Global output growth and volatility spillovers

Abstract

This article examines discernable patterns of real Gross Domestic Product (GDP) growth co-movements across 29 countries, using consistent time series data (1912–2008). Of these countries, only 12 are found to form three statistically significant groupings (i.e. G6-six Organization for Economic Co-operation and Development (OECD) European countries, G4-four Anglo-Saxon countries, and G2-two major Asian countries). One may then conclude that, inter alia, geographical proximity, cultural ties, and the level of socio-economic and financial ties among countries determine the global systematic co-movements of growth rates. Our results indicate that any recession in the US initially engulfs other Anglo-Saxon countries as well as G6 and G2 countries, before exerting its adverse knock-on effects to the rest of the world. A Multivariate Generalized Autoregressive Conditional Heteroscedasticity (MGARCH) model is also used to examine the transmission of GDP growth across these three groups and their corresponding volatility spillovers. We find significant bi-directional cross-mean spillovers between the G4 and G6 blocs. In terms of cross-volatility spillovers, the estimated persistence varies from a maximum 0.959 (G4–G6) to a minimum of 0.832 (G2–G4).

Keywords:

• GDP growth
• multivariate GARCH model
• volatility spillovers
• output co-movements

JEL Classification:

• C32
• E23
• E32

Acknowledgements

We wish to acknowledge Professor David Peel and the anonymous referee(s) whose useful inputs and comments improved an earlier version of this article. The usual caveat applies.

Notes

¹ The acronym, BEKK comes from the synthesis work done by Baba, Engle, Kraft and Kroner for the earlier version of Engle and Kroner's (1995) paper.

² Dincer (2007) used time series date (1982–1997) for 79 countries and found that two main factors influencing their economic performance are similarities of their degree of property rights protection and saving rates.

³ We also tested various DBEKK(p, q) specifications using AIC, SIC and HIC, where p = 1, 2 and 3 and q = 1, 2, and the results indicate that the DBEKK(1, 1) specification consistently has the lowest AIC (−6.67), SIC (−6.12) and HIC (−6.45) with a log-likelihood of 330.07.

⁴ Sweden has retained its own currency and remains outside the Eurozone.

⁵ This contrasts with the other two groupings where geographical proximity is likely to be more important. This is likely to be only significant for the US and Canada for the G4 group.

⁶ Particularly since 2003, with the commencement of negotiations on the establishment of a Japan–Korea free trade agreement.

⁷ The rapid growth of production networks in East Asia, including Japan and Korea, has brought about further integration of these two leading Asian economies (Ando and Kimura, 2005).