Does Decomposing Realized Volatility Help in Risk Prediction: Evidence from Chinese Mainland Stocks

31 Pages Posted: 14 Feb 2011  

Yin Liao

Australian National University (ANU); Financial Research Network (FIRN)

Date Written: January 13, 2011

Abstract

This article studies the risk forecasting properties of three realized volatility models for three Chinese individual stocks, and reveals the important role that jumps can play in risk prediction. I firstly investigate dynamic pattern of jumps in three Chinese stocks, and find that relative to developed markets, jumps in this emerging market are more predictable and account for a larger proportion in realized volatility. Further, I compare the Value-at-risk (VaR) forecasting performances of three commonly used realized volatility models for the three Chinese stocks. Two-step VaR backtesting shows that a newly proposed realized volatility forecasting model (Andersen et al, 2007), which separately treats jumps and the continuous sample path of the asset price, provides more accurate VaR prediction than two other competing models, that treat realized volatility as a single variable. These findings suggest that carefully modelling of jumps is important in risk prediction, especially for emerging markets where jumps play a stronger role than those in developed markets.

Keywords: Value-at-Risk, Realized volatility, Bi-power variation, Jumps, HAR-CJN model

JEL Classification: C13, C32, C52, C53, G17, G32

Suggested Citation

Liao, Yin, Does Decomposing Realized Volatility Help in Risk Prediction: Evidence from Chinese Mainland Stocks (January 13, 2011). Available at SSRN: https://ssrn.com/abstract=1739644 or http://dx.doi.org/10.2139/ssrn.1739644

Yin Liao (Contact Author)

Australian National University (ANU) ( email )

Canberra, Australian Capital Territory 2601
Australia

Financial Research Network (FIRN)

C/- University of Queensland Business School
St Lucia, 4071 Brisbane
Queensland
Australia

HOME PAGE: http://www.firn.org.au

Paper statistics

Downloads
57
Rank
305,276
Abstract Views
500