An investigation into the extent of beta instability in the Singapore stock market

Abstract

This paper explores the issue of beta instability in the Singaporean stock market over the period 1986 to 1993. Analysis of the full eight-year interval reveals a very high incidence of beta instability, namely, at about 40% of the individual stocks tested. However, this sample straddles a listing change between the Singapore and Kuala Lumpur markets which occurred at the beginning of 1990. Accordingly, to assess what impact this listing change may have on our results, we examine the four-year sub-periods either side of the listing change, as well as the four-year sub-period which straddles the listing change. A comparison of our results across these different smaller overlapping subperiods indicates an incidence of beta instability for individual stocks at about 20% or less – a somewhat lower level than that observed for the eight-year sample. Hence, while this analysis suggests an absence of a listing change effect on beta stability, a sample length effect may exist. Our results are largely insensitive to at least three different factors. First the results are insensitive to whether betas are estimated by OLS or the Dimson (1979)technique. Second the results are insensitive to whether the Straits Times Index or the Morgan Stanley Country Index for Singapore is used to measure the market returns. Finally the results are insensitive to survivorship bias.