The mining channel of climate-induced international migration in Latin America: instrumental-variable evidence and a comparison with agriculture

Mining is central to many Latin American economies, but its relationship with climate-related international migration remains insufficiently understood. We compare agricultural and mineral indicators in an instrumental-variable framework using a 17-country panel, with baseline estimation samples covering 1970–2017. Countries are stratified by multidimensional poverty, and climate exposure is aggregated under six spatial weighting schemes. Under maintained exclusion restrictions, a 1% contraction in mineral rents or depletion is associated with about 0.21% higher emigration in the higher-poverty stratum. The baseline estimate is significant at 10% under country-clustered wild-bootstrap- t inference and depends materially on Bolivia. Inference accommodating both weak identification and few countries does not establish its sign or magnitude. Pooled and middle-income estimates are smaller and sensitive to the instrument set; a between-stratum difference is not established. On a common sample and design, tests reject signed equality between agricultural and mining composite estimands under the reported family-wise adjustments. These comparisons do not establish the population ordering of absolute elasticities or the relative strength of separate causal channels. Baseline control and country-exclusion checks preserve the mining sign, while price-control specifications reveal weak conditional identification and imprecise positive estimates. The results support a descriptive comparison of climate-related mobility and sectoral income indicators, with explicit limits for causal and policy interpretation.