Economic Policy Uncertainty and Chinese Stock Market Volatility: A CARR-MIDAS Approach

Table 5

Out-of-sample forecast evaluation results for forecast window of 1000.

Horizon	CARR	CARR-MIDAS	CARR-MIDAS-CEPU	CARR-MIDAS-GEPU
Panel A: MSE loss function
1	6.3291e − 06	5.4761e − 06	3.8849e −  06	4.5401e − 06
5	8.6110e − 06	7.2356e − 06	4.7254e −  06	5.6592e − 06
10	1.0671e − 05	8.5469e − 06	5.1621e −  06	6.2875e − 06
22	1.4443e − 05	9.6989e − 06	5.2976e −  06	6.4651e − 06
Panel B: QLIKE loss function
1	5.6089e − 02	5.0384e − 02	3.7588e −  02	4.2935e − 02
5	7.3891e − 02	6.5760e − 02	4.6985e −  02	5.4085e − 02
10	8.9510e − 02	7.6431e − 02	5.2300e −  02	6.0477e − 02
22	1.1905e − 01	8.9079e − 02	5.8530e −  02	6.6474e − 02

MSE is the mean squared error, and QLIKE is the quasilikelihood. Bold entries indicate the model with the lowest loss value per horizon (in each row). Shaded entries indicate the model is included in the MCS at a 10% significance level.