Working Paper: NBER ID: w29535
Authors: Frank Schorfheide; Dongho Song
Abstract: We resuscitated the mixed-frequency vector autoregression (MF-VAR) developed in Schorfheide and Song (2015, JBES) to generate macroeconomic forecasts for the U.S. during the COVID-19 pandemic in real time. The model combines eleven time series observed at two frequencies: quarterly and monthly. We deliberately did not modify the model specification in view of the COVID-19 outbreak, except for the exclusion of crisis observations from the estimation sample. We compare the MF-VAR forecasts to the median forecast from the Survey of Professional Forecasters (SPF). While the MF-VAR performed poorly during 2020:Q2, subsequent forecasts were at par with the SPF forecasts. We show that excluding a few months of extreme observations is a promising way of handling VAR estimation going forward, as an alternative of a sophisticated modeling of outliers.
Keywords: No keywords provided
JEL Codes: C11; C32; C53
Edges that are evidenced by causal inference methods are in orange, and the rest are in light blue.
| Cause | Effect |
|---|---|
| mfvar model's performance during initial months of the pandemic (2020 Q2) (E17) | poor performance compared to SPF forecasts (E17) |
| absence of critical data during estimation period (C51) | model's failure to anticipate recession's magnitude (E17) |
| July 2020 onwards (Y50) | mfvar model generated accurate forecasts (E17) |
| model's ability to propagate large shocks through a persistent VAR structure (C32) | improvement in predictive capability (C53) |
| excluding extreme observations from estimation sample (C51) | improvement in forecast accuracy (C53) |
| model's adjustments (C51) | better forecasts post-pandemic onset (C53) |