332 LARGE AND SMALL CAP STOCKS IN EUROPEThe use of market indices, instead of portfolios, provides two advantages
to the practitioners of the market. Firstly, they can take signals directly from
market indices quotations, therefore, it is not necessary to build portfolios
and, secondly, the cost of implementing any potential trading rules can be
reduced due to the existence of derivative contracts on the large stock index
(Pardo and Torró, 2005).
Figure 17.1 displays the weekly evolution of the indices in the studied
period and preliminary data analysis is presented in Tables 17.1 and 17.2.
Table 17.1 displays a summary of the principal statistics for the returns.
It can be stated that all indices offer very similar statistics. All series present
significant skewness except the CAC40 and the FTSE. Moreover, all present
significant kurtosis and the Jarque-Bera statistic indicates that the hypoth-
esis of normality is rejected for all indices. On the other hand, all series
present significant autocorrelation and heteroskedasticity. Finally, although
equality in means between large and small indices of each country cannot
be rejected, the variances equality test is rejected.
Panel (A) in Table 17.2 displays returns, volatilities and correlation coeffi-
cients, year by year through the sample period for the French indices. Three
facts can be highlighted from this table. First, there are five years (1991,
1992, 1995, 2000 and 2003) in which both indices offer a different sign return
but the means equality hypothesis cannot be rejected. Secondly, for every
year, except 2000, the CAC40 volatility is larger than the MIDCAC volatil-
ity. Another appealing fact is that except in two years (1998 and 2000), the
hypothesis of variance equality is rejected. Finally, the correlation between
both indices has dropped over time, becoming very small.
Panel (B) in Table 17.2 displays returns, volatilities and correlation coeffi-
cients, year by year through the sample period for the German indices. Three
facts can be highlighted from this table. First, there are six years (1991, 1992,
1995, 1998, 2000 and 2004) in which both indices offer a different sign return
but the means equality hypothesis cannot be rejected. Secondly, for every
year, except 1996, the DAX is more volatile than the SDAX. Thirdly, except
in three years (1993, 1996 and 2004), the hypothesis of variance equality is
rejected. Finally, the correlation between both indices has decreased over
time, becoming very small.
Panel (C) in Table 17.2 displays returns, volatilities and correlation coef-
ficients, year by year through the sample period for the British indices. Four
facts can be highlighted from this table. First, there are three years (1998, 2000
and 2004) in which both indices offer a different sign but the means equal-
ity hypothesis cannot be rejected. Secondly, for every year, except 2000 and
2001, the FTSE volatility is larger than the SMALL CAP volatility. Thirdly, at
the beginning of the sample the hypothesis of variance equality is rejected
but, at the end of the sample the hypothesis of variance equality cannot be
rejected. Finally, thecorrelationbetweenbothindiceshasdroppedovertime.
Three facts must be highlighted from the previous analysis. Firstly, from
Tables 17.1 and 17.2 it can be accepted that there exist significant differences