In this research project, we show the importance of considering non-linearities when analyzing financial conditions and the macroeconomic-financial linkages.
Conclusions include:
- Based on a model fit criteria, the introduction of Markov switching in parameters and variances improves the fit of a macroeconomic VAR model with financial variables, with the best fit in an unrestricted model with two switches in coefficients and three switches in variances (2c3v).
- The introduction of Markov switching in parameters and specially in variances, also greatly improves the fit of a DSGE macroeconomic model with financial frictions in long-term debt instruments developed by Carlstrom, Fuerst and Paustian (2S2R3V).
- In the used DSGE model, when allowing for switching in the parameters capturing financial frictions and monetary policy and switching in shocks volatilities there are different, well defined, regimes of high and low financial frictions, high and low monetary policy response to the term premium and high (medium) and low credit shock volatilities regimes.
- If Markov switching in variances is ignored, there is an overestimation of the high coefficient regimes.
- The DSGE without Markov switching requires larger shocks relative to a model with Markov switching in parameters. Events that otherwise might be interpreted as a structural regime switch are accommodated by large shocks.
- The IRFs are markedly different depending on the regime the economy is in.
- The presence of high financial frictions and high financial shocks explained why the Fed had to respond aggressively cutting interest rates and the severity of the 2008 GDP contraction.