Are macro-financial linkages stable or time-varying? Evidence from Bayesian vector autoregressions

This paper investigates the importance of time-varying parameters in US macro-financial linkages. To do so, we adopt a flexible hybrid time-varying parameter Bayesian vector autoregression with stochastic volatility empirical framework. We find that, first, macro-financial linkages are mainly characterized as hybrid time-varying interactions, adequately captured by a combination of stochastic volatility, constant parameters on most lagged effects, and time-varying parameters that mainly capture the contemporaneous effects of macroeconomic variables on financial variables. Second, the relative change in the size of financial shocks, captured by their respective stochastic volatility components, is the main driver of the observed time-varying effects of financial variables on macroeconomic outcomes during periods of financial stress. Third, the combined contribution of credit spread, house and stock prices shocks to unemployment (GDP growth and inflation) fluctuates from approximately 20% (5%) in normal times to 60% (30%) during the Global Financial Crisis, thus indicating that financial shocks affect more importantly labor market outcomes. Fourth, macroeconomic variables respond more significantly to credit spread and house price shocks. Fifth, GDP growth and inflation react differently to financial shocks: while house price shocks and stock price shocks act as demand-type shocks by moving both variables in the same direction; credit spread shocks act as supply-type shocks by moving both variables in opposite directions.