THADAVILLIL JITHENDRANATHAN 275Table 14.3OLS regression output forex postreturns against the DCC
dummy
Period αβAdj. R^2 Obs.
(t-stat) (t-stat) (F-stat)
[All portfolios]
One month −0.0044 0.0077 0.0024 1340
(1.6540)* (2.0513) (4.2080)**
Three months −0.0032 0.0076 0.0007 1340
(0.8353) (1.4128) (1.9961)
Six months 0.0005 0.0072 0.0001 1340
(0.0962) (1.0264) (1.0536)
[Low-risk portfolios]
One month 0.0001 0.0045 0.0001 610
(0.0197) (0.9282) (0.8616)
Three months 0.0057 −0.0018 0.0001 610
(1.1507) (0.2605) (0.0679)
Six months 0.0157 −0.0044 0.0001 610
(0.0962) (1.0264) (1.0536)
[High risk portfolios]
One month −0.0081 0.0104 0.0034 732
(2.0543) (1.8604)** (3.4612)
Three months −0.0106 0.01553 0.0037 732
(1.8682) (1.9365) (3.7498)***
Six months −0.0122 0.01696 0.0021 732
(1.6261)*** (1.6320)*** (3.3672)***∗Significant at 1%;∗∗Significant at 5%;∗∗∗Significant at 10%.
portfolios are divided into two groups based on the standard deviations
of the efficient portfolios. For each month, the sample is divided into a set
of low-risk portfolios comprising of five of the lowest variance portfolios,
and another set of high-risk portfolios comprising of six portfolios with the
highest risk.
From the results it is clear that except for two out of the nine cases, the
ex postreturns of efficient portfolios created using the DCC model are higher
than those of the portfolios created using the rolling model. This point is
further proved in the regression analysis that follows.
The results of the regressions using equation (14.15) are given in
Table 14.3. Pooledex postreturns of efficient portfolios are regressed against
the dummy variable that has a value of one for those portfolios for which
the inputs were computed using the DCC model. Three sets of regressions