Economic Data Helps Explain a Pattern of Violence Against Myanmar’s Rohingya Minority
The leaders of Myanmar have rejected accusations of systematic state violence against the country’s Rohingya ethnic group. But new research co-authored by Yale SOM’s Mushfiq Mobarak shows that the violence and looting in rice-growing areas is tied to rice prices, suggesting an economic motivation for the attacks, and finds that the government response to conflicts involving the Rohingya is far harsher than in conflicts with other ethnic groups.
Rohingya refugees at a refugee camp in Bangladesh’s Cox’s Bazar district in May 2024.
In 2017, hundreds of thousands of Rohingya refugees, part of Myanmar’s Muslim minority, crossed the country’s northern border into Bangladesh, fleeing widespread state violence in their home state of Rakhine. The mass exodus captured the world's attention, as did the stories the refugees told: of indiscriminate violence and killing of civilians, sexual violence, arson, and other crimes committed by government forces.
After The Gambia brought a case of genocide-related claims against Myanmar at the International Court of Justice, Myanmar’s former head of state Aung San Suu Kyi dismissed the allegations as “inaccurate” and “exaggerated.” In interviews, she described violence against the Rohingya as essentially random, and often triggered by Rohingya militias themselves.
Two years later, Yale SOM’s Mushfiq Mobarak, an economist who has worked extensively in South Asia and is himself a Bangladeshi citizen, visited the refugee camps in Cox’s Bazar, where an estimated 750,000 Rohingya refugees have resettled, and spoke to people who had witnessed the violence themselves. With funding from several Yale institutions, NGOs, and the World Bank, he and his colleagues oversaw a representative survey of 4,616 Rohingya refugees, and a similar number of Bangladeshis living alongside them, to document what had happened to the Rohingya and how their arrival had affected the host community. Their study is forthcoming in the Economic Journal.
Hey, I’m AInsights
Ask me questions about the article and its underlying research. Let’s chat!
In the refugees’ experience, the violence, which included massacres and the razing and looting of villages, was far from random. It built on and intensified a long-standing pattern of discrimination and exclusion against the Rohingya population designed to disrupt their livelihoods and encourage self-deportation. For example, Mobarak describes meeting a former professor who had his teaching license taken away by the government, and had to move back to his rural hometown in Rakhine province. He began a small business selling medicine, only to then have his freedom of movement circumscribed through a network of government checkpoints. Although his family had lived in Rakhine for at least 150 years, his inability to earn a living left him no choice but to flee, leaving his land and property behind.
But these stories, as powerful and as numerous as they were, were too easy to refute by the Myanmar government in court. “That’s data, but that’s not a large sample of statistically analyzable, quantitative data,” says study co-author Jaya Wen, a professor at Harvard Business School who was a PhD student at Yale when the data collection commenced. “Aung San Suu Kyi and the Burmese government suggested that the refugees might be exaggerating their claims, as it could serve their interests,” adds Paula López-Peña, now a professor at Queens University, who oversaw the data collection in Bangladesh while a postdoctoral scholar at Yale.
The authors wondered if there was a way to use the tools of economics—data, statistics, pattern detection—to test claims made by Suu Kyi.
“We were thinking what would be useful is to see whether or not there are actually any systematic underlying motives to the violence,” says author C. Austin Davis of American University, also a former postdoctoral scholar at Yale. “That would help us understand whether or not it is random, as Suu Kyi is claiming, and coming from all sides—or, if it is motivated and systematic in one direction.”
The four authors analyzed the conflict and found that the state-perpetrated violence against civilians had economic drivers. In a country with multiple ongoing ethnic conflicts, they also identified a pattern of disproportionate response in conflicts involving the Rohingya people.
Their research adds to a small but growing body of work examining economic drivers of conflict, and the first to analyze an ongoing case under consideration at the International Court of Justice. The existing literature has put forth several ways that factors such as global commodity prices can encourage or discourage fighting. A rise in prices for, say, valuable minerals can increase the incentive for factions to expropriate valuable mining areas—or it can disincentivize fighting because the opportunity cost of waging war, instead of mining and selling the minerals, goes up. Or an international price increase could give a stronger group more resources with which to wage conflict, by allowing it to purchase more weapons or hire mercenaries.
The authors wondered whether and how the price of rice, Myanmar’s major agricultural product, might affect conflict there. But because the country’s government has kept the economy essentially closed for years, it was hard to come by reliable data. So Davis, López-Peña, Mobarak, and Wen had to generate the relevant datasets themselves. They began with the Armed Conflict Location and Event Data Project (ACLED) dataset, which codes episodes of political violence and social unrest worldwide collated from a range of news and information sources. In addition to the number of fatalities per incident, the dataset includes information on violence against civilians, and property crimes such as destruction and looting.
They combined this data with an existing index of domestic and international rice prices, as well as a database that maps the suitability of a country’s terrain for growing rice, down to the township level. The climate, soil, and terrain data needed to assess local conditions, Mobarak notes, did not come from inside Myanmar; instead, it had to be generated using satellite imagery and comparable data that had been gathered from neighboring Bangladesh.
This methodology highlights how economists can use technology to obtain data in information-poor environments, or in situations where access to a field site might not be possible. With the satellite imagery, he notes, “We can get information about Myanmar without having to rely on the Myanmar government allowing us to do any work on the ground.” This circumvents a common challenge with studying conflict settings and authoritarian regimes.
Although Myanmar only exports 9% of its rice output, the researchers focused on international prices to avoid any confounding effects of domestic rice price volatility on the domestic economy which could independently affect conflict propensity.
Using all of this data, they calculated how the likelihood of violence would change in a township with average rice suitability when rice prices increased from their 25th to 75th percentile. Such a price hike, they found, increases the probability of any violence against civilians in such locations by 55%, and state-sponsored violence against civilians by 39%. The likelihood of looting fully triples when rice prices rise. Rising international rice prices increases the returns to expropriation of harvests, but it also heightens the appeal of land theft. Mobarak explains. “Suddenly the value of that rice-growing land becomes more salient.”
The interaction of rising prices with an area’s suitability for rice growing is especially powerful in Rakhine state, home to nearly all of Myanmar’s Rohingya population. State forces were even more likely to react to an international rice price hike with violence against civilians and looting in Rakhine state: the statistical probability of violence there, the researchers found, was an order of magnitude larger than for their nationwide calculations. Those results aligned with their survey results from the Cox’s Bazar refugee camps, in which Rohingya refugees from Rakhine reported widespread looting and property theft: 60% lost animals, 56% lost agricultural equipment, and 79% lost land.
Together, these two data sets tell a consistent and compelling story. “The rice-price information doesn’t use any refugee narrative at all, but we are able to show the systematic patterns of violence,” Mobarak says. At the same time, the refugees surveyed “are saying things that are entirely and exactly consistent with what we see in the large sample data.”
The economic explanation, though, does not account for what appears to be especially harsh treatment of Rohingya, something that Suu Kyi has also denied. Myanmar is home to numerous ongoing conflicts, in which armed groups claiming to represent the interests of minority populations such as the Shan, Karen, and Kachin ethnic groups battle with state forces. This unfortunate reality allowed the authors to test whether the violence against the Rohingya was exceptional or merely comparable to the state’s response to other ethnic militias.
They found that government forces responded overwhelmingly to outbreaks in the Rohingya conflict, with additional attacks and a far higher rate of civilian casualties than in other ethnic conflicts. For example, for each bilateral event (a clash involving both the state forces and the ethnic militia), Rohingya are targeted with an average of six additional unilateral state attacks, compared with an average of 0.58 attacks against two other groups. Similarly, Rohingya civilian fatalities increase by an average of nearly 60 in the month following the instigating event, while civilian fatalities from state forces do not significantly increase for the other groups.
“In other conflicts, for every attack, there is a proportionate counterattack. For this one, there is a huge disproportionate reaction,” López-Peña says. “There’s directed animosity here. It’s much more harsh.”
There are many other ongoing conflicts around the world, Wen notes, that follow a similar pattern: rights suppression followed by violent outbursts. Economists could subject them to this same sort of rigorous analysis, in order to identify causes of conflict or provide clarity about whether a response is proportionate and in accordance with humanitarian principles and the laws of war.
“Environments with ongoing civil conflict, war, or oppression are often data-poor, and it’s not possible for social scientists to directly collect large-sample data,” says Davis. “Our indirect methods of inference employing satellite data or aggregated newspaper reports may be useful to shed light on such events, whether it’s in South Sudan, Myanmar, Ukraine, or Palestine.”