A long-term focus on growing cities in Africa
When Stanford University researchers, including Professors Pascaline Dupas and Marcel Fafchamps, set out to test a new way of assessing relative poverty, applying a peer-ranking methodology they devised to data from a large metropolis in Côte d’Ivoire and rural Indonesia, their method worked—in some ways better than other poverty measures.
But, in the urban setting, they ran up against a problem: Most people did not know their neighbors well enough to accurately rank their relative wealth.
“That surprised us—that people knew so little about their neighbors,” Fafchamps says. “There was a lot of variation, a lot of disagreement.”
The results didn’t discourage the team, however; instead, they opened a new line of inquiry: whether people are actively trying to hide their socioeconomic status from their peers. Research like this—rooted in large-scale, long-term data collection—is at the heart of the Stanford King Center on Global Development’s African Urbanization and Development Research Initiative (AUDRI), which was created nearly a decade ago to study how people and cities in developing countries are changing in response to large-scale migration from rural areas to urban ones. Through AUDRI, Dupas, Fafchamps, and many undergraduate and graduate students, predoctoral research fellows and postdocs working with them have constructed wide-ranging panel data sets, including surveys of individuals and businesses. The data is publicly available to anyone and has allowed researchers to conduct studies and experiments for scholarly purposes and in response to policymakers’ requests.
“The idea was to create projects that had a presence on the ground and a commitment to long-standing data collection,” Fafchamps says. Because research on development has historically focused on rural communities, “we thought that even just documenting cities and how they evolve over time is something that would have lasting value.”
“We started this thinking it would be a public good,” she says.
AUDRI’s work is focused on two of the largest cities in sub-Saharan Africa: Addis Ababa, Ethiopia and Abidjan, Côte d’Ivoire. The Côte d’Ivoire data is particularly compelling because there is relatively little research on Francophone countries on the continent.
A major benefit of the data is that it provides a foundation from which researchers can pivot to provide evidence on new issues as they arise. For instance, when the COVID-19 pandemic began to take hold, AUDRI researchers were able to quickly capture people’s experiences. Among AUDRI’s publications are a policy brief on “Households’ attitudes towards schools reopening in Ethiopia” and panel data evidence on the effects of the pandemic on people’s livelihoods in Côte d’Ivoire.
“One way the data has been very useful is to help governments think through what is going on” in real time, says Dupas, who was director of the King Center until June 30 when she left Stanford to take a faculty position at Princeton University. “We have this infrastructure in place, so we can deliver.”
In the forthcoming study about peer poverty rankings, “Eliciting Poverty Rankings from Urban or Rural Neighbors: Methodology and Empirical Evidence,” Dupas, Fafchamps, and Deivy Houeix, a former predoctoral fellow at the King Center who is now in a PhD program at the Massachusetts Institute of Technology, asked more than 500 respondents in 34 poor neighborhoods in Abidjan to rank the relative wealth of up to 14 other households in their area, and aggregated all the information provided. Although they did not receive enough information back from participants to yield accurate predictions, their method was effective when they applied it to data from rural Indonesia collected by other researchers in 2012.
“The methodology was definitely an innovation, and we demonstrated that the methodology works and that it does perform better than existing technologies,” Fafchamps says. “That adds to the toolkits of economists.”
Having accurate ways to measure relative income is important to ensure that government programs designed to alleviate poverty benefit the right people.
“These are countries where the overwhelming majority of the population doesn’t have a wage job,” Fafchamps says. “You can’t rely on pay slips.”
After the peer ranking data from Côte d’Ivoire came in, Dupas, Fafchamps, and current predoctoral fellow Laura Hernandez Nuñez began to wonder if there was a reason people didn’t know much about their neighbors’ income levels. They were familiar with existing studies on people’s concerns about their own income compared to others’ incomes—including a well-known experiment in which half of participants said they would rather earn $50,000 a year in a world in which others earned $25,000 than earn $100,000 a year while others earned $200,000.
“We were circling around the idea of social image,” says Hernandez Nuñez. “We wanted to try it in this setting.”
The team conducted a series of experiments designed to find out whether people are treated worse (e.g., more likely to be suspected of a crime or less likely to be selected for a job interview) based purely on their socioeconomic status; and whether they need, as a result, to try to prevent others from knowing their true socioeconomic status. The answer to both questions was yes. In one particularly striking example, when participants were told they were interviewing for a job and could receive either a free meal or a pen, they were less likely to choose the meal if they knew their choice would be revealed to the interviewer. In another, participants who had the option to donate money to a charity or to receive a basket of food and household goods were more likely to donate money if the enumerator was present, even if they had very little resources of their own.
Using the data from Addis Ababa, Fafchamps and Dupas are also working as part of a team studying management quality among firms in Ethiopia. They hypothesized that firms were not offering training to their employees out of a fear that competitors would then entice those employees away. But their interventions did not increase interest in management training. The researchers are still analyzing the results, but Fafchamps says the results indicate that firms in the country are “not worried about competition.”
“It’s always exciting when the data doesn’t agree with what is the most common understanding of a situation,” Dupas says. “We need to think of new theories to explain that.”
She adds that one benefit of AUDRI—and the King Center—is that it provides consistent funding over the long term so that scholars can try to answer new questions that emerge from their research.
“In some sense, AUDRI has been a proof-of-concept,” Dupas says. “It encouraged the King Center to pursue this model of long-term, interdisciplinary research, which has been transformative.”
The initiative has also inspired students to pursue the field of development economics. One of AUDRI’s former predoctoral fellows, Eva Lestant, now a PhD candidate at Stanford, is launching her own data collection effort in Côte d’Ivoire this summer; Hernandez Nuñez is starting a PhD program at Columbia University in the fall. Two former AUDRI predoctoral fellows have similarly started PhDs at top U.S. universities: Houeix at MIT and Daniel Agness at UC Berkeley.
AUDRI has been an “anchor” for development economics at Stanford, Fafchamps says.
“Now the community is thriving,” he says.