Latin American Journal of Central Banking

High-frequency inflation forecasting: A two-step machine learning methodology

Thursday, Jan 1, 2026 by Osmar Bolivar Latin American Journal of Central Banking

This paper introduces a two-step machine-learning method to produce daily and weekly inflation forecasts in developing economies where official CPI is monthly and delayed. A regularized (Ridge/L1) model outperforms econometric and survey benchmarks, validated via Kolmogorov-Smirnov testing.

GDP nowcasting: A machine learning and remote sensing data-based approach for Bolivia

Sunday, Sep 1, 2024 by Osmar Bolivar Latin American Journal of Central Banking

This paper proposes a GDP nowcasting strategy for Bolivia that combines machine learning with traditional indicators and satellite imagery (nighttime lights, built-up area, and vegetation and precipitation measures for major crops), narrowing the economic-activity release lag from six to two months. The approach won first place in the 2023 CEMLA Central Bank Award.