Research on RMB Exchange Rate Volatility Risk Based on MSGARCH-VaR Model

Table 4

Accuracy of VaR predictions. The table reports the actual over expected exceedance ratio (AE) and the statistic and probabilities of the conditional coverage (CC) test and the dynamic quantile (DQ) test for the one-day ahead 1% VaR.

	CC test			DQ test			AE
	RUSD	RJPY	RHKD	RUSD	RJPY	RHKD	RUSD	RJPY	RHKD
Single regime
GARCH-t	2.943 (0.230)	4.817 (0.900)	0.336 (0.845)	13.597 (0.059)	4.033 (0.776)	3.122 (0.873)	1.8	0.2	1.2
GJR-GARCH-t	0.918 (0.632)	4.817 (0.090)	0.336 (0.845)	8.898 0.260)	3.914 (0.790)	9.099 (0.246)	1.4	0.2	1.2
Two regimes
GARCH-t	1.799 (0.407)	4.817 (0.900)	0.101 (0.951)	10.043 (0.186)	3.485 (0.837)	0.788 (0.998)	1.6	0.2	1
GJR-GARCH-t	0.918 (0.632)	2.369 (0.306)	0.336 (0.845)	3.347 (0.851)	2.033 (0.958)	0.699 (0.998)	1.4	0.4	1.2

Note. Numbers in brackets are values and in bold are highlighted the Markov two-regime GARCH models that outperform their single-regime counterparts.