A new article by LCDS affiliated DPhil Micol Morellini, published in the European Journal of Population, proposes a novel method for evaluating migration theories by examining how well they reproduce the spatial patterns of migration flows.
Migration research has traditionally focused on identifying the factors that drive migration — such as economic opportunities, distance, or shared language — and evaluating theories using statistical measures like explained variance or the significance of model coefficients. But migration is fundamentally a spatial phenomenon: it describes how people move between places. Despite this, the spatial accuracy of migration models has rarely been assessed directly.
To address this gap, Morellini introduces a simulation-based approach that allows researchers to evaluate how accurately migration theories reproduce observed migration systems. Starting from an observed network of migration flows, the method generates synthetic migration systems based on the assumptions of a given theory. These simulated systems can then be compared to of real-world ones to assess how well the theory captures the spatial structure of migration.
The approach is applied to migration flows between European countries from 2002 to 2021 and illustrated using two well-known migration frameworks: the gravity model and migration systems theory. While both models perform well using conventional goodness-of-fit measures, the analysis reveals that they struggle to reproduce several important spatial features of the European migration system — including the strong reciprocity of migration flows and distinctive migration profiles of Eastern and Northern European countries.
These findings highlight an important methodological insight: good statistical fit does not necessarily mean that a theory accurately captures how migration is organised across space.
As Micol Morellini explains: “Migration theories are usually evaluated by how well they explain the drivers of migration. But migration is also about where people move. By using simulations to test the spatial patterns implied by different theories, we can better understand what they capture well — and where they fall short.”
By integrating simulation methods with spatial analysis, the study offers researchers a new tool for evaluating migration theories and understanding the complex structure of migration systems. The approach also demonstrates how computational methods can open new avenues for demographic research.
Read the article here.
