5.2. Correlation analysis
Accurate correlation assessment is one fundamental aspect in
portfolio theory. According to Corbet et Al. (2018), such important metric has
momentous implications on portfolio construction, diversification and
hedging.
15
Figure 2 illustrates pairwise correlation coefficients, which
give a first snapshot of average correlations between our different financial
assets. It is noteworthy that almost all correlation coefficients between
cryptocurrencies and traditional assets do not exceed 0.10. Regarding
cryptocurrencies, they range from 0.25 to 0.59 with Bitcoin and Litecoin
exhibiting the highest correlation. Relatively, traditional assets have more
varying correlations within themselves due to their global diversification.
Yet, this is only the average correlation for our sample period. That is why I
derive the multi-varying correlation through the Dynamic conditional
correlation model.
Tables 4, 5, 6 and 7 illustrate descriptive statistics of
dynamic conditional correlations between the innovations of each of the
cryptocurrencies and traditional assets. The multi-varying correlation analysis
will allow us to asses precisely the diversification and hedging benefits of
cryptocurrencies.
Table 4 depicts the DCC statistics for correlation pairs
within Bitcoin. The latter displays negative correlation among all the sample
period with the following asset categories: developed corporate bonds, global
real estate and Chinese equities. Hence, it acts as a strong hedge according to
the definition of Baur and Lucey (2010). Moreover, it has a correlation of
approximately zero with gold, MSCI emerging markets and sovereign bonds. It is
also notable that the standard deviation of those correlation pairs is very low
suggesting a stable correlation over time and high diversification benefits.
Regarding developed market equities, Bitcoin is negatively correlated to Nikkei
225 on average with a maximum value of 0.0194.
The highest average correlation from all conventional assets
is the pair with Eurostoxx 50 with a value of 0.0649. Nevertheless, it is very
stable among all the period. For S&P 500, DCC correlation is more dynamic
with a maximum spike of 0.27.
16
Table 5 reveals that Ripple cannot be regarded as a strong
hedge against any of the traditional assets. A further look into the
descriptive statistics of DCC correlations brings to light the noisy
correlation spikes. Moreover, the 25% and 75% quantiles show slightly positive
correlations with traditional assets over the whole sample period.
Table 6 shows that Dash has the highest correlation with
developed market equities within all the cryptocurrencies. Furthermore, dynamic
correlation is unstable for emerging markets equities and alternative
investments since it swings from negative to positive. Yet, it can be
considered as a strong hedge against developed corporate and sovereign bonds
only.
According to table 7, Litecoin possesses hedging capacities
against Japanese equities and global real estate. In addition, Low co-movements
with equity market indices are more persistent than for Ripple and Dash which
suggests better investment opportunities.
It is apparent that S&P 500 shows unstable dynamic
correlations with cryptocurrencies all over the sample period. For a better
assessment of hedging capacities, figure 2 plots its DCC correlation with
cryptocurrencies as well as gold. The latter is added as a reference point
since it is depicted as a safe haven against S&P 500. (Baur & Mc
Dermott 2016).
Ripple and Dash show mostly positive correlations as already
stated. Meanwhile, Bitcoin and Litecoin exhibit a wide range of positive and
negative correlation values. Albeit, with small periods and no persistence
while being negative. Gold, on the contrary, shows negative dynamic correlation
for successive several months. In this regard, safe haven and strong hedge
attributes should be excluded for Bitcoin and Litecoin. They can only be very
effective diversifiers against S&P 500.
| 
