epstein, E. J. (2005). The big picture: The new logic of money and power in Hollywood.
Mueller (2021), Why movies cost so much to make, https: //www. investopedia. com.
Murray, S. (2002). Harry Potter, Inc.: content recycling for corporate synergy. M/C Journal, 5(4).
Neelamegham, R., & Chintagunta, P. (1999). A Bayesian model to forecast new product performance in domestic and international markets. Marketing Science, 18(2), 115-136.
O’Callaghan, D., Greene, D., Conway, M., Carthy, J., & Cunningham, P. (2015). Down the (white) rabbit hole: The extreme right and online recommender systems. Social Science Computer Review, 33(4), 459-478.
Sebok, B. R. (2007). Convergent Hollywood, DVD, and the transformation of the home entertainment industries. The University of Texas at Austin.
Stephen (2016), Do hollywood movies make a profit?, https :// stephenfollows . com/hollywood-movies-make-a-prof it/.
The Numbers (2022), Movie index, https ://www. the-numbers. com/.
Vogel, H. L. (2020). Entertainment industry economics: A guide for financial analysis. Cambridge University Press.