Description

We estimate the full predictive distribution of GDP growth in a small open economy, using Chile as a case study. Following Sokol (2025), we implement a Bayesian conditional quantile regression framework that conditions on leads of domestic variables and on a broad information set summarized by foreign and domestic factors. Our results show that domestic labor-market and financial conditions substantially improve out-of-sample predictive accuracy, reduce quantile loss scores, and allow earlier identification of tail risks and contraction episodes relative to models relying solely on contemporaneous information of GDP growth. In addition, the framework enables the construction of conditional fan charts that exhibit asymmetries consistent with projected unemployment dynamics.