Euro Area business cycles in turbulent times: convergence or decoupling?

Abstract

We study the business cycle properties of the four largest European economies in the wake of the recent recession episodes. The analysis is based on the factors estimated from a multi-country and multi-sector data-rich environment. We measure alikeness of business cycles by studying the synchronization of up and down phases, the convergence properties of country fluctuations towards the Euro Area (EA) cycles and the contribution of the EA factor to national GDP volatilities. While the economic fluctuations of the four EA member states were similar before the global financial turmoil, we gather compelling evidence of an asymmetric behaviour of Spanish fluctuations relative to the EA one.

Keywords:

• hierarchical factor models
• international business cycles
• synchronization and convergence

JEL Classification:

• C51
• E32
• O52

Notes

¹ See the section ‘Why a data-rich environment?’ for an empirical justification.

² These numbers are based on total credit (including bank credit and debt issuances) collected from the BIS, see http://www.bis.org/statistics/credtopriv/credtopriv.xlsx. In particular, we computed annual growth rates based on the outstanding amounts. In our analysis, we use loan growth from the MFI data, which is a narrow credit definition focusing on bank credit.

³ Moench et al. (2013) use the dynamic hierarchical factor model to explore the different blocks/sectors in a large set of economic data of one particular country, the United States. Kose et al. (2012) study the global interdependencies of output, consumption and investment of 100 countries categorizing the layers of commonalities into country-specific, region-specific (industrialized, emerging markets and developing economies) and a global factor. In this article, we combine the multi-country and the multi-sector structures using EA data.

⁴ Hence, panel VAR and factor models are likely to have different characteristics and span a different informational space. Whether lags or current values of the endogenous variables provide superior information for the states of a theoretical model is an open question.

⁵ See Tables A1 and A2 for an overview of all variables.

⁶ For a more comprehensive discussion, see Harding and Pagan (2006).

⁷ We also estimated a version where we can consider mixed-frequency, quarterly data for macroeconomic series and monthly for financial series. See Section IV for a discussion.

⁸ Times series are reported in the appendix (Figs A1 and A2).

⁹ The definition and the list of MFIs is available in the ECB statistical database warehouse (Table A3).

¹⁰ See the section ‘Robustness’ for a justification on the number of country-specific factors.

¹¹ Quoting from Committee findings release available at CEPR Webpage.