with a t-ratio of 1.3), it is evident that these figures are much smaller
than before and are no longer statistically significant.We can therefore
conclude that the reversal (i.e. the subsequent price rise) is only partial
and over the whole event window the return is around−5%. Similarly,
the long-run post-event performance is now a negative −3.3%, although
it is not significant, where it was +40% previously. Given that
these outlier-removed results are comparable for additions but much
more sensible for deletions, we proceed to employ these rather than
the unadjusted data in all subsequent return-based analysis.
Studies involving the S&P500 have argued that since the announcement
regarding changes in the composition of the index aremadewhen
the market is closed, most of the announcement effect can never be
realised by traders since the price trades at a higher (lower) level immediately
at the open the day after the announcement for additions (deletions).
It is thus argued in this literature that it is important to examine
the open-to-close return on the event date rather than the typical daily
returnswhich are usually measured on a close-to-close basis. However,
for the sample of firms thatwe consider, it is clear fromthe returns that
investors know about index recompositions and trade on this information
at least a few days prior to themtaking place. As such, there is only
a minimal return on the event day and therefore no need to be
concerned with this issue.