The number of people moving outside their country of birth has grown rapidly in recent decades, with current estimates suggesting there are around 281 million migrants in the world.  Keeping track of such a huge population of people is difficult, and research from the RAND Institute suggests that social media might play a vital role.

The authors believe that by using data from the Facebook Marketing Application Programming Interface, they can effectively “nowcast” migrant numbers in real-time. The researchers gathered 22 months of data, with a focus on the “lived in” status people use on Facebook, which was used as a proxy for one’s country of origin.

They also gathered official data from a number of sources, including the US Census Bureau, the European Union, Eurostat, UN, and OECD, with a metadata analysis of each dataset undertaken to understand the key characteristics of each one and its relative strengths and weaknesses.

Shrinking the lag

The research was born out of the desire to provide more accurate portrayals of the flows of people around the world, with official data often coming with a time lag that fails to capture either the extent or the nature of migration, which obviously creates difficulties for policymakers.

The researchers also developed a Bayesian model to try and nowcast the official data sources after augmenting them with the data from Facebook.

“In other words, by fitting the model to multiple data sources of varying quality and time resolution, the aim was to predict official statistics (e.g. census) at time points for which these more accurate data sources are unavailable, but less reliable and more timely data sources are available (i.e. the Facebook data),” they explain. “In addition to using Facebook data, the model was designed to capture other features of the available data to improve nowcasting  accuracy, such as time trends, correlations between multiple official data sources, and similarities in migration patterns across corridors.”

The researchers hope that their nowcasts are able to provide a better early-warning system to help policymakers better anticipate shock events and rapid migration trends that might otherwise be appreciated too late to do anything meaningful.  As such, they hope that policymakers will be able to take a more evidence-based approach to their work.  They also believe that the approach provides a good illustration of how so-called “big data” can be combined with more traditional data in this field.

“The major benefit of this novel methodological tool is that it provides estimates of stocks of migrants in real-time, which can be used to highlight rapid and sudden changes in international migrant stocks,” the researchers conclude. “Ultimately, this tool could be used to provide an early-warning system for decision-makers and NGOs to respond to recent trends in international migration.”