presents the results for the sample of additions, pooled
across countries and indices, over the event window from day t − 5
to day t + 15. The final five rows display the cumulative returns for
the pre- and post-event windows, and also for a long-run window
comprising 180 post-event trading days. The returns are presented
as daily percentages along with the corresponding t-ratios. We
present the raw returns and also the abnormal returns calculated
using the single index model approach described above. We employ
the EPRA Developed Market and S&P500 indices as benchmarks in
columns 4–5 and 6–7 respectively. The final column presents the
number of firms in the sample, N, on each day during the event window.
It is clear that the figure is near the maximum available of 366
for most days, although a handful are lost before the event — due,
for example, to firms making their debut on the market and then
being listed in an EPRA index the same day or a day or two later.
The most salient feature is one of positive and highly statistically
significant positive returns until the day before index inclusion but
then an almost zero return on the event day. We therefore conclude
that market participants are aware that a company will enter the
index and exert net buying pressure on the days before index entry.
Focusing on the raw returns, these are positive on four of the five
final pre-event trading days, and are largest (1.34%) and most significant
(t-ratio = 7.66) on the day just before the stock enters the
index, cumulating in an average pre-event week's return of 2.4%.
The event day return is not statistically significant but there is a larger
(although still not significant) average return of 0.75% during the
three trading weeks following index inclusion. Over the longer-run
window, the average return is highly significant and positive at
around 13%.
It is clear that the use of abnormal or raw returns makes no qualitative
difference to the results, but subtracting the market-required
rates of return reduces both the short-run and long-run returns
compared to their unadjusted values, perhaps indicating that index
recompositions tend to occur around times of buoyancy in the overall
market