UNIVERSITY PARK, Pa. — Climate-driven hazards can push many to leave their home in search of livelihood elsewhere, but aspirations for the future may change where they decide to migrate, according to a new study led by researchers at Penn State. At the same time, income losses due to climate shocks, or extreme weather events like floods or droughts, may constrain migration options for vulnerable populations. The interdisciplinary team spanning three continents adapted a migration flow model to see how computational agents representing individuals in the same environment but with different aspirations might navigate climate-driven migration decisions in Senegal.
The model, detailed in a study published in Global Environmental Change, revealed that accounting for the balance of short-term investments, like the cost of moving and paying for further education, and long-term opportunities, like potential income after earning a degree, significantly shifted the geographic migration flow compared to agents operating without livelihood aspirations. These nuanced migration decisions may not be restricted to people living in Senegal, the researchers said. Understanding how people migrate, they explained, could also help policymakers support these populations and build resilience.
Field research, the researchers said, has long emphasized that individuals do not simply respond to present conditions when making decisions to migrate but also consider their future expectations and desires. For this study, the team’s model assigned agents to a livelihood in one of six representative categories, such as agriculture or trades, with income varied by time and location. They simulated 25 years in two different climate shock scenarios: a multi-year drought and rising sea levels. They compared these scenarios to a fixed climate state reference. The team also compared these scenarios to the same environments, but where the agents acted on long-term aspirations rather than just near-term livelihood opportunities. Over the simulated 25 years, the agents could make decisions about moving as they learned more information, like whether the soil could maintain a productive crop or if a new vocational program was available within a more reasonable distance.
The team’s findings show that accounting for those aspirations in the model suggests, on average across the climate shock scenarios, almost 50% more migrants than predicted in the simulations without aspirations, and often to atypical destinations. The team also found that government efforts like investment in vocational training in rural areas can act as shock absorbers, which help mitigate some of the climate impact by creating opportunities that require migrating smaller distances and possibly increasing income potential in the future.
“Our findings demonstrate the value of integrating a livelihood aspirations framework with modeling tools to understand the long-term consequences of climate shocks on migration and economic outcomes,” said Nicolas Choquette-Levy, assistant professor of climate risk and decision making at Penn State and lead author on the paper. “These insights can inform targeted investments in improved access to education and mobility to build resilience to these shocks.”
The team selected Senegal as the testbed for the modeling work because it is situated in the Sahel, a region under major climate threats to agricultural livelihood, and has what the researchers called a “culture of migration.” Migration plays a significant role in the strategies of individuals and households to improve their families’ circumstances and achieve a form of social mobility.
“This model offers a fresh approach to and understanding of migration dynamics in Senegal by combining three significant factors within a population characterized by its youth: social mobility, education and climate change,” said Véronique Petit, deputy director of the Centre Population et Développement — Université Paris Cité and co-author on the paper.
The team previously built a basic model to simulate migration flows driven by climate shocks, but they needed to incorporate additional decision-making factors regarding if, where and why people decided to migrate. The project started with in-person workshops to understand what influenced people to move. Andrew Reid Bell, the Schleifer Family associate professor of sustainability at Cornell University and paper co-author, explained that the workshops provided key information about how individuals viewed migration as a strategy for better education and economic stability. They incorporated this data into the model by allowing the computational agents to consider economic opportunities as a migration driver, along with climate constraints.
Co-author Fabien Cottier, associate professor at Hiroshima University and lecturer at the University of Geneva, explained that agent-based models enable researchers to see how individual choices contribute to population-level changes. The team’s model incorporated data from the Senegal census, other statistical models and the workshops that built in the heterogeneity of the population, Cottier said.
“We want to focus on the attributes that influence individual decision making, like location, social connections, resources and risk aversion,” said Choquette-Levy, also a faculty affiliate with the Institute of Energy and the Environment at Penn State. “We can see how the heterogeneity at the individual level bubbles up to the societal level. And we can look at those population-level trends and drill back down and see how they differ for demographic characteristics.”
Alex de Sherbinin, director and senior research scientist for the Center for Integrated Earth System Information at Columbia Climate School and paper co-author, said that wanting better education and economic opportunities leads more people to migrate, but climate shocks often limit the opportunity.
These limitations translated to atypical migrant destinations. Often, migration models predict movement from rural areas to urban ones, where there are more resources and opportunities for individual development. According to the researchers, the baseline scenario of Senegal migration showed most agents moved from the rural areas to urban areas, such as the capital city of Dakar. However, climate impacts shifted where people could move.
“Dakar is a coastal city, so sea level rise makes it a less desirable place to move or constrains its ability to support migrants,” Choquette-Levy said. “On the other hand, drought in rural areas restricts people’s financial capability to move or invest in things like a bigger farm. One of the few pathways that was left in the model was moving within the interior regions to attend one of the government-sponsored vocational schools.”
These vocational schools can absorb some of the impact from climate shocks by offering opportunities for further education and aspiration fulfillment without the need to migrate as far. The researchers explained that the schools are designed to help the people who are at the greatest risk of climate-induced losses.
“Vocational training is seen as one way to anchor or retain young people in their society by enabling them to obtain a qualification and a job,” Petit said. “It is also a strategy developed by migration governance stakeholders, who offer training in trades and agriculture to repatriated or returning migrants.”
Through the agent-based model, the researchers found that migration and vocational training can help mitigate the consequences of climate shocks in terms of income potential. Understanding how such opportunities intersect with livelihood aspirations could help better predict climate migration and inform policy, Choquette-Levy noted.
“From a policy perspective, one message is that areas that haven’t traditionally had to think about influxes of population might need to be more prepared for that as a function of future climate risks,” Choquette-Levy said.
The researchers said they hope to extend this model to further understand policy scenarios and explore how formal and informal education may help build resilience against climate changes.
Additional co-authors include Ndèye Coumba Diouf and Seynabou Sall of Initiative Prospective Agricole et Rurale; Aboubakr Gambo Boukary of Université Abdou Moumouni de Niamey; Etienne Piguet of University of Neuchâtel and Corentin Visée of University of Namur.
The research was supported by the U.S. National Science Foundation under award number 1934978 and the Department of Defense Minerva program under award number FA9550-22-1-0282. This content is the sole responsibility of the authors and does not necessarily reflect the views of the funders.