Working Paper: CEPR ID: DP9815
Authors: Claudia Foroni; Pierre Gurin; Massimiliano Marcellino
Abstract: This paper introduces regime switching parameters in the Mixed-Frequency VAR model. We first discuss estimation and inference for Markov-switching Mixed-Frequency VAR (MSMF-VAR) models. Next, we assess the finite sample performance of the technique in Monte-Carlo experiments. Finally, the MSMF-VAR model is applied to predict GDP growth and business cycle turning points in the euro area. Its performance is compared with that of a number of competing models, including linear and regime switching mixed data sampling (MIDAS) models. The results suggest that MSMF-VAR models are particularly useful to estimate the status of economic activity.
Keywords: Forecasting; Markov-Switching; MIDAS; Mixed-Frequency VAR; Nowcasting
JEL Codes: C53; E32; E37
Edges that are evidenced by causal inference methods are in orange, and the rest are in light blue.
| Cause | Effect |
|---|---|
| MSMFVAR model (C32) | improved estimation of economic activity status (E01) |
| MSMFVAR model (C32) | high-frequency estimates of low-frequency variables (C58) |
| MSMFVAR model estimated via Kalman filter (MSMFVARKF) (C32) | better estimates of in-sample regime probabilities (C51) |
| second regime identified by MSMFVAR model (C32) | higher volatility and lower GDP growth (E39) |
| MSMFVAR model (C32) | improved forecasting accuracy (C53) |
| MSMFVAR model (C32) | better prediction of GDP growth (E20) |
| MSMFVAR model (C32) | better prediction of business cycle turning points (E32) |