FinTech Adoption and Bank Credit Risk: Evidence of a Nonlinear Relationship from Latin America

We examine how FinTech adoption affects credit risk in a panel of commercial banks from Chile, Colombia, Mexico and Peru over 2010–2024, covering more than 90 percent of bank lending in each country. Because structured technology-expenditure data do not exist for these banking systems, we construct cumulative dictionary-based digitalization indices from a raw corpus of 785 annual-report PDFs across six technological dimensions. Panel estimates with Driscoll–Kraay inference document an inverted-U relationship between cumulative digitalization and nonperforming loans: early adoption raises NPLs as lending expands towards borrowers without credit histories, while beyond a threshold efficiency gains in screening dominate and credit risk declines. The pattern is specific to the IT-infrastructure dimension of adoption; it survives various robustness exercises. A Cournot framework with two borrower segments organises the opposing inclusion and efficiency channels that motivate the quadratic specification. The turning point falls with bank size, and slopes differ across countries, so the pooled estimates are averages over heterogeneous national markets. With 82% of observations below the threshold, these banking systems remain largely in the phase where digital adoption is associated with rising credit risk. Palabras claves: Transformación digital, Fintech, riesgo de crédito, banca, América Latina. JEL Classification-JE: G21, O33.