A Common European Business Cycle: Markov-Switching SUR with Cross-Sectional Weighting

We study the extraction of a common European business cycle from quarterly GDP growth when countries need not provide equally informative signals about the contemporaneous common regime. We show that the usual signal-extraction argument for enlarging a cross section can fail when country-specific regime behavior is imperfectly aligned: additional observations may then weaken rather than strengthen common-state identification. Motivated by this result, we develop a Markov-switching Seemingly Unrelated Regressions (MS–SUR) model with an endogenous positive-definite crosssectional weighting structure. The model distinguishes the informational influence of each economy in identifying the common regime from the strength of its cross-sectional commonality. A separate Markov process governs the innovation covariance matrix, allowing changes in the volatility environment to be separated from the weighting mechanism. Empirically, alternative treatments of the cross section generate materially different common-cycle signals, with the largest differences arising during periods of pronounced cross-country heterogeneity. The endogenous specification assigns approximately 95 percent of the posterior informational weight to Germany and the Netherlands, while the estimated commonality loadings produce a substantially different cross-country ranking. These results show that common-regime inference depends not only on the amount of cross-sectional information available, but also on which observations are most informative about the latent state.