by jpfeifer » Fri May 17, 2013 6:04 pm
conditional_variance_decomposition performs a forecast error variance decomposition at the specified horizons, i.e. it tells you for example how much of the variance of output is driven by TFP shocks.
shock_decomposition uses the Kalman smoother to compute the most likely shock realizations that lead to the observed values of the data, e.g. at time t=1, 50% of output deviation from steady state is due to a positive technology shock and 50% due to a monetary policy shock.