Pattern and determinants of tail-risk transmission between cryptocurrency markets: new evidence from recent crisis episodes

Financial Innovation

Table 7 Wald test statistics for time-varying Granger causality

	Wald statistics- recursive evolving-heteroskedasticity
Panel A: tail risk (CAViaR) at 1%
H0: TCI (1%) is Granger causality FGCI	32.871***
	− 7.023
	[10.047]
H0: TCI (1%) is Granger causality ADS	49.018***
	− 8.026
	[10.046]
H0: TCI (1%) is Granger causality GPR	21.921***
	− 6.39
	[9.543]
H0: TCI (1%) is Granger causality EPU	28.910***
	− 7.149
	[9.940]
H0: TCI (1%) is Granger causality TEU	32.791***
	− 7.098
	[9.114]
Panel B: tail risk (CAViaR) at 5%
H0: TCI (5%) is Granger causality FGCI	26.924***
	− 6.752
	[8.731]
H0: TCI (%%) is Granger causality ADS	48.819***
	− 7.776
	[10.090]
H0: TCI (5%) is Granger causality GPR	27.582***
	− 7.389
	[9.278]
H0: TCI (5%) is Granger causality EPU	30.200***
	− 6.699
	[8.663]
H0: TCI (5%) is Granger causality TEU	33.201***
	− 6.318
	[8.619]

Wald test statistics computed using recursive evolving-heteroskedasticity algorithms. We follow Shi et al. (2020) and set the minimum window size to 72 observations. The empirical distribution of the bootstrap test statistics at the 95th and 99th percentile are shown in parentheses and brackets, respectively. *, ** and *** indicate significance at 10%, 5% and 1% levels respectively