The effect of earnings/dividend announcements on open interest in exchange-traded call options : Australian evidence

2017-11-10T00:56:21Z (GMT) by Adrienne H. Whelan
Research in the United States has found that there is a significant decline. In option open interest prior to earnings announcements. The research reported herein uses Australian data to replicate the United States research. There are, however, some differences necessitated by data constraints. The major differences arise from (i) a shortage of suitable announcements, (ii) the Australian practice of announcing earnings and dividends simultaneously, (iii) the Australian practice of making semi-annual rather than quarterly announcements, (iv) the smaller set of available options and (v) thinner trading in the options market.Results indicate that although there are some similarities between Australia and the United States in the behaviour of open interest around earnings announcements, there are also some differences. The direction and magnitude of deviations from normal are similar for the time period from 9 days prior to the earnings announcement to 8 days after the announcement. Although the deviations from normal prior to this period are negative in both the Australian and United States studies, the magnitude of the deviations observed in Australia is considerably larger. The negative abnormal open interest behaviour observed after this time period in the United States does not occur in Australia. This research is predicated on the basis that there may be a difference in behaviour between the populations of announcement and non-announcement period open interest. Results of parametric tests of the Australian data used in this study implied that the population means were significantly different. However, statistical tests revealed that the data are not Normally distributed and therefore it was considered inappropriate to use only parametric procedures. Results of non-parametric tests implied that the population distributions were not significantly different. As financial markets in Australia and the United States have similar structures, these findings may also be true of the United states. Although the Australian and United States studies obtained broadly similar results, statistical tests of significance in the United States research relied on procedures that are not appropriate when the assumption of Normality does not hold.Results of the Australian research indicate that "abnormal" open interest is lower prior to an announcement than after an announcement. Given the cyclical nature of open interest and the use of a measure of "normal" which is constant over time, an upward trend in "abnormal" open interest is not surprising. The measure of "normal" open interest used in the United States study is inappropriate for data which is cyclical in nature. It is, therefore, inappropriate for Australian data and may also be inappropriate for United States data. This contention is supported by results which indicate that, in the United States, abnormal open interest may not return to zero in non-announcement periods when the length of the announcement period is extended. In the light of the results reported in this thesis, there is a need to re-examine the United States research results.