We propose a straightforward algorithm to estimate large Bayesian time-varying parameter vector autoregressions with mixture innovation components for each coefficient in the system. The computational burden becomes manageable by approximating the mixture indicators driving the time-variation in...