i (↓), j ( →) | Bitcoin | Ethereum | Binance | Cardano | Litecoin | Ripple | FROM others |
|---|
Bitcoin | 45.83 | 16.1 | 9.77 | 6.94 | 13.1 | 8.27 | 54.17 |
Ethereum | 14.83 | 43.39 | 10.83 | 9.04 | 13.47 | 8.43 | 56.61 |
Binance | 9.89 | 9.73 | 40.41 | 11.84 | 15.91 | 12.21 | 59.59 |
Cardano | 6.26 | 8.29 | 12.29 | 48.83 | 15.06 | 9.28 | 51.17 |
Litecoin | 11.7 | 11.33 | 14.65 | 12.89 | 36.41 | 13.03 | 63.59 |
Ripple | 7.91 | 7.59 | 12.78 | 10.6 | 14.88 | 46.24 | 53.76 |
TO others | 50.59 | 53.04 | 60.32 | 51.31 | 72.42 | 51.22 | 338.9 |
Net directional | − 3.57 | − 3.58 | 0.73 | 0.13 | 8.83 | − 2.55 | TCI = 56.48 |
- The spillover indices are calculated following the methodology of Antonakakis et al (2020). TCI in the lower right corner is the total connectedness index calculated by Eq. 15. The “TO others”, is the shock transmission from one variable to all other variables, and “FROM others”, is the shocks received from other variables by one variable, and “Net directional”, which is the difference between TO others and FROM others, are calculated by Eqs. 16, 17 and 18, respectively. To estimate TVP-VAR parameters, we follow Kumar et al (2022) and rely on Akaike Information Criterion (AIC) in choosing lags where GFEVD is built on a 100-days ahead forecast. Further, Bayes prior size was set to 200 days, and following Antonakakis et al (2020), forgetting factors used in the Kelman Filter are 0.99 for VAR forgetting factor and 0.96 for the Exponentially Weighted Moving Average (EWMA) forgetting factor. The results are in percentages
