Economist Nathan Nunn and his team measure the long-term impacts of the Tulsa massacre on Black communities in the US and find a pattern that resonates with an earlier analysis of the transatlantic slave trade, where violence begets long-term economic and social disadvantages that span generations.
By Michelle Nicholasen
Until recently, many Americans, both white and Black, were unaware of the massacre that took place in the Greenwood neighborhood of Tulsa, Oklahoma, 100 years ago this spring. It began with an alleged assault by a young Black man named Dick Rowland on a while female elevator operator named Sarah Page. The details are unknown, but the incident led to the arrest of Rowland the next day. Fearful of a lynching, a group of armed Black men arrived at the courthouse to defend Rowland. A firefight erupted, and the scene escalated to chaos, with widespread looting, arson, and death. In the most egregious act of racial violence in twentieth-century US history, thirty-five square blocks of a prosperous Black community were completely burned down by white attackers “deputized” by the local police. Martial law was declared, hundreds were killed in the rampage, and “Black Wall Street” was effectively erased from the map.
Nearly a century later, economist Nathan Nunn and his collaborators have measured the enduring effects of the massacre on Black communities in the US by looking at two factors: rates of home ownership and occupational status over time. Nunn’s previous research on the economic impact of the slave trade on African countries eerily parallels findings from the Tulsa study. We asked Nunn to give us a closer look at his methodology and to discuss what his findings reveal about historic memory.
Q: Looking at a one-time event, like the Tulsa massacre, is very different from your previous studies that analyze long-term processes like the slave trade or the use of the plough in early societies. Why did you think a single event like Tulsa would lend itself to a similar analysis?
A: Technically, the massacre can be described the same way as the slave trade: it's a temporary event of a finite period; only it’s a three-day event, rather than a 400-year event. We wanted to see if it had permanent lasting effects, at least up until today. Another reason is that it's such an exceptional event. Even among other forms of racial violence from the early twentieth-century US, it's exceptional in the amount of devastation, the amount of wealth that was destroyed.
Q: The event was localized to Tulsa; why did you think it would have far-reaching consequences?
A: The more I looked into it, the more it became clear by actually reading newspaper articles directly, that this was plausibly a warning to other Black communities in the United States.
The way the event was described by lots of individuals and in newspaper articles, was that of a warning. Most newspaper articles wrote about the event as if it was a good thing, along the lines of: it's actually better for everyone, including the African American community in the long run, because Greenwood was crime-ridden and there was prostitution, and alcohol and so on.
So, it was a terrible event for African Americans from Tulsa, but then the other question was: How did it affect Black communities elsewhere as well? It seemed plausible that it could have had an effect and so was worth looking into.
Q: Let's start with the Black residents in the city of Tulsa. You looked at home ownership and occupational status. Why did you choose those variables and what did you find?
A: We looked at home ownership, rather than renting, and the reason why that's the natural outcome is that if you owned a home, it was destroyed in the massacre—with almost certainty.
And it turns out, even though there was a decade of trying, you didn't get money back through insurance claims because race riots aren’t covered by insurance, so it’s a completely lost investment. So obviously that'll affect the African Americans within Tulsa because they had an asset loss.
But another question is—how is home ownership affected in other African American communities that just heard about the event through newspapers. And you could see why, if you thought, “oh there's a probability that the same thing will happen to my community,” that you might not invest in a property which will just be burned down or could be burned down, but instead you’ll rent.
Q: You looked at African Americans living both in Greenwood and in other parts of Tulsa, correct?
A: We looked inside and outside Greenwood simultaneously, and we found pretty large effects. For Black Tulsans, the likelihood of owning a home in the decades after the massacre if you were Black was reduced by 4.5 percentage points.
So, we found a large direct effect, which in some ways isn't surprising because their homes were destroyed and the whole neighborhood was rezoned.
Q: Before the massacre, a large share of Greenwood residents had white collar jobs, and there were some 191 businesses in the area. What happened afterward in terms of employment?
A: After the massacre, what you see over time is occupational downgrading. If you assign a score to each occupation in terms of its status or its earnings, you find that the massacre reduced the occupation score of Black Tulsans residents by 9.6 percent; the massacre also decreased the likelihood that they held white collar jobs by 2.8 percentage points.
Q: And then you looked at Black communities across the US using these same variables. What did you find outside of Oklahoma?
A: The starting question was: Did the Greenwood massacre have effects in other communities across the United States? It’s really impossible to answer which Black communities should be affected more from the events. So, we take a stand and ask—how did those communities learn about the massacre? Well, at the time it's going to be through newspaper coverage and so that's what we measure. Basically, how much newspaper coverage was there, do we find spillover effects, and are the spillover effects stronger for those locations that received more newspaper coverage? So, we look to see what happens to Black communities that are outside of Tulsa after the massacre and how this relates to the extent of newspaper coverage. We find that greater newspaper coverage of the massacre led to reduced rates of home ownership for Black communities in the decades that followed.
Q: Then you also add another condition: the level of segregation in a community. Why did you add that variable?
A: People brought this up in some of our talks, and I think it’s a valid hypothesis. What happened in Tulsa was really only possible because it was so segregated. If you have intermixed houses—Black-owned, white-owned, Black-owned, white-owned, etc., then white populations can’t easily burn the Black-owned homes down selectively. And so that is the other thing that may be important. If you're in a highly segregated area, you're going to think, “Oh, this is more likely to happen to me,” and therefore it will have larger effects on home ownership. Similar to newspaper coverage, we find that home ownership rates were lower for more segregated Black communities in the decades that followed the massacre.
Q: What was the overall impact on Black communities throughout the US?
A: There are spillover effects. They're weaker and also statistically noisier, but consistent.
So, if you're a Black community outside of Tulsa after the massacre and receive more newspaper exposure, then you have lower rates of home ownership subsequently, compared to previous years. And similarly, in places that are more segregated you see the same effect, so the two things seem to matter for home ownership.
Q: And you find exactly the same effect when you look at newspaper coverage?
A: Yes. When there's little or no news coverage of the event, then the effect is zero. When news coverage is higher, the negative effect on home ownership starts to be larger.
I’ll just mention one other thing which is interesting. We find that it's important to account for the spillover effects themselves when estimating the direct effect of the massacre on Black Tulsans. The reason is, if you don't account for them, what you're doing is you're comparing African American communities in Tulsa after 1921 relative to before with other African American communities in the United States. But, if there's spillover effects, then what we're thinking of as our “control” was actually also exposed to the massacre through newspaper coverage, and so are harmed as well. When you measure that difference, it's going to be smaller than the effect really is.
When we account for those spillovers, what we find is the direct effect on residents of Tulsa goes up by 100 percent. So the roughly 5 percent is actually an underestimate and ends up being more like a 10 percent reduction in home ownership. In other words, once we properly control for the fact that other communities also got a dose of treatment [newspaper exposure] then the estimated effect on the residents of Tulsa ends up being twice as large.
Q: You rely on newspaper coverage in other parts of the country as the “exposure” Blacks would have about the massacre, and you connect this to reduction in home ownership over time. Doesn’t this model make a lot of assumptions? For example, couldn’t there be other racial incidents going on in a location or region that could be affecting behavior, not just Tulsa?
A: Two things to mention: First, you could imagine that if it's after 1921, the Tulsa massacre is in the news and it affects the dynamics—the social and racial dynamics in Maine, for example—and so that's endogenous to it.
Also, there was a series of riots in 1919, and once you start looking into this, there's racial violence everywhere. So we try to control for the incidents that are the highest profile and may have had similar effects; so in our regressions we account for the wave of violence that happened in 1919.
If it was a violent year, that'll be captured in what we call “year-fixed effects.” In the case where certain counties are just always more violent than others, we have “count-fixed effects” to account for that. All of the estimations are always across three dimensions: space, race, and time.
Q: In your paper about the Atlantic slave trade, you showed that the more slaves taken from a particular area in Africa, the lower the per capita income of those countries today. That is a powerful finding, and I see a direct parallel to your Tulsa work. Do you as well?
A: In both cases a temporary event has had permanent impacts. A temporary event put these communities or countries on different trajectories. If it was an effect that mattered only at the time, then you would expect, after the events are over, that any effects should disappear. Instead, the events we’re looking at do seem to put the societies being examined on different paths of development.
In the case of Tulsa, that's why we thought it was important to look up until the year 2000 to see if those effects were declining over time. In fact, they are growing. Similar to the slave trade, the negative effects seem to be growing slightly over time. It’s not like things are getting better and the history or memory of that event is fading. If anything, the memory or importance of the event is growing.
Q: Are you saying that in both studies the negative consequences increase?
A: Yes. When you look up to 2000, you still see an effect of the event, whether it was the massacre or the slave trade, so, that's the first thing: there is a historical memory.
If you look at the Tulsa paper, it covers 1980–2000 and that's just because of data availability. If you look decade by decade, the effect of that event from 1921 is growing.
And in the slave trade paper, I think I looked at 1950 on, decade by decade, and the effect of the slave trade is growing as well, in terms of per capita income in these locations that lost the most individuals to the slave trade.
So, it's very consistent with just a different trajectory; as time goes on, the gap between the reality and the counterfactual—what would have been—gets further and further apart.
Q: Whereas you expect there to be recovery?
A: That's exactly right. I think most economic models would suggest recovery: there's one equilibrium, you perturb it, and this moves us away from the equilibrium. Then that event disappears, and you converge back to the one equilibrium. This work is telling you that the world is more complicated, and there are many possible paths—and not just one equilibrium that you are shocked away from but then move back to in the long run.
Q: What are the global or international implications of studies like this, sometimes referred to as the “persistence literature”?
A: At a minimum what they show is that these events that happened in the past still have legacies that matter today. They provide evidence for thinking about how we put populations or groups of people or different societies on equal footing. What our papers show is that these initial differences can grow over time. You can imagine that the gap between different groups will continue to grow 100 years from now, that African countries that had a lot of slaves exported will continue to be poor. So it raises the question for policies of, perhaps, reparations in the United States or similar reparations internationally to help former colonies develop.
These are just two events we’re talking about here. Historically, there were many, many more. History is experienced by all societies everywhere, so there are lots of dynamics like these around the world.
Studies like this can inspire us to ask: Is there something we can do—or should do—through our policies today to try to counteract some of these historical forces that lead to severe inequality and adverse outcomes?
Q: Otherwise, it’s a dark prognosis, right?
A: People often think this is all very bleak. They don't like it because it's too negative. But actually, the other way to look at it is that any policies that we implement right now, even if they are temporary, are potentially going to have effects that last for hundreds or thousands of years.
So it's a double-edged sword. Things that happened in the past, yeah, they last a long time.
But any corrections we make, even a temporary correction—so let's say affirmative action that equalizes the levels of university education across different groups—that might only take a few decades, or a generation or two, but we could have that forever; basically, absent any other shocks.
So that's the other neat thing about this work; it's showing you that temporary events, temporary policies, can have permanent lasting effects. A more positive way of thinking about it is that we can induce effects that are persistent and positive, and not just study events that happened long ago and are persistent and negative.
—Michelle Nicholasen, Editor and Content Producer, Weatherhead Center for International Affairs
Faculty Associate Nathan Nunn is the Frederic E. Abbe Professor of Economics at Harvard University. He studies economic development, political economy, economic history, and Africa.
- All that was left of his home after Tulsa Race Riot, 6-1-1921 | A postcard showing an unidentified man standing alone amidst the desolation and the ruins of what is described as his home. The placement of the ruins of Dunbar Elementary School in the background indicates that this photo was taken on either North Greenwood Ave. or North Frankfort Ave. Credit: McFarlin Library Special Collections, part of a collection of postcards on the Tulsa Race Massacre at McFarlin Library
- The Morning Tulsa Daily World | Front page of the newspaper on Saturday, June 4, 1921. Credit: Chronicling America: Historic American Newspapers, Lib. of Congress
- Newspaper clipping with the article, “Refuse to Pay Tulsa Negroes’ Damage Claims,” The Kansas City Kansan, Vol. XXV No. 332, June 2, 1921, pg. 1. Source: newspapers.com
- Fraction of Newspapers that Contain the Word ‘Tulsa’ on the Front Page, June 1–4, 1921, page 17. Credit: After The Burning: The Economic Effects of the 1921 Tulsa Race Massacre working paper by Albright A, Cook J, Feigenbaum J, Kincaide L, Long J, Nunn N
- Paths of Economic Development Since 1950, page 167. Credit: The Long Term Effects of Africa's Slave Trades paper by Nunn N, Quarterly Journal of Economics. 2008; 123 (1) : 139-176
- Entrance to Refugee Camp, Fairgrounds | During the massacre, approximately 6,000 Blacks were rounded up and detained in camps throughout the city, such as the Convention Hall and the Fairgrounds. Some were detained for up to six days. Credit: DR-6.08 Oklahoma, Tulsa Co. Riot Reports and Statistics, page 161. Records of the American National Red Cross, 1881–2008