Mini Post #5: The First BlocKKKbuster
Did movies praising the Ku Klux Klan make the Klan more popular?
I’ve decided to try something new: once a week, to write a shorter, narrower post focusing on one specific paper. Last week, I wrote about how people’s attitudes towards new housing affect housing policy. This is the tab with all the previous (and current) posts.
Onto the actual post: does seeing the infamously racist movie Birth of a Nation lead to higher rates of racial violence?
Racism, I think we can all agree, is bad. Movies, I also think we can all agree, are good. Movies about racism have been somewhat of a mixed bag: for every Killers of the Flower Moon, we get a Green Book. But do movies about racism matter? This weeks paper, titled “The Birth of a Nation: Media and Racial Hate” (2023) by Desmond Ang, tries to argue somewhat in the affirmative: racist movies make you racist.
Firstly, let’s explain the Ku Klux Klan (or “the Klan” or “KKK” for short). The Klan was originally founded after the Civil War, where a group of Confederate war veterans got together to terrorize the recently freed slaves by means of (largely public) violence. The group contained a hodge-podge of conservative and reactionary elements, but largely acted as a white supremacist paramilitary group - until it was disbanded by decisive federal legal actions throughout the early 1870s, and being shut down by Grand Wizard Nathaniel Bedford Forrest (a former Confederate General) in 1872.
Now, onto the movie. Birth of a Nation (originally titled “The Clansman”) by D.W. Griffith follows two Southern families around the 1860s and 1870s, and the plot largely exists to villify Black people and celebrate racism, including the Confederacy, the first Ku Klux Klan, and lynchings. Importantly for this post, the Ku Klux Klan are portrayed as unambiguous heroes, Birth of a Nation was a massive success, selling around 10 million tickets, equivalent to 20% of the white population, and was profoundly influential on filmmaking as a discipline, but to its pioneering camera work, and being the first American picture to both feature hundreds of extras in a single scene, and the first one with an orchestra score. It was also controversial upon release, with some cities such as Boston, Chicago, Denver, Pittsburgh, St. Louis, and Minneapolis, as well as the entire state of Ohio even banning it. President Woodrow Wilson (a racist and believer in eugenics) even screened it at the White House for his family and Cabinet, another first for a movie.
Okay, but how does one prove that a wildly popular piece of racist propaganda resulted in more racism? Firstly, interestingly, the movie came out just six months before the Ku Klux Klan was being refounded in the South, and the (first) Klan’s portrayed heroism can be seen as the conduit to KKK membership. For instance, the members of the Second Klan didn’t necessarily wear the uniforms of the first Klan, but rather, the ones portrayed as Klan regalia in the movie - a sign of its influenece on the organization. How do we quantify this influence? The movie’s release calendary and distribution strategy provides an intersting empirical strategy.
Let me explain: back in 1915, there were very few movies made, and seeing them was a hassle: rather than having a movie theater that played the same film all the time for a period, movies were available in different locations for a limited period, all at different times. This format, called the roadshow, was common until the 1960s (and started being more of a luxury experience in the 50s), and one such movie that got a roadshow release was Birth of a Nation. Because the movie was so popular, the roadshow lasted for around five years - at a time with few other movies playing.
The author thus puts together a database of screenings of the movie alongside their dates, as well as registers of movie theaters that played the movie. Interestingly, this was done by tracking advertisement in newspapers, and the ads largely didn’t mention that the movie was KKK propaganda, but rather, the epic scope, stirring emotions, and technical elements of the film. Additionally, the study compiles a database of Ku Klux Klan membership and chapters (such as the location of KKK locales, known as “klaverns”1), as well as a dataset of dates and locations of lynchings.
Moving onto methodology, the author notes that there wasn’t a geographic pattern for Birth screenings (i.e. the screenings weren’t more likely to be held in former Confederate counties). Secondly, and interestingly, there was little correlation between which counties showed Birth of a Nation and racial composition of the county, or with support for President Woodrow Wilson in the previous election. And while there is a correlation between Birth of a Nation screenings and number of Confederate monuments, this correlation inverts when counting monuments per capita, meaning that it’s probably statistical noise. Instead, the largest predictors for a county holding a Birth of a Nation screening are its population, its population density, how urban it was, and its newspaper and magazine penetration - that is, how big a market the county had for mass media at the time.
For the race riots and lynchings, the paper utilizes an event study format, and compares counties that held screenings of Birth of a Nation with counties that did not, before and after the film’s release. This is done via a simple regression of the number of racial violence incidents as a function of county-level fixed effects, time period fixed effects, statewide effects, and the variable of interest, which is whether or not the county screened the movie in a period, and the following six months. The results are that there is a correlation between Birth of a Nation screenings and racial violence, such that “screened counties” had four times more racial violence than non-screened counties. However, due to the small number of lynchings recorded across all counties in the period, the significance is not especially high, but the data does show an increase concentrated in rural or exurban, more Confederate, and Blacker counties, meaning that the movie might have increased racial violence through increased coordination among racists. Using a series of robustness checks, such as selecting a randomized subsample, alternative specifications for the regression, or even comparing exclusively among neighboring counties, results in a similar relationship between Birth of a Nation screenings and racial violence.
Because the event study methodology does not apply to long-run effects, and the second Klan operated across most of the early 20th century, the methodology for the Klan membership effects of Birth of a Nation are estimated differently. The methodology is a regression of whether a given county had or did not have a Klavern in 1930, given the same controlled for factors, and whether the film had or didn’t have a Birth of a Nation screening in 1915. But because it could be possible that (particularly since the movie was screened for five years) the Klan was purposefully coordinating screenings, the movie utilizes whether a county had a movie theater in 1914 as an instrument for screenings.
An instrumental variable, in economics, is a variable which is used as a proxy for another variable. This is useful when there is an unobservable variable that is highly correlated with the variable of interest, but not the proxy. Thus, for the proxy (called an instrument) to be useful, it has to meet two conditions: first, that it is correlated with the variable of interest, but not other variables (aka “relevance”), and secondly, that it is not correlated directly with either a relevant unobservable or with the original outcome of interest. In this case, because Birth of a Nation screenings might have similar, unmeasurable causes to Klan presence (i.e. racism), one can find whether there is a relationship between Birth of a Nation screenings and movie theater presence (provided the same controls and effects), and if there is (particularly a strong one), then one can find the relationship between movie theaters and KKK activity. The author runs the same regression between movie theaters and racial animus measures as was originally used to test whether counties that showed Birth of a Nation were inherently more racist, and finds the same results. One final controlling factor is that Kansas, a state where the Klan was very popular, banned the movie until 1920, which means that a number of Kansas counties should have no relationship between Klaverns and 1914 movie theaters if there are no confounders.
The results of the movie theater/racial violence regressions are simple: before the movie was released, there was no relationship between racial violence and movie theaters; after the movie was released, there is an immediate spike in lynchings and race riots. Likewise, the presence of a movie theater in a county in 1914 implied a 68.5% increase in likelihood of the same county having at least one Klavern in 1930, which holds true even after dropping a series of potential outliers. Adjusting the number of Klan venues for the adult white male population (i.e. the Klan aged demographic) results in an even larger effect of Birth of a Nation screenings, which, if extrapolated alongside the first set of results, would mean that the movie increased the ranks of the KKK by roughly one million members. Regressing the same variables by a series of state-level KKK membership estimates results in a similarly strong relationship, as well as an estimated increase of 1.5 million members.
One final result is whether the impacts of the movie on the Second Klan, which seem pretty convincing, remained over time. To test this, the authors run a similar regression to the one outlined above, but for the Third Klan of the 1960s rather than the Second Klan of the 20s, 30s, and 40s2. The Third Klan emerged over the 50s and 60s as a direct response to the Civil Rights movement and was far smaller and less organized. Using the same instrumental variables estimates for the number of Third Klan venues in the 1960s finds a persistent relationship, but rather than not being driven by any factor in particular, these results are largely concentrated among the same historically racist counties outlined a few paragraphs ago, meaning that the film’s persistence in the racist psyche are more dependent on pre-existing racial prejudice. Using a similar approach for contemporary hate crimes finds significant effects on hate crimes against Black people, but also other ethnic minorities, as well as LGBT individuals - in fact, hate crimes were 88% more likely in those counties than in counties that did not screen Birth of a Nation in 1915. It seems, however, that these contemporary effects are also driven by a persistence of existing prejudices; thus, the movie’s enduring legacy is more through permitting the Second Klan than by directly influencing the beliefs of non-racists.
To finish up, some links:
The paper in question
A recent NBER paper about how the success of movies is influenced by long-standing cultural and economic attitudes
A Substack post by Nicolás Azjenman about the same paper, in Spanish
My two blog posts on the economics of Oscar-nominated movies
Or “Klan Taverns”. Besides being responsible for horrific acts of racial violence, the Klan was also kind of silly, with a bizarre tendency to add “K” or “KL” at the beginning of random words (as shown in one of the more memorable Freakonomics chapters)
The Second Klan declined from the mid 20s until the mid 40s, largely as a result of coordinated pressure campaigns from Jewish, Black, and religious organizations, as well as several high-profile convictions of KKK officials of heinous crimes against White people, (largely women), which resulted in a swift backlash.