Forecasting co-volatilities via factor models with asymmetry and long memory in realized covariance

Abstract

Modelling covariance structures is known to suffer from the curse of dimensionality. In order to avoid this problem for forecasting, the authors propose a new factor multivariate stochastic volatility (fMSV) model for realized covariance measures that accommodates asymmetry and long memory. Using the basic structure of the fMSV model, the authors extend the dynamic correlation MSV model, the conditional/stochastic Wishart autoregressive models, the matrix-exponential MSV model, and the Cholesky MSV model. Empirical results for 7 financial asset returns for US stock returns indicate that the new fMSV models outperform existing dynamic conditional correlation models for forecasting future covariances. Among the new fMSV models, the Cholesky MSV model with long memory and asymmetry shows stable and better forecasting performance for one-day, five-day and ten-day horizons in the periods before, during and after the global financial crisis..

Bayesian Parametric and Semiparametric Factor Models for Large Realized Covariance Matrices

“…The approach by Tao et al (2011) and extensions in Shen et al (2015) and Asai & McAleer (2015) decompose the RCOV matrix in a similar fashion to Engle et al (1990). Asai & McAleer (2015) model the decomposed factor in a number of ways including time-series models with long-memory, asymmetric effects and as a conditional autoregressive Wishart model. Shen et al (2015) focus on a diagonal model of the latter Wishart specification.…”

“…For instance, econometric forecasting gains are demonstrated in Golosnoy et al (2012), Asai & McAleer (2015) and Jin & Maheu (2013 while improvements in portfolio choice are found in Fleming et al (2003), Jin & Maheu (2013) and Callot et al (2017). through density forecasts of RCOV.…”

Frontiers in Time Series and Financial Econometrics: An overview

“…The broad theme of the special issue covers several exciting topics in time series and financial econometrics, including estimation, specification testing and prediction, The purpose of the special issue is to highlight a number of areas of research in which novel statistical, financial econometric and time series methods have contributed significantly to frontiers in time series and financial econometrics, specifically forecasting co-volatilities via factor models with asymmetry and long memory in realized covariance (Asai and McAleer (2015)), prediction of Lévy-driven CARMA processes (Brockwell and Lindner (2015)), functional index coefficient models with variable selection (Cai, Juhl and Yang (2015)), LASSO estimation of threshold autoregressive models (Chan, Yau and Zhang (2015)), high dimensional stochastic regression with latent factors, endogeneity and nonlinearity (Chang, Guo and Yao (2015)), sign-based portmanteau test for ARCH-type models with heavy-tailed innovations (Chen and Zhu (2015)), toward optimal model averaging in regression models with time series errors (Cheng, Ing and Yu (2015)), high dimensional dynamic stochastic copula models (Creal and Tsay (2015)), a misspecification test for multiplicative error models of non-negative time series processes (Gao, Kim and Saart (2015)), sample quantile analysis for long-memory stochastic volatility models (Ho (2015)), testing for independence between functional time series (Horvath and Rice, (2015)), statistical inference for panel dynamic simultaneous equations models (Hsiao and Zhou (2015)), specification tests of calibrated option pricing models (Jarrow and Kwok (2015)), asymptotic inference in multiple-threshold double autoregressive models (Li, Ling and Zakoian (2015)), a new hyperbolic GARCH model (Li,Li and 5 Li (2015)), intraday value-at-risk: an asymmetric autoregressive conditional duration approach (Liu and Tse (2015)), refinements in maximum likelihood inference on spatial autocorrelation in panel data (Robinson and Rossi (2015)), statistical inference of conditional quantiles in nonlinear time series models (So and Chung (2015)), quasi-likelihood estimation of a threshold diffusion process (Su and Chan (2015)), threshold models in time series analysis -some reflections (Tong (2015)), and generaliz...…”

A Multivariate Asymmetric Long Memory Conditional Volatility Model with X, Regularity and Asymptotics

“…Numerous instances of asymmetry have been considered for multivariate stochastic volatility models (see, for example, , , 2009a, 2009b, 2011), but not as frequently for conditional volatility (see McAleer et al, 2009) and realized volatility models (see, for example, Asai and McAleer (2015), Asai et al (2012), Asai et al (2011a, 20011b)). …”

“…If a stochastic process has a unit variance, the use of first differences may be convenient. When the ACF exhibits slowly declining correlations that do not converge to zero, a fractional differencing or long memory model may be useful (see, for example, Asai and McAleer (2015)). This will be discussed further in Section 3.…”