The Invisible Channel: How Labor Migration Quietly Synchronizes Global Business Cycles

This paper aims to assess whether labor migration networks are a medium for business cycle transmission across economies. To achieve this, this study utilizes a set of causal machine learning algorithms, such as causal forests and double machine learning, to estimate the heterogeneous treatment effects of labor migration on GDP correlation. To obtain this, this study utilizes a set of bilateral migration stock data from the OECD database for 1990-2023, as well as quarterly GDP data from the World Bank. The results show that a one-standard-deviation increase in labor migration network strength can result in a 0.08-0.12 percentage point increase in GDP correlation for a given pair of economies. The results are also robust to various checks, including instrumental variable estimation, where historical patterns of labor migration and geographical distance are utilized as instruments for causal inference.