Partial pooling with cross-country priors

A structural Bayesian vector autoregression model predicts that - when accompanied by a decline in consumer confidence - a one-percent decrease in house prices is associated with a contraction of economic activity by 0.2 to 1.2 percent after one year. Results point to important second-round effects and additional exercises highlight the amplifying role of (i ) the mortgage rate and (ii ) consumers' expectations. A novel econometric approach exploits information available from the cross section. Shrinkage towards a cross-country average model helps to compensate for small country samples and reduces estimation uncertainty. As a by-product, the method delivers measures of cross-country heterogeneity.