Underlying Mechanisms in the Pioneers and Followers Macro-Pattern: Using Agent Based Models Informed by Survey Data.

Jose Ignacio Carrasco , University of Oxford, Centre on Migration, Policy and Society (COMPAS).
Joelle Mak, London School of Hygiene and Tropical Medicine
Martin Hinsch, University of Southampton

We use international migration data from Colombia to examine the role of micro-level cohort characteristics and environmental in explaining changes in international migration rates over time, by drawing on an Agent Based Model informed by survey data. We seek to understand under what conditions qualitative changes between stages in migration rates emerge. We use the Latin American Migration Project (LAMP) data in this study supplemented with five macro data sources. We use a series of Agent-Based Models (ABM) informed by an empirical parametrization and enhanced by theory to model individual out-migration probabilities. We use LAMP to estimate a discrete-time model of the effect of a number of variables related to sociodemographics and life-course dynamics as well as macro-factors on individuals' probability to migrate. We design an ABM in which the migration decision is based on the regression outputs to simulate individual life-course trajectories, compute migration probabilities based on an individual's state as well as the current external factors. Our results allow for some interesting conclusions. Although the first version of our ABM is a direct stochastic implementation of the statistical model overall migration rates as well as the change of rates over time clearly differ from the empirical data. This suggests that the static statistical analysis fails to capture an important part of the dynamics of the process. The next instance of the model will add new features based on migration theories. We will use Approximate Bayesian Computation to calibrate the ABM against the observed LAMP migration rates.

See extended abstract

 Presented in Session P1. Postercafe