Page 113 - Contributed Paper Session (CPS) - Volume 6

CPS1837 Qiguang Dong et al.
2
MSE HMSE MAE HMAE QLIKE R LOG

M35 0 0 0 0 0 0 0 0 0 0 0 0
M36 0 0 0 0 0 0 0 0 0 0 0 0
(2) Compared the models based on different distribution, we can find
that, the amount of survived models based on distributions ged and sged are
more than the survived models based on the rest distributions. This results
suggest that the symmetric and skewed generalized error distributions mostly
approximate the actual distribution of the volatility series than normal and
student-t distributions.
(3) In all 36 models, the models which survived the most loss functions is
M24, namely LnRVsGARCH-sged. It survived the first 5 loss functions, and the
following model is M33 (LnRV-gjrGARCHged). For the sake of robustness, this
paper utilizes the short term ARMA model instead of ARFIMA in model
specification to construct another 36 forecasting models for robust test. The
MCS test results are presented in table4. The results are consistent with the
results shown in table 3, and the model M24 is still outperformed.

4. Discussion and Conclusion
This paper utilizes the realized volatility and logarithm-transformed
realized volatility to forecast the actual volatility of CSI300 stock index. We
construct 36 long memory ARFIMA models for forecasting, and then applying
the out-of-sample rolling time window forecasting combined with Model
Confidence Set test to evaluate and compare the predictive ability of the
models specified. For the sake of robustness, we conduct the same procedure
to 36 short memory ARMA models and the empirical results are similar. The
empirical results show that: (1) Both RV and LnRV series have a long memory
due to both Hurst indexes are greater than 0.5 and smaller than 1. (2)The
symmetric and skewed generalized error distributions ged and sged are
employed more accuracy than normal and student-t distributions. (3) The
model LnRV-sGARCH-sged is outperformed than the rest in the long memory
model as well as in the short memory model.

References
1. Andersen, T. G., & Bollerslev, T. (1998). Answering the skeptics: Yes,
standard volatility models do provide accurate forecasts. International
economic review, 885-905.
2. Andersen, T. G., Bollerslev, T., Diebold, F. X., & Labys, P. (2003). Modeling
and forecasting realized volatility. Econometrica, 71(2), 579-625.

102 | I S I W S C 2 0 1 9