This paper deals with the phenomenon of poverty-trap regimes in Mexico, that is, self-reinforcing mechanisms in which municipalities which start poor remain poor. We develop a coordination game of poverty traps driven by strategic interactions of economic agents: people choose to complete or not their education levels since it might be excessively costly and unprofitable. A one-shot game is constructed and then converted into a system of differential equations in which strategies that perform relatively better become more abundant in the population. Applying evolutionary games and symbolic-regimes dynamics (nonparametric and nonlinear techniques), we show that Mexican regions are in poverty-trap regimes (stable and dynamically evolving low-level equilibria) characterized by incomplete education and low income since initial conditions (education and income per capita) are such (very precarious) that poverty is the stable steady-state situation. We examine scenarios to show that to overcome the high-poverty regime by the year 2030, it is necessary to reduce incomplete education by 10% in the 5-year periods 2020–2025 and 2025–2030 and increase per-capita income by 10% in both periods.