Origin-Country Social Protection Regimes and Migration Intentions: Evidence from Tunisia

Abstract

This replication package reproduces all results reported in "Origin-Country Social Protection Regimes and Migration Intentions: Evidence from Tunisia" (submitted to Economic Modelling). It supersedes earlier versions of this dataset. The analysis uses microdata from the non-migrant module of the Tunisia Household International Migration Survey (Tunisia-HIMS, 2021), produced by the Institut National de la Statistique (INS). The analytical sample consists of 4,228 non-migrants who were employed before the COVID-19 outbreak. The study compares the migration intentions of workers covered by contributory (employment-based) social insurance or by state (means-tested) assistance with those of comparable unprotected workers. The main estimator is inverse probability weighted regression adjustment (IPWRA) with survey weights, with each regime's effect on the treated estimated among the workers of that regime. Robustness checks include double machine learning, propensity score trimming, Oster (2019) sensitivity bounds, alternative classifications of the outcome and of the treatment, additional controls, standard errors that account for the survey design, and stricter outcomes based on concrete migration plans. Tests of equal effects across subgroups are adjusted for multiple testing. All tables and figures are reproduced from the INS microdata by a single master do-file, which builds every variable and runs every estimation in sequence without manual intervention. The INS data are freely available from the INS website and are not redistributed here. The README file explains how to obtain the data, which software and packages are needed, how to run the do-file, and which output corresponds to each table and figure of the paper.