Climate change network
Abstract
This project investigates two important aspects of international migration among country pairs. On the one hand, this project proposes a global framework to study the relationship between climate change and bilateral migration. The dataset we build spans the 30-year period over 1990-2017 and includes 150 countries all over the world involving more than 15.000 country pairs. On the other hand, this project analyzes the role of neighboring countries on the migration flows. This is important because although there is extensive micro evidence showing that network effects are important factors explaining people’s individual decisions to emigrate, at the macro level, there is only one paper addressing the impact of neighboring countries on migration flows at the regional (groups of countries) level.
Our work has four main contributions. First, we include a wide set of environmental variables with a worldwide coverage which allows us to have a higher capability to incorporate climate change aspects into the migration analysis. Since climate change influences migration decisions differently across countries and depending on the countries’ economic development, this study will hence permit us to design country-specific policies to give response to climate-induced migration. Second, we consider a large set of country-specific socio-economic, demographic, health, political, and governance factors characterizing the countries of origin and destination in the bilateral migration flows. This extensive list of determinants will permit us to explore the contextual causes of migration and the dynamics of the country-specific characteristics influencing international migration. It will also allow us to disentangle the patterns distinguishing developed and developing economies through time.
Third, relying on spatial autoregressive interaction models estimated through Bayesian modeling and MCMC simulation, we quantify, for the very first time, to what extent neighboring countries affect migration flows between country pairs, accounting at the same time for climate change and controlling for an extensive set of determinants of international migration. Hence, our work sharply contrasts with the existing macro literature on migration which, up to present, has not accounted for these multiple factors simultaneously. Importantly, our spatial autoregressive methodology has the capacity to accommodate multiple spatial weight matrices (capturing several sources of proximity between countries leading to the origin- and destination-spatial dependence), a different list of countries as origins and destinations in the bilateral migration flows, as well as different characteristics for the origin and the destination countries.
Fourth, the existing procedures typically assume linear relationships between the socio-economic, demographic, political, environmental, weather and climate change determinants of migration flows. However, the connectivity matrices that are generated when registering migrants moving from one country to the other are dynamic and the inter-relations among them are non-linear. For instance, how the intensity of migration from one country to the other interplays with the evolution of other migration flows is an empirical question that the existing literature has not yet responded. To account for these non-linear relations, as a second step in the analysis, we rely on functional connectivity, and high-order functional connectivity (which accounts for functional interactions involving three and more countries), to characterize the collective dynamics in the migration networks. Importantly, thanks to this network analysis framework, we assess whether:
- a) (satellite) heat and rain maps stimulate or inhibit the connectivity of migration networks through time;
- b) higher order migration connectivity between groups of countries arises when countries are classified by their level of development and income, their regions, among others.
Project Dates: April 2023 - April 2026
Contact in TSE: Thibault Laurent