An examination of the equity market price linkage between Australia and the European Union using leveraged bootstrap method

Abstract

The paper examines the equity market price interaction between Australia and the European Union – represented by the UK, Germany and France – based on the Toda–Yamamoto causality test, which is bootstrapped with leveraged adjustments. A new information criterion is used to choose the optimal lag order. Weekly MSCI data covering the period 1988 to 2001 is used, divided into two subperiods to allow for a structural break arising from the ERM crisis of 1992. Results show that, during the period before the ERM crisis, no significant causal links exist between Australia and any of three EU countries. During the period after the ERM crisis, Australia also had no causal links with Germany and France but it had with the UK, with causality running from the UK to Australia but not vice-versa. Thus, Australian investors may find the German and French, but not the UK, equity markets, attractive venues for their international diversification. German and French, but not British, investors may also obtain the same benefit from the Australian equity market.

Keywords:

• equity market integration
• leveraged bootstrap
• causality
• European Union

Notes

Australia was hardly affected at all by the Asian crisis. Since 2000, Australia has been the fastest growing economy among the OECD countries (see IMF World Economic Outlook, 2002).

See, Roca (2000).

See Samant (1999).

See Shapiro (1999), pp. 104–7 for a concise treatment of the ERM crisis.

The existence of structural break was confirmed by tests for parameter stability.

Elton and Gruber (1995) show that even if risk-adjusted returns in the foreign market is lower than that of the domestic market, it still pays to diversify provided the correlation is low enough. The risk-adjusted returns in the foreign market need only to be above a certain proportion of the domestic risk-adjusted return where the proportion is the correlation coefficient. This is more formally stated as follows:

or

where:

By causality, we mean causality in the Granger sense. That is, we are interested to find out whether one variable precedes another variable or not.

For more details on leverage adjustment, see Davison, and Hinkley (1999) and Hacker and Hatemi-J (2002). The latter authors introduce this adjustment for multivariate equation cases.

A program procedure written in Gauss to conduct leveraged bootstrap simulations is available on request from the authors.