A quantitatively validated Latin American model for open-data corruption risk detection in public procurement

Abstract

This research presents the Latin American Integrated Risk Model (LAIRM), a formally delineated and reproducible computational framework aimed at detecting wrongdoing, especially in public buying processes across Latin America. The study’s scope is limited to procurement-related corruption, as opposed to other forms of public-sector malfeasance, to enable precise indicator construction and empirical validation, using open contracting data. The LAIRM was developed to address limitations in existing international benchmarks. It uses a mathematical framework based on principal-agent and institutional theories to combine weighted factors, such as non-competitive methods and vendor opacity. Cross-validation and risk tests with open data from Mexico, Brazil, and Colombia showed that the model was correct. The F1-score achieved by LAIRM was 15% higher than that of the World Bank’s Red Flags framework, with an estimated 18–25% improvement in early detection rates. The study demonstrates that the evidence-based method is applicable to large and small-scale projects. Subsequent plans to the study will ensure proper program implementation and the accuracy of the results.