Estimating the effects of the Eurosystem's asset purchase programme at the country level

We assess the macroeconomic effects of the Eurosystem's asset purchases on the four largest euro area economies using simulation exercises that combine unconventional monetary policy shocks with a fixed policy rate for the duration of the purchase programme. We identify unconventional monetary policy shocks in a large Bayesian vector autoregressive (BVAR) model as shocks to the term structure of interest rates using zero and sign restrictions. We propose a multi-country model in which we impose identification assumptions mainly on euro area aggregate financial variables and on country averages of output and price responses. Furthermore, the multi-country structure allows testing for cross-country differences in the effects of the asset purchase programme in a statistically rigorous way using the posterior of the difference between the country-specific effects. We estimate positive output effects in all countries as well as positive effects on bank lending to firms. Effects on HICP inflation, generally, are much weaker. We find substantial cross-country heterogeneity with the largest price level effects in Spain while output effects were smallest in France and inflation effects were smallest in Italy.