Economist Deniz Sanin was at Starbucks when a bag of specialty coffee from Rwanda caught her eye.

“I Googled it right away,” she recalled. “It turns out, there’s been a coffee boom in the country.”

Four years later, the Economics Department visiting scholar is finishing up a domestic violence study inspired by that chance encounter. Her research was made possible by the rapid expansion in the 21st century of Rwanda’s cooperative mills, which hire women as seasonal workers to help process coffee cherries. Sanin’s analyses show these wage earners are subject to less abuse by their husbands during the harvest season, just when they’re busy at the mills.

“The results are not all cheerful,” said Sanin, whose working paper is under review with the American Economic Review. “But the good news is, now we can use them to shape policy.” 

The development economist brings a longtime interest in women’s issues, particularly outside of wealthier countries like the U.S. She grew up in Istanbul in an environment of “female empowerment,” with an academic/public health expert mother and financial analyst father. The family proved a rarity in Turkey, which has persistent gender gaps in education, by prioritizing their daughter’s education.

Sanin dove into the scholarship on domestic violence and women’s financial empowerment while pursuing her Ph.D. at Georgetown University. She found a 2010 study which confirmed higher earnings meant lower rates of domestic violencefor women in California. But in the developing world, research showed the threat of violence can actually increase for women with access to more money — via family resources or anti-poverty programs that offer cash transfers.

“I was heartbroken to learn it can go the other way,” said Sanin, who vowed to find a natural experiment to investigate the benefits of women earning money outside the home.

In late 2019, Sanin was researching the impacts of Rwanda’s 2008 domestic violence law, which criminalized spousal abuse while allowing women to unliterally divorce violent husbands. Then a trip to that Washington, D.C., Starbucks sparked her curiosity about the country’s coffee industry and gradually pieces started to fall into place.

Sanin immediately found a 2011 case study detailing Rwanda’s shift to higher-quality coffee production, with farm communities building more than 200 cooperatives in the East African country between 2002 and 2012. She also learned women traditionally did the cleaning, drying, and sorting of coffee cherries during the spring-summer harvest in home operations. 

Meanwhile, researching the 2008 legislation brought Sanin into contact with Rwanda’s newly digitized records on monthly domestic violence hospitalizations. She was already acquainted with the country’s Demographic and Health Survey, a source of robust self-reported data collected every five years on domestic violence, labor market outcomes, household spending, and more. 

“All of a sudden, I had my natural experiment,” said Sanin, now an assistant professor at the University of South Carolina.

A male coffee farmer in Rwanda, she learned, traditionally sells the family’s yield at a local market. But those fortunate enough to live near one of the growing number of neighborhood mills earn much more because the membership-based entities export to foreign countries. 

Wives still had to help their husbands harvest. The facilities, however, needed laborers to help with the processing tasks that hadn’t been mechanized, and women were the obvious choice. 

“Since the tasks are female-dominated,” Sanin writes in her paper, “a mill enables the wife in the catchment area to transition from an unpaid family worker in the couple’s plot to a wage worker in the mill for the same tasks as before.” 

In other words, the mills doubly tie a husband’s economic interests to his wife’s ability to perform labor. Looking at data as far back as 2005 allowed Sanin to confirm that domestic violence had decreased for couples living in areas served by a newly opened mill. Wives in these neighborhoods became 29 percent less likely to self-report an incident over the previous 12 months.

Mean hospitalizations for domestic violence within the year (women)

Further analysis revealed this was particularly true during the harvest’s June-to-July peak, with hospitals in catchment areas showing a 14 percent decrease in domestic violence patients. 

“I was not expecting this,” Sanin said. “When I first requested the data from the Minister of Health, I thought I probably didn’t need the monthly data.” 

She then broadened her examination, performing a similar analysis with data on Rwanda’s potato farming regions, where women lack the same opportunities to monetize traditional tasks. “I found no change in hospitalizations in these areas during the potato harvest,” Sanin said.

She also looked at nearby Ethiopia, where social acceptance of domestic violence is high — and divorce remains far more stigmatized. “There was recently a randomized control trial there, with factory jobs randomly given to women,” Sanin said. But within that cultural context, she explained, “researchers did not find an effect on domestic violence.” 

Sanin’s paper parses several possibilities for the variations. One is that women with income have increased bargaining power in their marriages. To test this theory, Sanin analyzed self-reported household decision-making, traditionally left to the husband under Rwandan social norms. Women living in the mills’ catchment areas proved slightly more likely to make financial and contraception decisions either alone or jointly with spouses.

A wife’s income could also relieve financial stressors on the couple. Sanin tested this by comparing self-reported monthly household consumption to the domestic violence hospitalization data.

“I find that monthly consumption was higher right after the harvest season relative to the pre-harvest, although the domestic violence hospitalizations were the same before and right after the harvest,” said Sanin, who continued work on the paper as a 2021-22 visiting research fellow with the Kennedy School’s Women and Public Policy Program.

Separate analyses were used to test the “exposure reduction” theory (that is, the simple fact that working outside the home means less time with abusive partners). “The majority of the couples here are coffee farmers, but there are some where the wife is a coffee farmer and the husband is a truck driver or construction worker,” Sanin explained. “They plausibly did not see each other during the work hours both before and after the mill opening. But once she started earning more, there was still a decrease in domestic violence.” 

The bottom line, she concluded, is that her study shows domestic violence rising and falling according to men’s economic self-interests. “During the harvest season, when a husband’s income is a function of the wife’s labor, it’s really costly for him to incapacitate her,” Sanin said.

But her analysis also points to two conditions necessary for female employment to reduce domestic violence. “You need a context in which divorce is a credible threat, where women can walk out of the marriage,” Sanin said. “And in such a context, it also matters whether the husband derives some kind of economic benefit from the wife’s physical productivity at work.”