HELENA CHULIÁ AND HIPÒLIT TORRÓ 337Table 17.2Continued
Panel C: returns, volatilities and correlations of British indices seriesYear Annualized Returns (%) Annualized Volatilities (%) Correlation(3)
FTSE SMALLCAP Mean FTSE SMALLCAP Levene(2)
Test(1) Test1993 20.65 22.06 0.09 11.98 6.81 10.08∗ 0.63
1994 −11.18 −6.19 0.27 15.29 9.35 15.69∗ 0.55
1995 17.19 10.27 0.58 10.51 5.26 20.87∗ 0.60
1996 7.45 7.61 0.25 9.01 6.71 4.79∗ −0.07
1997 21.84 6.47 0.93 14.88 7.33 23.26∗ 0.02
1998 13.59 −11.07 0.94 20.23 16.38 4.49∗ 0.11
1999 11.04 39.56 1.35 18.44 10.25 18.92∗ −0.03
2000 −7.50 2.62 0.25 13.86 14.24 0.002 0.02
2001 −16.72 −21.00 0.15 18.67 20.95 0.07 −0.13
2002 −33.20 −32.59 0.02 28.01 15.79 7.34∗ 0.19
2003 17.47 34.67 0.74 18.90 13.59 0.08 0.53
2004 (4) 0.65 −11.23 0.64 10.76 8.24 0.99 0.56
Notes: Data frequency is weekly.
(1) This column displays the means equality test. Significant coefficients at 95% of confidence level
are highlighted with one asterisk (∗). (2) This column displays the variances equality test known as
Levene test. Significant coefficients at 95% of confidence level are highlighted with one asterisk (∗).
(3) This column displays the annual correlation between both indices. (4) This row displays the results
for the period 7 January to 25 August 2004.
and significant for all countries. Therefore, in these markets the risk is val-
ued. This is a surprising result because most studies find a positive but
non-significant relationship between expected return and risk (French,
Schwert and Stambaugh, 1997; Campbell and Hentschel, 1992) or a neg-
ative and significant relationship (Campbell, 1987; Officer, 1973; Glosten,
Jagannatham and Runkle, 1993). Moreover, this result is consistent with the
volatility feedback hypothesis. If volatility is priced, an anticipated increase
in volatility raises the required return on equity, and therefore there will be
observed time-varying risk premiums. Secondly, coefficientsg 11 andg 22 are
significant, showing that in both indices (large cap and small cap), nega-
tive asymmetric firm volatility is important for their own dynamic, with the
exception of the SDAX index. For this index, the estimated coefficientg 22 is
not significant. Thirdly, coefficientsg 12 andg 21 are both significant, showing
that cross-relationships between negative shocks in both markets are also
significant.
The encompassing model restrictions on the ADC were rejected. This
result means that the ADC cannot be reduced to any nested model. The esti-
mated values ofφ 12 (close but significantly different to 1) andρ 12 (close but