Evaluating the Robustness of a VAR-ECM Model in Recession Tracking: A Post-COVID-19 Analysis

Abstract

Forecasting recessions remains one of the most elusive goals in macroeconomic modeling. This thesis replicates and extends the model proposed by Boczon and Richard (2020), which combines a DSGE framework with an Error Correction Model and time-varying parameters. The extended analysis recalibrates the model and evaluates sensitivity to calibration choices, variable estimation, and data revisions. I assess the model’s robustness by extending the dataset through 2024, incorporating the COVID-19 recession, a period marked by extreme economic volatility and a historic drop in hours worked. While the model tracks past trends well, its forecasting accuracy weakens during the COVID-19 crisis. These findings suggest that while the original hybrid framework remains valuable, its robustness is limited under extreme volatility.