Choi and Zhao

Document Sample
Choi and Zhao Powered By Docstoc
					               Cross-autocorrelation in the New Zealand Stock Market

Daniel FS Choi and Xin Zhao, Department of Finance, University of Waikato, Private Bag
3105, Hamilton, New Zealand


Lo and MacKinlay (1990) find seminally the evidence of portfolio cross-autocorrelation in
NYSE-AMEX securities. This means that the returns of large-capitalization stocks always
lead those of smaller stocks. Chang et al. (1999) find evidence of cross-autocorrelation in all
six of the Asian markets: Hong Kong, Japan, Singapore, South Korea, Taiwan, and Thailand.
They further point out that the evidence is strongest in the larger, less restrictive markets of
the US, Hong Kong, and Japan. The New Zealand stock market is an open and deregulated
market which is shown to be globally integrated after 1984 (Chay and Eleswarapu 2001).
However, the New Zealand Stock Exchange (NZSE) is composed of firms with extreme
capitalization differentials (Courtenay and Cahan 2003). Bartholdy and Riding (1994) present
evidence that thin trading in the NZSE is extensive.

In this paper, we examine the evidence of cross-autocorrelation in the NZSE. Because the
distribution of firm size is skewed, we rank the capitalization of firms in descending order and
group the first fifteen percent as the large-size group and the rest eighty-five percent equally
into the medium-size, small-size, and very-small-size groups. Within each size group, we
split the firms into high and low volume firms. To minimize problems due to thin trading, we
delete the returns of a stock on the day not traded and on the following day.

Our results show that the returns of the medium-size portfolio lead the returns of the small-
size portfolio, and the returns of the small-size portfolio lead the returns of the very-small-size
portfolio. In the medium-size and small-size groups, the returns on high-volume portfolio
lead the returns on low-volume portfolio. We note that the returns of the large-size portfolio
do not lead the returns of the medium-size portfolio. This is expected. Over the sample
period, the mean capitalization of large firms ranges from ten to thirty times of that of the
medium firms. The market-wide information affecting large firms might not be relevant to
the medium firms. We conclude that the evidence of cross-autocorrelation exists in the New
Zealand market among medium-size and small-size firms. However, the very-small-size
group of firms have serious thin trading problem while the large-size firms and medium-size
firms are affected by different information sources.

Shared By: