Sparse vector heterogeneous autoregressive modeling for realized volatility
- PDF / 2,079,519 Bytes
- 16 Pages / 439.37 x 666.142 pts Page_size
- 38 Downloads / 186 Views
Online ISSN 2005-2863 Print ISSN 1226-3192
RESEARCH ARTICLE
Sparse vector heterogeneous autoregressive modeling for realized volatility Changryong Baek1 · Minsu Park1 Received: 10 June 2020 / Accepted: 23 September 2020 © Korean Statistical Society 2020
Abstract We propose a sparse vector heterogeneous autoregressive (VHAR) model for realized volatility forecasting. As a multivariate extension of a heterogeneous autoregressive model, a VHAR model can consider the dynamics of multinational stock volatilities in a compact manner. A sparse VHAR is estimated using adaptive lasso and some theoretical properties are provided. In practice, our sparse VHAR model can improve forecasting performance and explicitly show the connectivity between stock markets. In particular, our empirical analysis shows that the NASDAQ market had the strongest influence on stock market volatility worldwide in the 2010s. Keywords Sparse vector heterogeneous autoregressive model · Realized volatility · Heterogeneous autoregressive (HAR) model · Stock market linkage
1 Introduction Multivariate volatility is becoming increasingly important owing to its key roles in understanding the co-movements between stock markets, portfolio management, and risk management. For example, because of financial integration, the volatility in one market reacts to innovations in other markets. Thus, developing joint models that encapsulate the dynamics of multinational realized volatility (RV) linkages and provide reliable predictions is inevitable. We particularly focus on the RV as a proxy for underlying asset price volatility. This approach follows the groundbreaking study This work was partly supported by the National Research Foundation of Korea (NRF2019R1F1A1057104). * Changryong Baek [email protected] Minsu Park [email protected] 1
Department of Statistics, Sungkyunkwan University, 25‑2, Sungkyunkwan‑ro, Seoul 03063, Korea
13
Vol.:(0123456789)
Journal of the Korean Statistical Society
of Andersen et al. (2003), who show that the RV, defined as a low-frequency (typically 5 min) sum of intraday squared returns, provides the best approximation of real volatility. A prominent feature of the RV is its strong persistence, which can be easily confused with changes in the mean (Baek and Pipiras 2014; Lee et al. 2015). Rather than differentiating the two models, Corsi (2009) proposed a heterogeneous autoregressive (HAR) model, which gained popularity owing to its simple but remarkable forecasting performance. Later, researchers extended the HAR model by including jumps, leverage effects, GARCH-type errors, and structural breaks (see Song and Baek (2019) and the references therein). This study also extends the HAR model by incorporating volatility linkages between stock markets. This approach is partly motivated by the factor-augmented HAR model of Kim and Baek (2019), who show that RV forecasting can be improved by adding foreign stock market factors to the vanilla HAR model. In this study, we consider the vector heterogeneous autoregressive (VHAR) mode
Data Loading...
