Working Paper: CEPR ID: DP9705
Authors: Francesco Bianchi
Abstract: I develop a toolbox to analyze the properties of multivariate Markov-switching models. I first derive analytical formulas for the evolution of first and second moments, taking into account the possibility of regime changes. The formulas are then used to characterize the evolution of expectations and uncertainty, the propagation of the shocks, the contribution of the shocks to the overall volatility, and the welfare implications of regime changes in general equilibrium models. Then, I show how the methods can be used to capture the link between uncertainty and the state of the economy. Finally, I generalize Campbell's VAR implementation of Campbell and Shiller's present value decomposition to allow for parameter instability. The applications reveal the importance of taking into account the effects of regime changes on agents' expectations, welfare, and uncertainty. All results are derived analytically, do not require numerical integration, and are therefore suitable for structural estimation.
Keywords: Bayesian methods; DSGE; Impulse responses; Markov-switching; Uncertainty; VAR; Welfare
JEL Codes: C11; C32; E31; E52; G12
Edges that are evidenced by causal inference methods are in orange, and the rest are in light blue.
| Cause | Effect |
|---|---|
| Regime changes (P39) | Agents' expectations (D84) |
| Regime changes (P39) | Uncertainty (D89) |
| High volatility regime (E32) | Greater uncertainty about short-run outcomes (D89) |
| High volatility regime (E32) | Lower uncertainty about long-run outcomes (D89) |
| Neglecting regime changes (P39) | Overstatement of the importance of specific shocks (E32) |
| Regime changes (P39) | Misleading welfare calculations (D69) |
| Regime changes (P39) | Expected quadratic deviations from steady states (C62) |
| Uncertainty (D89) | Welfare assessments (I38) |