Regime-dependent relationships among the stock markets of the US, Australia and New Zealand: a Markov-switching VAR approach

Abstract

Adopting a multivariate Markov-switching-VAR model (Krolzig, 1997) and a recently developed regime-dependent impulse response analysis technique (Ehrmann et al., 2003), this article investigates the dynamic relationships among the stock markets of the US, Australia and New Zealand. Our results reveal the existence of two different regimes in the three stock markets. We find that the correlations among the three markets are significantly higher in a bear regime than in a bull regime. In addition, the responses of each of the three markets to shocks in the other two markets are stronger and more persistent in the bear regime than in the bull regime. Finally, our findings imply that for the New Zealand stock market, the Australian stock market is more influential than the US stock market, and for the Australian stock market, the US stock market is more influential than the New Zealand stock market.

Keywords:

• Markov switching VAR
• regime-dependent impulse response
• Hansen test
• stock markets
• dynamic relationships

JEL Classification:

• F36
• G15
• G11

Acknowledgements

This research is partially supported by University of Macau, California State University, Hong Kong Baptist University, and the Research Grants Council of Hong Kong. The third author would like to thank Robert B. Miller and Howard E. Thompson for their continuous guidance and encouragement.

Notes

¹ There are numerous studies that investigate the dynamic relationships between stock markets in different regions, such as the relationships between the US stock market and other countries' stock markets; see, for example, Campbell and Hamao (1992), Bekaert and Harvey (1995), Bracker et al. (1999), Hamori and Imamura (2000), Masih and Masih (2001), Fraser and Oyefeso (2005), Wong et al. (2006), Qiao et al. (2008), and You and Daigler (2010). However, papers studying stock markets in the Australasian region are limited. In this region, the economies of Australia and New Zealand are most influential. We thus choose to study the relationships between the US stock market (the world's largest) and the stock markets of Australia and New Zealand.

² Underpinning the close investment and trade ties between Australia and New Zealand is the Australia-New Zealand Closer Economic Relations Trade Agreement (ANZCERTA), which came into effect on 1 January 1983. ANZCERTA also provides for removal of impediments to investment flows, such as liberalizing government purchasing orders and free movement of labour, although there is no provision for the free movement of capital. ANZCERTA provides companies of both countries with greater access to the markets of the other. Thus, companies in both countries consider the other as the natural avenue for overseas expansion or, alternatively, seek a cross-exchange listing to gain access to a larger pool of capital and consequently realize a lower cost of capital.

³ Janakiramanan and Lamba (1998) argue that when two countries share geographic proximity, these markets are more than likely to interact with each other.

⁴ Readers may also refer to Lo and MacKinlay (1988) for the details of this approach, including how to deal with the situation in which Wednesday is a holiday.

⁵ One of these conditions is that all parameters must be identified under the null hypothesis of a single regime in order to keep the information matrix nonsingular. However, under the null of a single regime, parameters related to the second regime cannot be identified. Thus, the classical likelihood surface is flat with respect to these unidentified ‘nuisance’ parameters.

⁶ The first grid setting is: p_u (p ₂₂) = 0.70, 0.75, 0.80, 0.85, 0.90; μ₂ = −0.02, 0.00, 0.02, 0.04, 0.06, 0.08, 0.10, 0.12, 0.14, 0.16; σ₂ = 2.0, 2.1, 2.2, 2.3, 2.4, 2.5, 2.6, 2.7, 2.8, 2.9. Grid 1 has 2500 points. The second grid setting is: p_n (p ₂₂) = 0.4, 0.5, 0.6, 0.7, 0.8, 0.9; μ₂ = −0.3, −0.2, −0.1, 0, 0.1, 0.2, 0.3, 0.4, 0.5, 0.6; σ₂ = 1.50, 1.65, 1.80, 1.95, 2.10, 2.25, 2.40, 2.55, 2.70, 2.85. Grid 2 has 3600 points. We also experimented with other grid settings, and the conclusion drawn from these results are qualitatively the same. For details of the Hansen test, please refer to Hansen (1992, 1996) and Fong and Kim (2001).

⁷ Please note that the estimation of the MS-VAR model is extremely computationally intensive. For a 3-variable, 2-regime MS-VAR (2) model, 56 parameters need to be estimated. If the p increases to 10, 200 parameters need to be estimated.

⁸ In the related literature on regime-switching models, such as Krolzig (1997), when the estimated mean is negative, we take it as a sign of a bear regime, and when the estimated mean is positive, we take it as a sign of a bull regime.

⁹ Panel E of Table 3 also reports the AIC for a linear VAR and an alternative MS-VAR. It is obvious that the AIC favours the MS-VAR model over the linear VAR model. In addition, the above Hansen test result and estimation results of MS-VAR indicate that there exist two regimes. All these provide evidence that the MS-VAR is more appropriate than the linear VAR for analysing the dynamic relationships among the Australia, New Zealand and the US markets.