Ask almost anyone and they’ll agree that minorities are underrepresented in Hollywood movies. The data collected for this study confirms this intuition. From a survey of over 1,000 of the top grossing films in the US from 2004-2013, only 15% of films featured nonwhite actors in leading roles.
Not only are there limited roles for nonwhite actors, but in some cases studios will engage in “racebending” where they change the race or ethnicity of a character. The long-anticipated Dragonball and The Last Airbender films, both heavily criticized for racebending, were both critical and commercial flops. (For some comic relief and commentary, see video below)
Even when movies do star nonwhite leading actors, they may not be recognized as such by third party websites such as IMDb.com. I discovered these “micro-biases” while collecting data firsthand. Notice in the following pictures that based on the movie posters, black women appear to be in the leading roles. In The Secret Life of Bees, Queen Latifah’s name is listed first and her face is featured most prominently on the poster. However, Dakota Fanning is listed first in IMDb’s ‘Stars’ section. In The Family that Preys, Alfre Woodard stands in the foreground of the poster but is again listed second. I could not find a methodological explanation for this pattern on IMDb’s website.
The social and economic impact of the movie industry is huge. In the US and Canada, 68% of the population (228 million people) went to the movies at least once in 2013. With an $11 billion industry at stake, how do movie studios take race into account when making casting decisions? Studies have found that favorite films and film genres are a function of race, age, and gender (Fischoff). With this in mind, I investigate the following questions:
- What is the effect of casting a nonwhite actor in the lead role on box office performance?
- Which factors are correlated with the casting decision?
Two reports have attempted to answer the first question. UCLA’s Bunche Center released a Hollywood Diversity Report in 2014 that compared median global box office revenue by minority cast share. Median revenue peaked for 21-30% minority cast share films (close to the US population), but the report studied only 172 films in 2011.
Columbia University’s study on The Latino Media Gap focused solely on Latino representation but also investigated the impact of diversity on financial performance. The study used mainly case study evidence, citing successes such as The Fast and the Furious franchise and ABC’s Grey’s Anatomy to support the thesis that diversity pays.
This investigation covers 1,070 films over ten years (2004-2013) representing the top grossing movies released wide in the US. Animated films were excluded from this study. 909 films featured white leading actors; 161 featured nonwhite leading actors. The data collected includes:
- Box office performance from boxofficemojo.com
- Budget data from the-numbers.com
- Genre, cast, and rating data from imdb.com
- Ticket price inflation from the National Association of Theatre Owners
The much larger sample size, 1,070 compared to at most 600, provides an opportunity to conduct a quantitative analysis of the impact of race on box office performance. However, the data is limited to the race of the lead actor and not the entire cast.
A naive regression of gross revenue on the race of the leading role produces predictable but misleading results:
A nonwhite leading role equates to a $17,400,000 decrease in gross revenue, significant at the 99.9% level. However, there are several problems with this model:
- The number of theaters at which the movie was released is directly related to the gross revenue of the movie.
- There is almost certainly omitted variable bias. Budget, genre, and rating are all likely to be related to race and affect gross revenue.
- The gross revenue measure does not correct for ticket price inflation over the years.
Films featuring white and nonwhite leading actors are in fact different for a number of metrics. The following table provides median values and the percentage difference (medians were used to mitigate the effects of outliers):
Because of these differences, all of which were significant at at least the 95% level except Profit Margin, a number of controls were introduced to a stronger, revised model:
- tickets is the estimated tickets sales of the movie calculated by dividing the gross revenue by the average ticket price for that year. This corrects for ticket price inflation over time.
- X represents a vector of control variables (see next section).
- nonwhitelead is a dummy variable for the race of the leading role (1 = nonwhite leading role).
The control variables include:
- theaters: the number of theaters at which the movie was released
- deltatheaters: the difference in the number of opening weekend and final theaters (this is a good indicator of how well theater owners predicted the movie would perform).
- daysintheater: the number of days during which the movie was shown
- budget: production budget (whether advertising is included in this number is unclear).
- year: year of release
- month: month of release
- studio: production studio
- series: dummy variable for sequels and series (1 = sequel or series; 0 = first in series or otherwise)
- metascore: Metacritic’s Metascore
- maingenre: first listed genre
- secondarygenre: second listed genre
Key Findings
After controlling for these variables, we find that films featuring a nonwhite leading actor experience a 9.5% boost in ticket sales significant at the 95% level (p-value = 0.018) over comparable films. All other control variables are significant at the 99.9% level. To put this 9.5% boost in perspective:
- Each additional day that a movie is shown in theaters produces a 1.1% increase
- Being part of a series increases ticket sales by 22.6%
- A budget increase of $10,000,000 increases ticket sales by 2.3%
- Movies opening in December have 23.2% increased ticket sales
- Adventure/horror films underperform comparable films (by budget, rating, release scale, etc.) of other genres by 89%
A good question was raised during a discussion of these results: What is a ‘comparable’ film?
Ideally, a comparable film is a movie that is identical in all of the control variables apart from the race of the leading role. These comparisons can be made in the regression model but may not exist as real films. Film quality, box office competition, and actor popularity are only some of the factors that this study cannot fully control. However, to put this comparison into context, I’ll provide two examples of comparable films.
The Great Debaters / Freedom Writers
The films, both released in the same year, feature similar plot lines. The Great Debaters (2007), featuring Denzel Washington, is: “A drama based on the true story of Melvin B. Tolson, a professor at Wiley College Texas. In 1935, he inspired students to form the school’s first debate team, which went on to challenge Harvard in the national championship” (IMDb.com). For Freedom Writers (2007), featuring Hilary Swank: “A young teacher inspires her class of at-risk students to learn tolerance, apply themselves, and pursue education beyond high school” (IMDb.com). The gender difference provides a potential complication, but other than that the films are very similar. A comparison of their key attributes are as follows:
.
Ray / Walk the Line
Released one year apart, these films both chronicle the lives of famous musicians: Ray Charles and Johnny Cash. Their comparisons are below:
Consistent with both sets of examples is the fewer number of theaters for films featuring black leading actors. Because of this, absolute gross revenue remains capped at a certain point regardless of the strength of the movies’ performance. This is a good distinction to make. Absolute metrics are still weak for movies featuring nonwhite actors. Hancock featuring Will Smith, the most widely released movie featuring a nonwhite actor at 3,965 theaters, is still 11% behind the most widely released movie featuring a white actor (Twilight: Eclipse; 4,468 theaters). I Am Legend, also starring Will Smith, is the top performing nonwhite lead film having sold an estimated 37,266,426 tickets. Compare this to The Avengers which sold 78,311,295 tickets, more than double I Am Legend’s sales. This performance difference captures the effect of the series factor on ticket sales but also highlights the absence of nonwhite leading actors in major blockbuster films.
While collecting the data, I also recorded an indicator variable for black-oriented films. Many people associate these films with the works of Tyler Perry. For my designation, I included films that, beyond featuring a black leading actor, were also marketed toward a black audience. Overall, 37% of movies featuring nonwhite leading actors were also black films.
Including this variable in the same regression model as before significantly changed my results. The coefficient for nonwhite leading actor decreased to 2.6% and was no longer significant (p-value = 0.553). However, the black film factor provided a 21.8% boost in ticket sales significant at the 99% level (p-value = 0.004).
Keep in mind that this boost is against comparable films, a concept that becomes increasingly difficult to illustrate for black films. Regardless, the performance of black films remains weak on absolute terms. The top performing black film by ticket sales in the study, Dreamgirls — and this designation is highly debatable — is ranked at 171st out of 1,070 (16th percentile). The Butler, also debatable as a designated black film, is ranked at 191. Norbit is third at 204.
Part of the issue comes from white audiences’ perception of black films. Andrew J. Weaver of Indiana University writes in “The Role of Actors’ Race in White Audiences’ Selective Exposure to Movies”:
Movie producers are often reluctant to cast more than a few minority actors in otherwise race-neutral movies for fear that the White audience will largely avoid such films (Weaver).
According to Will Smith who starred in the 2005 hit romantic comedy Hitch:
… there’s sort of an accepted myth that if you have two Black actors, a male and a female, in the lead of a romantic comedy, that people around the world don’t want to see it. We spend $50 something million making this movie and the studio would think that was tough on their investment (Weaver).
The Washington Post asked the question quite starkly: “Do white people watch black movies?” The general consensus appears to be a quiet ‘no.’ The relative strength of black film performance is then most likely due to their appeal to black moviegoers. Black films often feature family-centric plot lines and strong female (even if they are cross-dressing men) characters, a respite from the crime-ridden, gun-slinging roles for black actors in many other Hollywood films. If black films have risen from a desire to counteract these mainstream images, then there may also be markets for Latino and Asian audiences as well. In television, this movement has already begun. Univision and Korean dramas are dominating at-home screen time for Latino and Asian audiences, respectively. The opportunity for the Latino movie market is immense considering that Latinos are overrepresented among active moviegoers:
Overall, minorities are underrepresented in leading roles in Hollywood films by a factor of more than 2. Films that feature nonwhite actors, while they may outperform comparable white films, are limited in their absolute performance. Their budgets, number of theaters, and ticket sales fall behind Hollywood’s biggest successes. Black films are especially notable for their relative ticket outperformance in excess of 20%, but again their absolute performance remains weak. Their relative strength does suggest a market opportunity for underserved minority markets.
Download the full paper here.
K Does Cases 