Predicting Formal Employment in Argentina Using Machine Learning Algorithms

Abstract

Formal employment is associated with improved quality of life, as it ensures access to legal protections and benefits. In Argentina, a large proportion of the workforce is in the informal sector, and official labor market data experience significant delays in publication, hindering timely policy implementation. This work applies Machine Learning algorithms to Permanent Household Survey (EPH) data to predict formal employment status. Following CRISP-DM methodology, we implemented multiple imputation for missing data and compared various predictive models. XGBoost demonstrated superior performance with balanced sensitivity and specificity metrics. The results highlight the value of advanced predictive models for anticipating changes in formal employment, enabling more agile and effective labor policy decisions.