This paper analyses the evolution of systematic risk of banking industries in eight advanced countries using weekly data from 1990 to 2012. The estimation of time-varying betas is done by means of a Bayesian state space model with stochastic volatility, whose results are contrasted with those of the standard M-GARCH and rolling-regression models. We show that both country specific and global events affect the perceived systematic risk, while the impact of the latter differs largely across countries. Finally, our results do not support the previous findings that systematic risk of the banking sector was underestimated before the last financial crisis.
CAPM Time-varying Beta Multivariate GARCH Bayesian State Space Models Stochastic Volatility