@techreport{oai:ir.ide.go.jp:00053021, author = {Hayakawa, Kazunobu and Keola, Souknilanh and Silaphet, Korrakoun and Yamanouchi, Kenta}, month = {Mar}, note = {application/pdf, IDP000847_001, This study applies causal forests, a machine learning framework developed for causal inferences, to estimate the impacts of international Mekong River bridges on the location of foreign firms in 140 districts in Laos. The dependent (target) variable is the number of foreign firms from the economic census aggregated into 10 industries. We generated 72 explanatory (features) variables from remotely sensed data and an online routing system to capture factors that may influence location choices of resource, market, efficiency, and strategic asset-seeking foreign direct investment. We found the average impact of the three Mekong international bridges to attract approximately one foreign firm per district per year. In addition, the predicted impacts of two provincial international bridges (the second and third) were more localized near the bridges and in less varied industrial activities compared with that of the first internal bridge in the capital. Our causal forest identified market size, access to neighboring countries, and several types of land resources as the top features that would attract foreign firms to districts in Laos.}, title = {Estimating the impacts of international bridges on foreign firm locations: a machine learning approach}, year = {2022} }