Forecasting Inflation in Argentina. A Probabilistic Approach

Tomas Marinozzi


2022 - Trabajo ganador del primer puesto (compartido) del Premio Anual de Investigación Económica “Dr. Raúl Prebisch” (Jóvenes Profesionales). Resumen: Probability forecasts are gaining popularity in the macroeconomic discipline as point forecasts lack the ability to capture the level of uncertainty in fundamental variables like inflation, growth, exchange rate or unemployment. This paper explored the use of probability forecasts to predict inflation in Argentina. Scoring rules were used to evaluate several autoregressive models relative to a benchmark. Results showed that parsimonious univariate models had a performance relatively similar to that of the multivariate models around central scenarios but fail to capture tail risks, particularly at longer horizons.