Stop, in the name of COVID!

In the wake of the COVID-19 pandemic, travel restrictions implemented to prevent its spread, like the suspension of international transit and closure of borders, first put into place in March 2020, often suddenly, have created complex, fast-evolving networks of restrictions between the countries of origin and destination of migrants and would-be migrants. These restrictions have had a particularly noteworthy impact on migrants from North and West Africa, who have reported experiencing greater impacts from the pandemic on their journeys than migrants from any other region in the world, as flows registered through key transit points in West and Central Africa and irregular arrivals to Europe plummeted, especially along the Western Mediterranean Route key to migrants from North and West Africa. The International Organization for Migration (IOM) has postulated that international migrant stocks have fallen well short of their pre-pandemic projections in West/North Africa, Europe, and globally, by more than 2 million, due to travel restrictions. However, this is not testable with migration data from traditional sources like censuses and population surveys, which on top of pre-existing timeliness and granularity limitations, have had data collection operations delayed, canceled, interrupted, or data quality otherwise seriously compromised by the pandemic. Recognizing these challenges, key migration stakeholders, including the IOM, have called for the use of data from alternative sources, including social media, to fill in these gaps. Inspired by this call, we endeavor to test the hypothesis that COVID-related travel restrictions reduced migrant stock compared to what it would have been in the absence of such restrictions using estimates of expats, or individuals living in a given destination country who formerly lived in a given origin country, from Facebook’s advertising platform. We take advantage of the quasi-natural experiment provided by different countries’ staggered adoption of different levels of travel restrictions, which we formulate as a treatment, and attempt to control for non-travel restriction-related factors that may be simultaneously influencing migration, using the method developed by Arellano and Bond for estimating dynamic linear panel models. Looking specifically at four key origin countries in North and West Africa, Côte d’Ivoire, Algeria, Morocco, and Senegal, and their 23 key destination countries, we estimate that a destination country implementing a total entry ban over the course of a month may have expected a 3.39% reduction in migrant stock compared to the counterfactual in which no travel restrictions were implemented. However, when taking pandemic-related mortality, broader restrictions on activity and movement, and the onset of the global pandemic itself into account, we estimate that a destination country implementing an entry ban over the course of a month may expect a 5.47% increase in migrant stock. While further research is needed on both the impact of the COVID-19 pandemic on migration and using social media data to obtain accurate migration estimates, travel restrictions do not appear to have been effective in curbing migration in the countries that implement them in the context of the wider disruptions wrought by the pandemic.