RESUMEN:

The Mexico–United States bilateral migration border corridor is the most heavily traveled in the world. Poverty has traditionally been identified as the primary driver of migration; however, analyzing this relationship poses significant challenges due to its nonlinear nature, the interaction with remittances, and the pronounced regional heterogeneity across Mexico. The main objective of this study is to determine the most robust econometric specification for analyzing this phenomenon by comparing the Ordinary Least Squares (OLS) model with the Two-Way Fixed Effects (TWFE) model. A balanced state-level panel dataset was constructed using temporal lags, OLS estimations, the within method, and Least Squares Dummy Variables (LSDV). In addition, Newey–West robust standard errors were employed to address heteroskedasticity and autocorrelation issues. The results confirm the statistical superiority of the TWFE model, reducing the residual standard deviation by 54.38% and increasing the Cox–Snell fit measure by 80.53% relative to the OLS model, while improving the AIC and BIC criteria by 41.63% and 29.49%, respectively. A critical poverty threshold of 53.81% was identified (rising to 67.17% in the high-poverty stratum), validating the existence of a liquidity trap and the catalytic effect of remittances. The implementation of two-way fixed effects proved to be the optimal approach for ensuring consistent inference by isolating unobservable heterogeneity and capturing the pure within-state variation, thereby providing a robust framework for modeling migration dynamics in Mexico. Specifically, these results can support decision-making related to migration public policies between Mexico and the United States, considering poverty as a topic of discussion.

Keywords: Mexico–United States migration, poverty, econometric model, two-way fixed effects.